A program

to assess and monitor

soil quality in Canada

Soil quality evaluation program

summary

Edited by D.F. Acton

Centre for Land and Biological Resources Research,

Research Branch,
Agriculture and Agri-Food Canada,
Ottawa

Funding for this program was provided

through the National Soil Conservation Program,
under Federal-Provincial Agreements on Soil and Water
Conservation and Development in Alberta, Saskatchewan,
Manitoba, Ontario, New Brunswick, Nova Scotia and
Prince Edward Island.

Research Branch
Agriculture Canada
1994
This publication may be cited as Acton, D.F., Ed. 1994. A program to assess and monitor soil
quality in Canada. Soil Quality Evaluation Program Summary Report. Centre for Land and
Biological Resources Research, Research Branch, Agriculture and Agri-Food Canada, Ottawa,
ON. 201 pp.

Centre for Land and Biological Resources Research Contribution No. 93-49

Copies of this publication are available from:

Centre for Land and Biological Resources Research Research Branch

Agriculture and Agri-Food Canada
OTTAWA, ON K1A 0C6
TABLE OF CONTENTS

Chapter

Executive Summary

1. Introduction. D.F. Acton and G.A. Padbury.

2. A Conceptual Framework for Soil Quality Assessment and Monitoring. D.F. Acton and G.A.
Padbury.

3. GIS-Based System to Assess Soil Quality. KB. MacDonald, F. Wang, W. Fraser and I. Jarvis.

4. Land Use Analysis and Monitoring System To Assess Soil Quality. J.C. Hiley and E. C. Huffman.

5. Benchmark Sites for Assessing Soil Quality Change. C. Wang, B.D. Walker, H.W. Rees, L.M. Kozak
M.C. Nolin, W. Michalyna, KT Webb, D.A. Holmstrom, D. King, E.A. Kenney and E.F. Woodrow.

6. Wind Erosion Prediction and Assessment. MS. Bullock F.J. Larney, W. Michalyna, A. Moulin and
G.A. Padbury.

7. Water Erosion Prediction and Monitoring. G.J. Wall, E.A. Pringle and D.R. Coote.

8. Evaluating Changes in Soil Organic Matter. E.G. Gregorich, C.M Monreal, B.H. Ellert, D.A. Angers,
and M.R. Carter.

9. Assessing and Monitoring Soil Salinization. R.G. Eilers, W. D. Eilers, W.J. Stolte, M Trudell and
R. Stein.

10. Assessing Change in Soil Physical Quality. R.A. McBride, D. W. Veenhof G.C. Topp, Y. T. Galganov
and J.L.B. Culley.

11. Prediction of Agrochemical Migration. W. D. Reynolds, R. de Jong, S.R. Vieira and R.S. Clemente.

12. Industrial Organic Compounds in Selected Canadian Municipal Sludges and Agricultural Soils. MD.
Webber.

13. Soil Quality and Sustainable Land Management. M Cann, J Dumanski, S. Gameda, and M Brklacich.

14. SOILCROP: An Innovative Conservation Farmers' Expert System for Conservation Farm Planning.
S. Gameda and J. Dumanski

15. Relationships of Simulated Erosion and Soil Amendments to Soil Productivity. R.C. Izaurralde, M.
Nyborg , E.D. Solberg, M.C. Quiroga Jakas, S.S. Malhi, R.F. Grant and D.S. Chanasyk
EXECUTIVE SUMMARY

Over the past few decades, numerous concerns have been raised in Canada over the issue of soil
degradation. These concerns were often of a general nature, sometimes conclusions were
contradictory; there was insufficient precision and quantification in much of the analysis.

The Soil Quality Evaluation Program was initiated in 1989 in response to a requirement of the
National Soil Conservation Program (NSCP) to monitor soil and associated environmental quality
for the agricultural regions of Canada. A ten-year program was established within the Research
Branch of Agriculture Canada to develop capabilities and to assess the status of soil quality in
Canada. The focus has been on systems development and data collection, however, these various
capabilities are currently being used to analyze and report on soil quality in Canada.

The report to follow identifies the requirement and provides a framework for soil quality evaluation.
It summarizes the development of improved capabilities for assessing soil quality and for analyzing
the impact of soil degradation on soil quality and crop productivity. It also provides some insights
into the status of soil quality in Canada that have been forthcoming as part of the system
developments. An overview of these various developments is presented below.

A framework for soil quality evaluation has been developed which considers soil quality as the
capacity of the soil to produce crops in a sustainable manner without impacting negatively on the
environment. This recognizes the important role played by soils in providing a media for plant
growth, regulating and partitioning the flow of water and gases in the environment and in
providing an effective buffer for the environment. The framework proposes that soil quality
evaluation for environmental reporting should involve large-area assessment of critical attributes or
of processes known to modify soil quality and that these assessments should rely on predicted
change to these attributes or processes based on the application of simulation models to standard
databases of soils, land use, climate and topography. In addition, it recognizes the need for an
independent capability to validate and calibrate this predictive capability.

Specific studies were established to understand more completely and, using computer models, to
simulate various soil degradation processes as a basis for soil quality evaluation. This has resulted
in the development and testing of a series of capabilities that provide or have the potential to provide
independent indicators of soil degradation risk as well as composite indicators of soil quality. These
studies include:
1. The Wind Erosion Research Model (WERM), currently being developed in the USA, has
been validated at a site near Lethbridge. In a "worst case" management scenario, erosion
losses amounted to 144 tonnes per hectare (an average topsoil thickness of 14.4 mm) from
16 erosion events over a two year period. Associated studies led to the development of an
equation to relate erosion to crop productivity. Tillage, crop residue decomposition, soils and
erosion components of the WERM model were studied at various locations throughout the
Prairie region to provide the basis for incorporating crop, soil and weather conditions
prevailing on the Prairies into the model. Wind erosion monitoring sites across the Prairies
reported single event losses as high as 7 tonnes per hectare and total seasonal losses of 10
tonnes per hectare under typical land management in 1992.

i
2. The Water Erosion Prediction (WEPP) hillslope model has identified the most critical input
parameters for conditions in Canada and subsequent model development has recognized
these requirements. Validation studies in Ontario have indicated that, under no-till
conditions, WEPP tends to overpredict surface runoff volume, whereas, in British Columbia,
plot-scale investigations have shown that it underpredicts soil and water losses. Evaluation
of climate data suggest that many of the winter erosion components of WEPP are not well
modelled for conditions in Canada.

Baseline monitoring of soil erosion rates on field-scale plots in Quebec, New Brunswick and
Prince Edward Island, using tracer techniques, indicates levels of soil redistribution an order
of magnitude above the often proposed tolerance level of 5 tonnes per hectare per year.

3. Long- and short-term changes in soil organic carbon and nitrogen for Canadian climate, soils
and crop management systems were simulated using the CENTURY model.

Studies on long-term rotations on the Prairies confirmed previous research that management
practices, type of rotation, and rate of soil erosion influence the level of soil organic matter.
In one study, the amount of nitrogen in a Brown soil cropped for 23 years with fertilization
decreased by 3% under wheat-fallow but increased by 11% under continuous wheat. The
dynamics of carbon paralleled those of nitrogen. Contrary to previous claims, current studies
indicate that physical removal of soil by erosion may outweigh the effects of biochemical
oxidation as a contributing factor to the loss of soil organic matter.

In a study at 22 sites in Eastern Canada, cultivated soils appear to reach equilibrium levels
at which the mass of carbon is about 35% less and the mass of nitrogen 20% less than in
adjacent forest soils. Losses were greatest in sandy soils. Several poorly drained soils with
tile drainage had greater levels of carbon and nitrogen than corresponding forested soils.
Also, long-term application of fertilizers was shown to increase carbon levels in many
cultivated soils.

4. Soil salinity was monitored and investigated at seven sites on the Canadian Prairies.
Preliminary evaluation suggests that the extent of salinity is not increasing at any of the sites,
although there is fluctuation in concentration of salts. Contaminant transport models have
been used to simulate salinization processes at each site to determine the relative importance
of land use, geologic and climatic factors in controlling salinity.

5. Research was undertaken to determine if typical or improved crop production practices have
affected the physical behaviour of agricultural soils. Procedures were established to measure
an integrated indicator of soil physical quality, the Non-limiting Water Range (NLWR).
Measurements at nine locations across Canada indicate that the increasing soil strength
which may result from cultivation may be affecting the availability of soil water more than
has been previously recognized. About 20% of the soil horizons studied are showing
inadequate porosity for aeration, leading to possible restrictive root development; and some
of the intensively cultivated soils show a reluctance to wet. The impact of these forms of

ii
 degradation of soil structure on crop yield has yet to be determined. The application of the
NLWR concept provided indications of some of the possible causes of soil physical
degradation.

The magnitude of sub-soil limitations to crop productivity arising from high soil strength in
structured soils in the Regional Municipality of Haldimand-Norfolk, Ontario were reasonably
well predicted from fundamental and widely accessible data on soil physical properties,
generalized soil surveys and crop cover.

6. Methodology was developed to predict low-level, non-point source contamination of

groundwater resources due to the migration of agrochemicals through the soil root zone.
Although the Leaching Estimation And CHemistry Model (LEACHM) and associated
geostatistics and GIS methodology still requires further development and testing, the
preliminary results are very encouraging. Application of the methodology to predict,
characterize and quantify atrazine migration through the soil profiles in the Grand River
watershed of Ontario produced plausible results that are consistent with measurement of
atrazine contamination in wells in Ontario.

Soil contamination with industrial organic compounds such as are found in pesticides or their
alteration products was found to be very low or not detectable in 30 agricultural soils from
across Canada. The exceptions were residues of pesticides currently in use and the detected
levels were consistent with their use in crop production and are not known to represent a
significant hazard to the environment.

An evaluation of the persistence of organic chemicals derived from municipal sludge from
Hamilton and Sarnia concluded that land application of these sludges according to
recommended Ontario practice does not represent a significant hazard from organic
chemicals to agriculture and the environment.

Several approaches were used to evaluate the impact of soil quality change on crop yield and
sustainable production. An annotated bibliography of research which focused on the effects of soil
degradation on crop productivity was prepared and augmented by a survey of innovative
conservation farmers to produce a framework for evaluating sustainable land management and a
prototype expert system for soil conservation planning and research. Complementary studies of two
artificially eroded soils in Alberta further elucidated on the relationship between topsoil removed,
amendments and crop productivity. A USA model, the Erosion Productivity Impact Calculator
(EPIC) satisfactorily predicted grain yield, above-ground biomass, and grain nitrogen at low levels
of simulated erosion; accounted for the effects of commercial fertilizers and manure in restoring lost
productivity; and detected the slight advantage of commercial fertilizer over cattle manure in
restoring lost productivity.

iii
A prototype of an operational geographic information system was developed in Manitoba and
southern Ontario to assess the current status and trends in soil quality. This system uses data from
the Soil Landscapes of Canada map to assess inherent soil quality for crop production. Susceptibility
to change to this inherent soil quality is then predicted by applying knowledge of the soil-modifying
processes in conjunction with land and management data. Although the current prototype system
is limited, further development will provide a system that should meet the requirements to
periodically assess soil quality to support the State of Environment (SOE) reporting activity in
Canada.

As a basis for the analysis of land use and management, files linking Statistics Canada Census of
Agriculture data to land resource databases have been completed for the Prairies and Ontario. These
files provide access to more than 100 farm characterization variables that are spatially stratified by
soil and landscape criteria. A generic spatial decision support system has been developed to facilitate
integrated analysis of soils and census-derived land use information for these broad-scale
assessments. The high potential of remote sensing for assessments of soil quality at regional and
local levels was demonstrated using Landsat (TM) imagery.

A network of experimental sites was established to monitor soil quality at 23 benchmark sites
representing typical farming systems and dominant soils and landscapes within the major
agricultural regions of Canada. Baseline data collection has been completed and the preparation of
databases will be completed by 1995. On-site monitoring of climate, crop yield, land use and
management practices will enable the assessment of soil quality change by periodic resampling of
the sites. The sites also will be used to validate computer models which relate soil degradation to
crop production and, as mentioned above, to serve as a basis for validating predicted change to soil
quality.

A capability to monitor soil quality has been developed and tested. An assessment of the main
degradation factors at some of the more susceptible locations reveals that soil degradation continues
to be a serious threat to soil and environmental quality and crop productivity. Over the next few
years, efforts will be concentrated on developing a fully national assessment of the current situation
and trends over time. This will provide a scientifically-sound, quantitative measure of soil quality,
and its economic impact in terms of productivity, as a function of time. This will provide a firm
basis for the development of future land use and conservation policies.

iv
CHAPTER 1
INTRODUCTION

D.F. Acton and G.A. Padbury

Agriculture and Agri-Food Canada, Research Branch,

CLBRR Saskatchewan Land Resource Unit, Saskatoon, Saskatchewan

PROGRAM INITIATION

Reasons for Initiation

"Despite its widespread severity and global impact, soil degradation remains an emotional
rhetoric rather than a precise and quantifiable scientific entity" (Lal and Stewart, 1990).

Agriculture and food production constitute one of the largest economic activities in Canada.
Although Canada is the second largest country in the world, only 5% of its land is capable of
sustained production of field crops and only one-half of this land is considered to be prime
agricultural land.

Over the past few decades, concerns have been raised in Canada over the issue of soil degradation.
Nevertheless, most assessments dealing with this issue are of a general nature and are contradictory.
In the above quotation, Lal and Stewart (1990) describe the need for more precise and quantitative
assessments of soil degradation. The lack of precision and quantification can be attributed to the use
of relationships that have not been adequately tested or, if tested, have been found wanting. It is also
important to note that there are only a few sites across Canada where soil loss or alteration is
measured on an ongoing basis. These measurements are primarily directed toward a single cause of
soil or environmental change, such as water erosion, under a limited number of land uses and
management practices.

We know that changes in certain soil characteristics are the inevitable consequence of "breaking"
the land. Further changes resulting from soil degradation processes such as water and wind erosion,
salinization, loss of organic matter and nutrients, and compaction have been a consequence of
certain kinds of land use and management. The impact of degradation on crop yield is often unclear.
In spite of the alarming rates of degradation reported over the past decade or two, crop yields have
been increasing. Nowland and Halstead (1986) have suggested that soil degradation is beginning
to slow the rate of increase in agricultural production, but precise measurements of this relationship
have been as elusive as measurement of degradation itself. Resolution of this question is paramount
to determining the sustainability of crop-based agricultural production in Canada.

1-1
Clearly, precise quantification of the kind, severity and extent of the various processes that modify
soil quality has to be coupled with the impact of these processes on soil qualities that determine the
capability of soils to produce crops.

The Nature of the Requirement

Soil quality is often considered in terms of suitability for crop production (Van Diepen et al., 1991)
but, more recently, this has been extended to the reclamation of disturbed lands (Alberta Soils
Advisory Committee, 1989) and non-agricultural, often point-source pollution of agricultural lands
(Batjes and Bridges, 1993; Sheppard et al., 1992).

The need to include soil quality in the environmental assessment process is recognized by agencies
with mandates for environmental reporting on a national basis (SOER, 1991). The added requirement
for improved capabilities for the development and analysis of programs and policies related to
sustainable land management (Environmental Sustainability Bureau of Agriculture and Agri-Food
Canada, personal communication; OECD, 1991; IBSRAM, 1991; Federal-Provincial Agriculture
Committee, 1990) has necessitated the development of criteria and indicators for soil quality
assessment that are applicable over broad regions and within the broader sustainability context.
In recognition that the maintenance of soil quality for the production of food crops ultimately rests
with the producer, it is critical that the development of capabilities for national assessments be
complemented with capabilities to assess soil quality under various farming practices so as to more
directly serve landowners and local communities (Larson and Pierce, 1991; Sparling, 1991;
Hamblin, 1991).

PROGRAM PLANNING

Identification of a Monitoring Requirement

The working group (1986) report to Agriculture Ministers on Soil and Water Conservation recom-
mended the development of a comprehensive program to monitor soil degradation in Canada. The
National Soil Conservation Program (NSCP) provided the opportunity for action on this
recommendation by including soil quality monitoring as one of the components of a proposed
program. After extensive consultation, the Soil Quality Evaluation Program (SQEP) was established
by the Centre for Land and Biological Resources Research as the first phase in the development
of a soil quality monitoring program for Canada.

Implementation Plan

The plan for the SQEP was to develop a comprehensive national program to monitor the quality of
the agricultural soils of Canada. A number of research stations, Prairie Farm Rehabilitation
Administration (PFRA) and other federal departments, provincial departments and universities
collaborated with the Centre for Land and Biological Resources Research (CLBRR) in the
implementation of this program. Program delivery relied heavily on resources from the NSCP
provided through second-party agreements with PFRA for the Prairies and the Yukon and regional

1-2
offices of the Agri-Food Development Branch of Agriculture and Agri-Food Canada for Ontario and
Prince Edward Island. In Nova Scotia and New Brunswick, resources were provided through
third-party arrangements involving provincial ministries. Financial resources from the NSCP
provided the major source of funding for the SQEP activities but many of the partners direct very
significant levels of ongoing support in terms of personnel and operating expenses to this program.

Monitoring Approaches

Two fundamental strategies or approaches to the assessment of soil quality change have been
followed. In the first approach, baseline levels of soil properties considered important to soil quality
were determined at specific sites, and soil quality change is being evaluated from periodic
re-assessment of these properties (Martin, 1989; Tabi et al., 1990).

In the second approach, change to soil quality is being evaluated by using models or equations which
predict the direction and magnitude of change to an attribute or series of attributes based upon the
prevailing land management and other conditions of the area (Larson and Pierce, 1991).

Program Objectives

The objectives of the SQEP were: i) To develop a general procedure or framework for evaluating
soil quality, ii) To develop and evaluate capabilities to integrate soil and related databases at
appropriate scales for improved assessments of the current status and change to soil quality, iii) To
develop capabilities for monitoring and evaluating land use and land management in relation to
their effect on soil quality, iv) To establish a network of benchmark sites to monitor soil quality
change, v) To develop predictive technologies for evaluating soil quality change from wind and
water erosion, change in organic matter and structure, soil salinization and additions of agricultural
and industrial chemicals that will permit improved assessments of soil and related environmental
quality, and vi) To provide a framework for evaluating soil quality within the context of
environmentally and economically sustainable land management.

PROGRAM COMPONENTS

A series of integrated studies were established. Some of them were directed to applying, improving
upon, or developing a series of predictive models that simulate soil and environmental degradation
for conditions in Canada, while others aimed to develop capabilities to monitor land use and
management and to conduct assessments of soil quality change by applying these aforementioned
process models to appropriate soil, land use and other databases in a Geographic Information
System framework. A network of soil quality Benchmark Sites will provide an independent measure
for assessing soil quality change from the predictive approach described above.

The initial scope of each study was established by a series of workshops. It is summarized in the
remainder of this chapter.

1-3
Criteria and Standards - This study will identify the key criteria that can be used to measure
change to soil and water quality and to develop a general procedure for environmental
assessments involving soil.

Analyses and Assessment System - This study will develop the capability to make regional and
national assessments of soil and environmental quality by superimposing predicted change to
soil quality onto current soil, land use and climate databases.

Land Use - This study will develop a land use and management analysis capability for
application of soil-modifying process models as well as to provide a capability for specific land
use analysis and monitoring. Statistics Canada, Census of Agriculture data linked to physical
rather than traditional political regions will be central to the proposed capability. Also, the
feasibility of integrating information obtained by remote sensing into this Census database will
be assessed.

Benchmark Sites - Soil properties critical to evaluation of the potential land resources for crop
production and the constraints of land management on these resources will be measured at a
network of sites representing the major agricultural land resource areas and farming systems in
Canada. Changes to critical properties will be assessed by repeated measurement. The
information will be used to validate several of the predictive systems in the SQEP as well as to
provide a potential for calibration of the state of soil quality as a basis for subsequent re-
assessments.

Wind Erosion - This study will develop an improved capability to predict wind erosion loss
under Canadian conditions. The first phase of the study will focus on the prairie region where
the problem is considered most severe. The major thrust in this study will be to evaluate the
suitability of a wind erosion model being developed by the United States Department of
Agriculture for use in Canada.

Water Erosion - This study will measure soil loss from water erosion for selected soils and
farming systems across Canada and, from this, develop an improved capability to predict water
erosion loss. Approaches will include: monitoring of water erosion on plots, measuring soil
movement within typical fields and landscapes across the country using the 137Cs technique, and
rainfall simulation studies. All approaches will increase the capability to utilize existing models
and to evaluate the suitability, for Canadian conditions, of a new water erosion model under
development in the United States Department of Agriculture.

Organic Matter Change - This study will develop the capability to predict change to the amount
and kind of organic matter as a result of land use, length and kind of crop rotation, kind and
frequency of tillage and other management practices. Soils on long-term crop rotation plots at
research stations and universities will be used to evaluate the applicability of this model when
applied to soils and farming practices across Canada.

Soil Salinity - This study will develop the capability to predict changes to soil salinity levels
which may result from particular land use and management practices. This will be undertaken

1-4
 by developing a series of sites across the prairie provinces where soil salinity and associated soil
moisture, groundwater levels and land use will be monitored. The information obtained will be
used to develop a process-based model that simulates conditions encountered in the field.

Soil Structure - This study will develop procedures to measure soil physical qualities related to
the structures of agricultural soils in all regions of Canada, assess the current status of these
qualities, determine how these qualities are affected by crop, machinery traffic, tillage as well
as inherent soil properties over the short and long term, and develop relationships between those
soil structure parameters and economic and environmental sustainability. In one approach,
change to soil porosity will be measured at a series of sites across Canada. Another approach
involves a pilot study in Ontario where compaction risks due to spring and autumn traffic for
diverse agricultural landscapes will be estimated.

Organic and Inorganic Additions - This study will develop predictive capabilities to assess the
risk of groundwater contamination from additions of agrochemicals (fertilizers, pesticides) and,
secondly, evaluate the impact of additions of sewage sludge on the quality of agricultural soils.
Soil Quality and Sustainable Land Management - The long-term objective of these studies is to
develop a framework for assessing land resource and environmental sustainability. In the shorter
term, the studies will establish, through a comprehensive literature review, the impact of soil
degradation on crop yield and further develop this relationship by the utilization of an expert
system approach. They will also evaluate the impact of soil erosion on crop yield through field
experimentation on artifically eroded soils, with comparisons to results predicted by simulation
modelling techniques.

The ultimate goal for the SQEP is to develop a national capability to assess soil and associated
environmental quality so that farmers, extension advisers, policy makers and the public can have
the required knowledge to safeguard and preserve the agricultural lands of Canada for generations
to come.

The output will be practical methods for evaluating the impacts of land management practices on soil
quality as well as for the analysis of risk of degradation from erosion, salinity, etc. This will serve
as soil conservation and environmental protection planning tools at local and regional scales. The
output also will provide a systematic basis for monitoring soil quality, regionally and nationally, as
a component of programs aimed to report on soil and environmental quality.

REFERENCES

Alberta Soils Advisory Committee. 1987. Soil quality criteria relative to disturbance and reclamation.
Alberta Agriculture. Edmonton, AB. 56 pp.

Batjes, N.H. and Bridges, E.M. 1993. Soil vulnerability to pollution in Europe. Soil Use and Manage. 9:
25-29.

1-5
IBSRAM. 1991. Evaluation for sustainable land management in the developing world. Volume I. Towards
the development of an international framework. Bangkok, Thailand: International Board for Soil
Research and Management. IBSRAM Proceedings No. 12.

Federal-Provincial Agriculture Committee on Environmental Sustainability. 1990. Growing Together.

Report to Ministers of Agriculture. Agriculture Canada. Ottawa, ON. 41 pp.

Hamblin, A. 1991. Sustainable agricultural systems: What are the appropriate measures for soil structure?
Aust. J. Soil Res. 29: 709-715.

Lal, R. and Stewart, B.A. 1990. Need for action: research and development priorities. In R. Lal and B.A.
Stewart, Ed. Advances in Soil Science 11: 331-336. Springer-Verlag, New York.

Larson, W.E. and Pierce, F.J. 1991. Conservation and enhancement of soil quality. In: Evaluation for
sustainable land management in the developing world. IBSRAM Proc. No. 12 (2), 175-203. Bangkok,
Thailand: International Board for Soil Research and Management, 1991.

Martin, S. 1989. Observatoire de la qualite des sols. Service des Technologies Propres et des Dechets,
Secretariat D'Etat Aupres Du Premier Ministre, Charge De L'Environment. Neuilly-sur-Seine.

Nowland, J.L. and Halstead, R.L. 1986. Land, a fragile resource. Research Branch, Agriculture Canada.
Publ. 5221/B. Ottawa, ON.

OECD. 1991. Environmental indicators, a preliminary set. Organization for Economic Cooperation and
Development, Paris.

Sheppard, S.C., Gaudet, C., Sheppard, M.I., Cureton, P.M., and Wong, M.P. 1992. The development
of assessment and remediation guidelines for contaminated soils, a review of the science. Can. J. Soil
Sci. 72: 359-394.

S.O.E.R., Environment Canada. 1991. State of Canada's environment. Supply and Services Canada.
Ottawa, ON.

Sparling, G.P. 1991. Organic matter C and microbial biomass C as indicators of sustainable land use. In
Evaluation for sustainable land management in the developing world. IBSRAM Proc. No. 12(2):
563-580. Bangkok, Thailand: International Board for Soil Research and Management, 1991.

Tabi, M., Tardif, L., Carrier, D., Laflamme, G., and Rompre, M. 1990. Inventaire Des Problemes De
Degradation Des Sols Agricoles Du Quebec. Rapport Synthese. Gouvernement du Quebec. Publication
90-130156. 71 pp.

Van Diepen, C.A., van Keulen, H., Wolf, J., and Berkhout, J.A.A. 1991. Land evaluation: from intuition
to quantification. In B.A. Stewart, Ed. Advances in Soil Science 15: 139204. Springer-Verlag, New
York.

Working Group. 1986. Working Group report to Agriculture Ministers on Soil and Water Conservation
and Development. Agriculture Canada, Ottawa, ON.

1-6
CHAPTER 2
A CONCEPTUAL FRAMEWORK FOR SOIL QUALITY ASSESSMENT AND
MONITORING

D.F. Acton and G.A. Padbury

Agriculture and Agri-Food Canada, Research Branch,

CLBRR Saskatchewan Land Resource Unit, Saskatoon, Saskatchewan

INTRODUCTION

This conceptual framework is intended to provide a structure for organizing our knowledge of the
factors that determine the quality of a soil so as to facilitate the development of procedures for
assessing and monitoring soil quality in Canada.

The framework considers soil quality as the capacity of the soil to produce crops in a sustainable
manner without impacting negatively on the environment. It recognizes that soil quality is
determined to a large part by the inherent properties of the soil but that land use and management
practices may substantially change these inherent properties. Two general approaches to soil quality
assessment and monitoring are recognized and integrated into the framework. The framework
utilizes research on land use and soil management to predict change to soil quality. A capability to
assess and monitor soil quality and soil quality change can be developed by linking these predictive
capabilities with appropriate databases. Direct measurement of change to critical attributes serves
as a means to validate predicted values.

In the presentation to follow, the term evaluation of soil quality implies any consideration of soil
quality in time and space. Soil quality assessment, on the other hand, is an evaluation of the capacity
of the soil to perform a particular function at a given point in time. Monitoring represents the act of
repetitive measurement of this capacity.

SOIL QUALITY

Soil Functions

Soil quality describes a capacity or capability of the soil to perform certain functions in a sustainable
manner. McKeague (personal communication) described soil quality as the degree of excellence of
the soil in terms of its capacity for sustainable crop production while preserving a healthy
environment. Karlen et al.(1990) described soil quality in terms of the physical, chemical and
biological attributes of the soil, including the water that is retained, transmitted through, or runs off
the soil. Anderson and Gregorich (1984) considered soil quality to be the sustaining capability of
a soil to accept, store and recycle water, nutrients and energy. Larson and Pierce (1991), recognizing
the dual importance of environmental sustainability and sustainable crop production, embraced

2-1
many of these concepts as they considered soil quality to be the capacity of soil to function within
the ecosystem boundaries and interact positively with the environment external to that system. They
suggest that the quality of a soil should be considered as a composite of its chemical, physical and
biological properties as they perform three critical functions: i) provide a medium for plant growth,
ii) regulate and partition water flow through the environment, and iii) serve as an effective
environmental buffer. The growing concern for climate change and the impact of land management,
including soil quality, on this change suggest that gaseous partitioning could be considered as a
fourth critical function.

The four critical functions of soils alluded to above reflect the current desire of producers, other
resource managers, and the public for environmentally sustainable crop production systems.
Accordingly, and for ease of presentation, these four critical functions are placed into two groups,
each representing expectations placed on soils by various segments of society:

(i) sustainable crop production - the capacity of the soil to produce crops, and
(ii) environmental sustainability - the capacity of the soil to maintain and protect the integrity of
the environment in the immediate and in adjacent ecosystems.

Soil Capacity

The determination of the soil's capacity to perform the critical functions mentioned above requires
consideration of many, often complex factors and relationships; the scope of which may be better
appreciated with reference to Larson and Pierce (1992) and Anderson and Gregorich (1984). They
consider a quality soil has: i) an adequate capacity to accept, hold and release water to plants,
streams and groundwater, ii) an adequate capacity to accept, hold and release nutrients and other
chemical constituents, iii) a physical condition which promotes and sustains the unhindered
development of roots, including conditions which promote aeration, iv) an adequate capacity to
sustain soil biological processes, including the capacity to accept, store and recycle the energy
contained in organic matter, and v) an adequate capacity to respond to management and resist
degradation. Finally, there must be vi) an absence of unsuitable chemical conditions such as acidity
or salinity, that would be deleterious to plant growth.

SOIL QUALITY CHANGE

Soil Quality in Time and Space

This framework aims to provide a basis for the development of procedures for regional and national
assessments of soil quality and soil quality change. As such, it must consider procedures for
determining change over time at a particular site or local area and coupling these with procedures
for extrapolating from point sources or small areas to larger areas. Hamblin (1991) suggests that all
research methodologies are limited either by constraints of space, in that they are too specific to be
extrapolated to whole regions, or by constraints of time, being measured over too short a period to
be predictive of any long-term trend. She suggests that to extrapolate from point-sources of
measurement in time and space requires some form of modelling or statistical manipulation.

2-2
An approach to address the time dimension in the above question is provided by Larson and Pierce
(1992). They propose soil quality (Q) be considered as the state of existence of soil relative to a
standard or in terms of degree of excellence and that it be defined as the sum of all of the individual
soil qualities or attributes such as texture and organic matter. In that the maintenance and
enhancement of soil quality requires knowledge about changes in soil quality (dQ) and not solely the
magnitude of Q, it is necessary to consider change to the sum of all soil qualities between two points
in time. Individually, any soil quality may be aggrading or degrading, but it is the collective impact of
all soil qualities that determine dQ.

Through consideration of the dynamic change in soil quality, the relationship dQ/dt can be used as a
measuring device for evaluating the impact of land management systems with respect to either soil
degrading or soil aggrading processes. A minimum set of soil attributes could be chosen as indicators
of soil quality and these could be monitored over time. These minimum data sets should be selected
from those soil attributes in which quantitative changes can be measured in a few years time.
Alternatively, these data sets could be used to identify indirect measures of a soil quality attribute or
groups of attributes which could serve as surrogate indicators of Q and, hence, provide a means of
monitoring Q and its change in response to land management and soil degradation processes.

One approach to the evaluation of soil quality and soil quality change over large areas involves the
selection of sites that are representative of the agricultural land resource areas and farming systems of
the region or country and periodically measuring selected attributes important to soil quality at these
sites. Alternatively, soil data bases in combination with land use and management information provide
a basis for the assessment of soil quality wherever this information has been collected. In this
approach, dynamic soil quality is predicted from information that describes land use and management
and models or other techniques that establish a quantifiable relationship between soil attributes and
associated land use and management.

Soil Quality Factors

The framework considers land management practices as the causes of change to a soil's quality and
provides examples of soil processes that may be impacted by the various practices. It proposes that the
considerable knowledge and analytical capabilities that have been developed to relate land
management to soil degradation (i.e. wind and water erosion) can provide a basis for determining the
impact of land management on soil quality. This provides a focus for the selection of the most relevant
processes and attributes and, in turn, the most appropriate set of criteria and indicators for the
assessment of soil quality for any critical function. Elaboration of the land management and soil
factors and the soil-modifying processes is provided below. Proposals for the development of criteria
and indicators are presented in following sections.

Land Management Factors

Land management impacts on a number of soil processes that determine soil quality. Greer (personal
communication) has listed crop type and rotation, tillage, traffic, and additions of fertilizers,
pesticides and animal, human and industrial waste as examples of management practices that affect
soil quality. He also has indicated the kinds of soil processes that are moderated by each of these

2-3
practices (Fig. 2.1). When used in conjunction with the factors determining soil quality enunciated
above by Larson and Pierce and Anderson and Gregorich, it is possible to select the most appropriate
processes for assessing change in soil quality for each critical function.

Soil Factors

Soil processes such as infiltration and respiration, which determine the soil's capacity to perform a
specific function, and critical attributes such as soil depth and organic carbon content, which can be
used to quantify these processes, are termed the factors of soil quality. Lists of these factors for each
of the crop production and environmental functions, shown in Fig. 2.1, have been developed by Greer
(personal communication). Consideration of these factors in developing criteria for assessing soil
quality change recognizes that some factors are more important determinants of the capacity of a soil
to perform specific functions, are more responsive to land use and management pressures, and are
more easily measured than others and, hence, are more suitable for use as criteria for determining soil
quality change.

Soil-Modifying Processes

Processes such as wind and water erosion, salinization, oxidation and mineralization of organic
matter which degrade or aggrade the soil are referred to as soil-modifying processes. They are
recognized as processes affecting soil quality for both critical functions (Fig. 2.1). Water erosion, for
example, has long been recognized to affect the critical function of crop production. More recent
concern involves its effect on the quality of water in streams or lakes. The critical point, as it relates
to the environmental function, is not whether water erosion is impacting on surface water quality but
whether water erosion is impacting on soil quality by influencing its capacity to accept, hold and
release water to plants, streams and other water bodies. These soil-modifying processes can be used
to evaluate change to individual or composite groups of soil attributes that have been or will be
established for the two critical functions.

Work is on-going in the development of physically-based simulation models for wind and water
erosion, loss in soil organic matter, and in the development of models or alternate capabilities to
predict change in soil quality related to soil structure change and soil salinity. A model is also being
evaluated to predict the movement of chemicals to groundwater. These models have the capability to
predict change to many of the attributes that determine soil quality and hence enable us to measure
change to the capacity of a soil to perform a critical function.

OTHER DIMENSIONS OF SOIL QUALITY ASSESSMENT

Assessment of soil quality, in essence, implies an evaluation of the capacity of a soil at a particular
locale to perform specified functions. This is to say, the assessment addresses soil quality change
within the boundaries of the ecosystem.

The suggestion by Larson and Pierce (1991) that soils interact positively with the environment
external to the ecosystem in which they are located implies that assessments must recognize that the
pedosphere is inextricably connected to the lithosphere, biosphere, atmosphere, and hydrosphere. A

2-4
soil's capacity to function within that pedosphere, and, in particular to respond to land use and
management practices, will determine the potential for change to any of the related spheres (Arnold
et al., 1990). However, the framework is not directed to the assessment of soil-induced environmental
change in a global context but simply to provide the capability to assess, if required, whether these
environmental influences prevail in any particular assessment.

In a similar vein, the assessment of soil quality change will focus on those man-induced changes
directly related to the agricultural sector. Changes induced by transportation or industrial
contamination, for example, are not considered.

This framework is generic in nature, considering, for example, that all crops have the same basic
growth requirements. As such, some assessments may require the identification of a more specific
purpose than that provided by the critical functions described above.

The utilization of the framework depends upon the scale of the assessment. For example, is the area
of interest a field, a farm or some larger region.

SOIL QUALITY STANDARDS

The framework, to this point, identifies the factors that determine soil quality, but it does not place any
comparative or judgement value on them. The final step of an assessment is to indicate the degree of
excellence of the soil for the use in question, and the final step of monitoring is to indicate the
direction and magnitude of change. As such, assessments require a standard for the comparison of soil
quality and monitoring requires a standard for soil quality change. Before proceeding to an approach
for the development of standards, a definition of a soil quality standard is required.

Soil Quality Standards

Standards of soil quality include any defined basis for determining a stated condition (or for comparing
changes in soil condition) that indicates the health, quality, or productive potential of a soil for a
specified use. They may also indicate a level at which the soil cannot suffer additional change without
showing significant adverse effects and/or irreversible damage. These standards are use-specific and
are not intended to be a rating or a limitation scheme for a particular use (soil capability and many
other soil interpretations are different concepts).

The development of standards for soil quality requires consideration of criteria and thresholds and
the assessment or monitoring of soil quality or soil quality change requires the development of
indicators. The meaning and application of these and related terms is provided, as follows, as a basis
for the full development of this framework.

Soil Quality Criteria

A criterion is a standard or a rule against which a judgement can be made. It may be a field or
laboratory test or measure such as the determination of total soil carbon. It may be an equation that
predicts soil loss from water erosion, or it may be a model that predicts soil salinization (Fig. 2-1). As

2-5
such, these criteria identify factors, numerical relationships, formulae or more complex models that
provide a set of conditions or rules which enable us to observe, measure or predict the dynamic state
of soil quality. They are the "rule-base" for a set of standards. For example, a standard could be
established for levels of nitrate leached below the rooting zone. The same limit could be used for all
soils or limits could be soil specific. A set of criteria would define the limits and, perhaps, how they
would be applied in assessment and monitoring.

Thresholds

Thresholds are a particular form of a standard denoting levels beyond which soil quality undergoes
intolerable and perhaps irreversible change.

SOIL QUALITY INDICATORS

Indicators provide a set of measurable or observable attributes which reflect changes to processes and
hence to attributes governing soil quality. They are often short-cuts to the development of rigorous
criteria and standards for soil quality assessment and monitoring. A comprehensive capability to
evaluate soil quality, as indicated earlier, would address the capacity of a soil to function for a specific
purpose by evaluating all of the important attributes that contribute to this capacity. It is impractical
if not impossible to develop criteria that would address all of these attributes for all, perhaps any,
evaluation. Instead, a limited number of key attributes are chosen to represent the condition to be
assessed or the change thereto to be monitored. These key attributes or indicators take many forms,
to be considered in the following in terms of soil quality attributes, surrogate or proxy indicators and
soil quality elements.

Soil Quality Attributes

A soil quality attribute is a measurable soil property that influences the capacity of the soil to perform
a specific function. An exemplary list of critical attributes is provided in Fig. 2-1. Some attributes,
however, are more suitable than others for measuring soil quality change. Attributes such as soil depth,
soil organic matter and electrical conductivity are often selected to represent properties most affected
by soil degradation processes (Arshad and Coen, 1992). Attributes that are more sensitive to
management are most desirable. In Chapter 8 of this report, attributes such as microbial biomass,
amino acids and soil enzymes are proposed as highly sensitive soil quality attributes.

Surrogate or Proxy Indicators

The framework provides the opportunity to select indicators that are either direct or indirect measures
of land management practices that clearly impact on soil quality. Crop cover, including residues,
exemplifies this type of proxy indicator; rates of addition of fertilizer, manure, or sludge are others.

Surrogate or proxy indicators of soil quality also may be derived from consideration of the
soil-modifying processes. They include risk to soil erosion, compaction, salinization or they could
include measures or estimates of soil loss from erosion or extent or severity of soil salinization,
compaction, or contaminated land, for example.

2-6
Fig. 2-1. Outline of a procedure for selecting criteria for the
evaluation of soil quality change.

2-7
Elements of Soil Quality

The critical soil functions described previously must be determined before soil quality can be assessed.
These functions are scale-independent, remaining the same for plots, fields, districts or larger regions.
The information used to assess soil quality, however, is dependent upon scale. This has necessitated
the identification of scale-dependent components of the critical functions, termed elements of soil
quality, as basic units of assessment.

Four soil quality elements have been established to provide a basis for quantification of the crop
production function: available porosity, nutrient retention, physical rooting conditions, and chemical
rooting conditions. Similarly, elements such as surface water recharge, water storage in the rooting
zone, water release for plant growth, water transmission below the rooting zone, and the capacity of
the soil to absorb, retain and recycle contaminants are proposed as elements for the assessment of the
various environmental functions (Fig. 2-1).

OUTLINE OF ASSESSMENT AND MONITORING PROCEDURES

A framework for soil quality evaluation is shown in Fig. 2-2. The pathways in such a framework are
as follows:

1. Determine the objectives of the evaluation. Begin by establishing the critical soil quality functions
that are important to the assessment. Are they related to crop production or to environmental
protection, or both? What are the dimensions of the assessment? Is it related to a single event or
to multiple events over a protracted time frame? Is it for a site, field, farm or for an extended
region? Is it focused within the ecosystem boundaries or on the environment external to that
ecosystem? Once the functions and dimensions have been defined, it is possible to initiate the
selection of criteria. Final selection of criteria may require consideration of available data and data
management capabilities.

2. Data collection and management. The specific data requirements will depend upon the purpose
and the dimensions of the assessment, but they always will include soils, topography, climate, and
land use and management. In assessment or monitoring over large areas, it may be advantageous
to integrate the various data sets, that is, to organize the data so that each area under consideration
has a description of all the relevant information. It also may be advantageous to organize the data
so that the initial analysis can be conducted at a detailed scale with facility to generalize the
information in the final analysis. "Nested databases", whereby groups of map polygons at a detailed
or large scale fit, or, are nested within polygons at successively smaller scales, facilitate such
analysis.

3. Make a final selection of the criteria for the assessment. The criteria are soil attributes contained
in the database. They may be resident data or they may be data to be collected as part of the
evaluation. They may be single attributes such as clay content or pH or they may be an integration
of a number of soil attributes into more comprehensive elements of soil quality.

2-8
Fig. 2-2. Framework for soil quality evaluation.

2-9
4. Predict the condition of, or change to, soil quality. A series of soil-modifying process models have
been developed that could facilitate the analysis of change to soil quality in space and time. It is
not always necessary or even possible to operate these models in their entirety. It is sometimes
adequate to develop more simple mathematical relationships for this analysis.

5. The output of an analysis following this framework is a measured or predicted condition of soil
attributes or elements that reflect a change to, or a new state of, soil quality.

ACKNOWLEDGEMENTS

The authors wish to acknowledge the many colleagues that provided suggestions and criticisms in the
development of this framework, particularly: D.W. Anderson, M.A. Arshad, G.M. Coen, W.W.
Pettapiece, and K.B. MacDonald. They also thank D.J. Mittelholtz and C.T. Stushnoff of the
University of Saskatchewan for the preparation of illustrations.

REFERENCES

Anderson D.W. and Gregorich, E.G. 1984. Effect of soil erosion on soil quality and productivity.
Pages 105-113 in Soil Erosion and Land Degradation. Proceedings of the Second Annual Western
Provincial Conference on the Rationalization of Water and Soil Research and Management. Sask.
Inst. of Pedology, Univ. of Sask., Saskatoon, SK.

Arnold, R.W., Szabolcs, I., and Targulian, V.O. 1990. Global soil change. Report of an
IIASA-ISSS-UNEP task force on the role of soil in global change. International Institute for
Applied Systems Analysis, Laxenburg.

Arshad, M. and Coen, G.M. 1992. Characterization of soil quality: physical and chemical criteria.
Amer. J. Ahern. Agric. 7: 25-31.

Hamblin, A. 1991. Sustainable agricultural systems: What are the appropriate measures for soil
structure? Aust. J. Soil Res.29:709-715.

Karlen, D.L., Erbach, D.C., Kaspar, T.C., Colvin, D.S., Barry, E.C. and Timmons, D.R. 1990.
Soil tilth: a review of past perceptions and future needs. Soil Sci. Soc. Am. J.54:153-161.

Larson, W.E. and F.J. Pierce. 1991. Conservation and enhancement of soil quality. In: Evaluation
for sustainable land management in the developing world. IBSRAM Proc. no. 12(2), 175-203.
Bangkok, Thailand: International Board for Soil Research and Management, 1991.

2-10
CHAPTER 3
GIS-BASED SYSTEM TO ASSESS SOIL QUALITY

K.B. MacDonald1, F. Wang1, W. Fraser1, G.W. Lelyk2 and I. Jarvis1

1
Agriculture and Agri-Food Canada, Research Branch, CLBRR Ontario Land Resource Unit,
Guelph, Ontario
2
Agriculture and Agri-Food Canada, Research Branch, CLBRR Manitoba Land Resource Unit,
Winnipeg, Manitoba

INTRODUCTION

The purpose of this study was to: i) develop an operational geographic information system (GIS)
that integrates standard databases of soil, topography, climate and land use at appropriate scales
for improved national and regional assessments of the current status and trends in soil quality, and
ii) demonstrate, in selected agricultural areas of Canada, the current status and potential for change
to soil quality.

The development of the GIS necessitated: i) definition of concepts and the development of a
general framework for soil quality assessment as documented in the preceding chapter of this
report, ii) preparing a 'minimum' set of standardized data bases, and iii) developing a set of
algorithms and GIS routines.

The current prototype system is limited; the algorithms require further development and testing,
and some of the data requires updating and correction. In its final form, the system should meet
the requirements to periodically assess soil quality and its changes to support the State of
Environment (SOE) reporting activity in Canada.

CONCEPTS AND OPERATIONAL FRAMEWORK

Concepts

During the course of this study a broad range of the concepts and definitions used to characterize
soil quality were reviewed. In some cases, it was necessary to name and define aspects of soil
quality for which no appropriate description could be found. The debates about whether these
concepts and definitions are the best possible will continue; however, the meanings represented
here should be kept in mind while reading and interpreting this chapter and other reports from this
study. The operational definition of soil quality developed by Larson and Pierce (1992), as
presented in the preceding chapter of this report, has been adopted for use in this study.

A definition of soil quality requires determining a desired use or function to be carried out on the
land. Several general land functions have been identified: production, environmental buffering,
and partitioning of water and gases. In any farming system, each of these functions proceed

3-1
concurrently at land management levels somewhere between low (natural or undisturbed) and
highly manipulated (or managed). For example, a land area which is used for intensive agricultural
production may also be required to serve as an environmental buffer to retain nutrients from
manures in the rooting zone and to ensure that precipitation is partitioned into soil storage rather
than allowing surface runoff and contamination of adjacent water.

The actual process of assessing soil quality requires that data from a variety of sources be
combined into components or elements of the function of interest. This process, as it was
implemented in this study, is illustrated in Fig. 3-1. The definitions associated with the various
components of this figure were presented in the preceding chapter of this report.

Procedures were developed to characterize the capacity of a soil to perform a function as defined
by Larson and Pierce (1992). Of the three functions alluded to above (production, environmental
buffering and water partitioning), production was chosen for pilot development. Each of the soil
quality functions is quite complex and its characterization could involve increasing levels of detail
right down to microscopic. For the system to be workable, characterization must be based on a
limited number of components or elements which themselves can be characterized by components
in a nested hierarchy (Allen et al, 1984). The general approach taken was to identify components
or elements of each function and choose the best data sources and attributes to estimate the
elements for the scale of interest. In reality, the elements of soil quality related to production tend
to characterize physical, chemical and biological activity of the soil and, as such, they are closely
related to the capacity of the soil to support other functions.

The information used to estimate soil quality elements is normally stored in standard,
computerized databases. The databases used in this study were: land resource, topography,
climate, and land use and management. Items of information stored in a database are termed
attributes.

Reliability must be considered when information is extracted from one or more database and
combined into estimates of soil quality. Determining the reliability of attributes requires
consideration of: i) spatial reliability, that is, the measure of the uncertainty associated with the
location and extent of an attribute and ii) attribute reliability, the confidence with which the value
associated with an item can be used to represent that item. This concept includes both the inherent
variability of the attribute and also the precision or size of the class represented by a single value.

In this study, information collected for the 1991 Census of Agriculture has been summarized on
the basis of soil landscape polygon boundaries, as described in the chapter to follow, to provide
for a more logical area for natural resource analysis and to allow for improved registration with
other data assembled on the same spatial basis.

3-2
 Fig. 3-1. An hierarchial framework of attributes and elements of major
functions of soil quality.

A General Approach to Soil Quality Assessment

The quality of a soil indicates how well it can function for one or more purposes. Assessments of
soil quality over time provide a measure of sustainability of the land management practices. Soil
quality assessment procedures can be described in several simple steps. These steps prevail
irrespective of whether the area being assessed is a site or field or an entire province or country.

3-3
They are also followed when the assessment is carried out intuitively and subjectively (by a farmer
or field extension worker) or more explicitly by a researcher using computerized databases, GIS
and computer models.

Basic steps in soil quality assessment

1. Estimate the existing (inherent) soil quality for one or more specific functions using basic
land resource information (e.g. a deep well-drained soil with adequate capacity to retain
and supply nutrients provides good quality for crop production and also for retention and
degradation of toxic organic materials).

2. Estimate the physical conditions causing land to be susceptible to change in quality using
basic topographic and land resource information (e.g. steep slopes and silty surface
textures make the soil susceptible to erosion by water).

3. Estimate the man-made conditions causing land to be susceptible to change using land
use and management information and trends (e.g. intensive row-cropping up and down
the slope make water erosion and soil organic matter loss more likely).

4. Combine steps 1, 2, and 3 over time to predict changes to land resource quality: i) by
subjective estimation, or ii) through monitoring and recording new values of land resource
data, or iii) through the use of models and historical or representative climate data.

5. Using the land resource data estimated at some time in the future, re-evaluate the quality
of the resource for a specified use.

The overall concept of quality embodies both the capacity of the soil to carry out a specific
function and its ability to maintain or improve functionality for a variety of possible uses. In
general terms, this approach suggests that three aspects of soil quality are considered: inherent soil
quality, soil quality susceptibility to change, and soil quality change. Definition of these terms as
used in this study are listed below.

Inherent Soil Quality (ISQ) refers to those properties of the soil which contribute to the capacity
of the soil to support a specific critical function and which are relatively unchanging through time.
Normally, this aspect of soil quality is estimated from the values calculated for the elements. In
the case of the production function and associated elements, ISQ does not deal specifically with
landscape-related features such as slope. Estimates of inherent soil quality can be based exclusively
on land resource data and represent the soil quality at the time the data were collected.

Soil Quality Susceptibility to Change (SQS) is an estimate of the ease with which soil quality
can be changed by typical current and/or proposed practices. It is estimated, based on knowledge
of the area under study and assumptions about the kinds of soil-modifying processes which are
important. These assessments are based on land resource, topographic, and land use and manage-
ment data. It consists of a combination of two aspects: i) Biophysical Susceptibility to Change
(BSC) - characterized by aspects of the land resource which make it more or less likely that a

3-4
soil-modifying process will change some basic land resource attributes and that this change will
result in a net change to the capacity of the soil to support the critical function of interest. It is
represented by threshold values of individual attributes and combinations of attributes, and ii)
Land Use and Management Susceptibility to Change (LUMSC) - aspects of past, current or
proposed land use which make the soil more or less susceptible to processes which modify soil
quality.

Soil Quality Change (SQC) refers to the change in resource quality (for a use) over time (t2 - t1)
or space; it has both magnitude and direction. Dynamic soil quality represents the direction aspect.
This assessment is normally based on land resource data used in conjunction with land use and
management, topographic, and climatic data. These databases are usually integrated and evaluated
over time.

While soil quality change is of key interest, very often the data available are not sufficiently
precise, either in terms of spatial resolution or the attribute values, to allow the calculation of
changes in soil quality. This is particularly true at national and regional scales where soil inventory
reports are the principal source of land resource information. With regard to soil inventory, this
kind of analysis is not the intent of the data but rather the inventory data purports to describe the
state of the land resource in one area relative to another (Hudson, 1990).

Although the characterization of soil quality status (inherent quality) was one of the objectives of
this study, it was also the intent to provide some indication of soil quality susceptibility to change
resulting from soil-modifying processes and to identify the requirements to quantify soil quality
change. A second set of procedures was developed to identify and map this susceptibility of soil
quality to change. These procedures encompass assessments of the effects of biophysical
characteristics and land use and management practices on soil quality.

Procedures to determine soil quality change were not developed; it was felt that the data available
were inappropriate for this assessment. In the future, it may be possible to evaluate soil quality
change using process-based models in combination with land resource, climate, socio-economic,
and remotely sensed data.

The various procedures referred to above cannot be directly compared with previous evaluation
systems for Canada (for example, Wang, et al. 1991; MacDonald and Brklacich, 1992; Brklacich
and MacDonald, 1992) as the approach, objectives, databases and technical environments are not
the same.

3-5
 METHODS
GIS Systems

The system was developed using Arc/Info1 GIS, operating on a VAX mainframe and MicroVAX
computers, and PAMAP GIS installed on a personal computer. The rating algorithms were
developed and implemented in dBase IV and linked with the GIS.

Kinds, Sources and Quality of Data

The major data sets used for the assessment at national scales are: i) Soil Landscape of Canada
(SLC) map and database at a scale of 1:1 million, ii) Canadian Soil Carbon Data Base at a scale
of 1:1 million (also includes some soil-landscape attributes such as slope steepness, iii) Soil Name
File (SNF) and Soil Layer File (SLF) which can be linked with the above data sets; iv) vectorized
Advanced Very High Resolution Radiometer (AVHRR) 1989 composite with 1 x 1 km resolution,
v) 1986 and 1991 Census of Agriculture at the Consolidated Census Subdivision (CCSD) level
(Census of Agriculture data was also available on the basis of SLC polygons). Note: only the
CCSD-level data was available at the time of writing this report.

The land resource data were obtained from the National Soil Data Base (NSDB) of Agriculture and
Agri-Food Canada and the Census of Agriculture data were purchased from Statistics Canada. The
AVHRR data were purchased from the Manitoba Centre for Remote Sensing and vectorized by
Energy, Mine and Resource Canada.

In general, the spatial resolution of the information is approximately ± one kilometre and
appropriate for national scale (1:1 million) assessment. The quality of the attribute information was
quite variable, particularly for the land resource layer which was compiled from a variety of data
sources ranging from expert estimates to summaries from detailed soil surveys.

Inherent Soil Quality (ISQ) Assessment Procedures

Inherent soil quality (ISQ) is estimated from land resource data by identifying a set of soil quality
elements and then defining combinations of soil attributes to characterize each of the elements for
the functions (crop production, environmental buffering, etc.) in question. It is recognized that
elements chosen for one function may also apply to the assessment of other functions. For
example, an element such as nutrient retention chosen to assess inherent soil quality for crop
production also has relevance in the assessment of the soil's capacity for environmental buffering.

In the assessment of the crop production function at the national level, where the land resource data
are very generalized, the crop properties represent a "generic" crop to allow for inter-regional

__________________
1
The mention of a trademark, proprietary product or vendor in any part of this report does not imply
endorsement by Agriculture and Agri-Food Canada to the exclusion of other products or vendors.

3-6
comparison. At detailed levels of assessment it is deemed essential to define properties for the
crops that are specific to the region under consideration. As such, the weighting assigned to the
elements may differ for different broad classes of crops, e.g. annual compared to perennial field
crops, or agricultural compared to forestry. The basis of the logic used for these assessments will
be developed more fully in the following section.

The Agronomic Interpretations Working Group (1992) has developed a land suitability rating
system for spring seeded small grains (LSRS). It is a national system for rating land suitability for
the production of crops. Rating factors have been developed for spring grains and the system is
currently being tested. The procedures are objective and provide a separate rating of the climate,
soil and landscape factors. Although the ISQ rating only requires soil attributes, appraisal of
climate and landscape are normally required to accommodate the requirements of the crop under
consideration. In that the LSRS provides an adequate basis for rating ISQ, the logic and the
algorithms used for LSRS have been adopted.

Procedures to estimate inherent soil quality were developed to use data from the standard attribute
files associated with maps in the NSDB. These programs are designed to access data in the polygon
or soil map unit files (SMUF) as well as the soil names (SNF) and soil layer files (SLF). They can
be operated in conjunction with generalized maps (as reported here) as well as detailed soil survey
maps (at various scales ranging from 1:10,000 to 1:125,000). For this implementation, the
estimates include a limited number of soil conditions and processes related to production. These
procedures can be modified to accommodate alternative crops and will be improved as the
relationships between soil attributes and production are better understood.

A preliminary assessment of the land resource data was carried out to define the boundary
conditions for the assessment. The values for the national-level crop assessment are defined as in
Table 3-l. It is recognized that the rooting depth limits are not necessarily ideal for any crop;
however, they have been chosen to allow the specific production elements to have meaning at both
depth extremes.

Four soil quality elements were selected for the assessment of inherent soil quality for the crop
production function: available porosity, nutrient retention, chemical rooting conditions and
physical rooting conditions. Estimation procedures were developed for all soils in Canada which
meet basic criteria for inclusion. Details of the estimation procedures for each of these elements
are outlined in the following sections.

Available porosity

Available porosity was chosen as one of the four soil quality elements for crop production so as
to assess the soil's capacity to provide air and water, two vital elements for plant growth. This
capacity is determined, in the first instance, by the volume of pores present in the soil (porosity).
The capacity to supply air, Aeration Porosity, and the capacity to supply water, Soil Water
Holding Capacity, are considered as independent "sub-elements", the most limiting of which is
used as the final rating value for the ISQ for that element.

3-7
Table 3-1. Restrictions and exclusions of soil characteristics to assess ISQ for national-level
assessment
Excluded Conditions:
Climate < 1050 Effective Growing Degree Days (EGDD)
Landscape
Maximum Surface Stoniness Class 3
Maximum Slope Class 30%
Soil
Organic soil exclude
Cryosolic soil exclude
Drainage class Very Poor exclude
Water exclude
Rooting depth < 20 cm
Salinity > 16 cm
pH < 3.5
pH > 10
Air-filled porosity (surface 20 cm) < 5%
Nutrient retention (surface 20 cm) <8
Root Restrictions:
Maximum rooting depth 80 cm
Minimum rooting depth 40 cm
Maximum bulk density (clay > 40) 1.60
Maximum bulk density (clay < 40) 1.45
Root Restricting layers Duric, Ortstein, Placic, Fragipan

The initial rating of porosity for aeration and moisture is based on soil bulk density and moisture
retention characteristics. Final ratings also depend upon soil drainage class and climatic zone. For
example, a well drained sand with a low available water holding capacity is a severe constraint to
crop production in more-arid areas, such as the Brown soil zone in western Canada. Where water
tables are present, such as imperfect, poor, and very poorly drained soils, this moisture supply may
compensate for a low available water holding capacity. On the other hand, a water table within the
rooting zone will reduce the porosity available for gaseous exchange (aeration porosity), which
may be an increased constraint on crop production.

In that comprehensive soil climatic zone attribute fields are not currently available in standard
NSDB soil data files, climate attributes were approximated using attributes for taxa in the
Canadian System for Soil Classification (i.e. great group, subgroup, and drainage) as contained
in the Soil Names File. Table 3-2a illustrates the "soil climatic zones" that were recognized.

The aeration porosity is calculated for each soil horizon in the Soil Layer file and these values are
summed to obtain a total aeration porosity (in cm of air) for the entire soil. This is calculated in
two stages; the surface layer (20 cm) and the subsurface layer. The formula used is as follows:

PORETOT(cm) = PORETOT(cm) + THICKNESS * [(2.65 - BD)/2.65 - 0.01 * KP33]

3-8
where,
THICKNESS = the thickness of the soil layer (cm). This is the absolute value of the upper
depth minus the lower depth, as recorded in the Soil Layer file.

BD = Bulk Density, g cm-3. *

KP33 = % water, by volume, that is retained by the soil at 1/3 atmosphere suction. This is
multiplied by 0.01 to convert from % to a ratio of the soil volume.

KP33 approximates field capacity.

_____________________
* A density value of 2.65 gcm-3 for the solid mineral soil material is used for all horizons where the
organic carbon is less than 2.0%. Where the organic carbon values are between 2% and 17%, a value
of 2.54 g cm-3 is used and for soil layers with organic carbon values of >17% (ie., organic layers), a
value of 1.00 g cm-3 for the solid soil material is used.

Table 3-2a. Soil climatic zones as defined by soil taxonomic attributes

SOIL
CLIMATIC SOIL NAMES FILE ATTRIBUTE VALUES
ZONE NO. P-PE1 (actual soil SUB GROUP.GREAT GROUP codes)
1. < -350 All well-drained Brown Chernozemic and Solonetzic subgroups.
(O.B, R.B, CA.B, E.B, SZ.B, B.SZ, B.SS, B.SO)

2. - 350 to All well-drained Dark Brown Chernozemic and Solonetzic

- 300 subgroups. (O.DB, R.DB, CA.DB, E.DB, SZ.DB, DB.SZ, DB.SS,
DB.SO). Also O.R, CU.R, O.HR, CU.HR for Province = "MN".

3. - 300 to All well-drained Black and Dark Gray Chernozemic and

- 200 Solonetzic, and Gray Brown Luvisolic subgroups. (O.BL, R.BL,
CA.BL, E.BL, SZ.BL, O.DG, R.DG, CA.DG, SZ.DG, O.GBL,
BR.GBL, PZ.GBL, BL.SZ, BL.SS, DG.SS, G.SS, BL.SO, DG.SO).
Also O.R, CU.R, O.HR, CU.HR for Province = "AL" or "SK".

4. > - 200 All remaining soil Great group + Subgroup combinations, where
soil drainage is not poor or very poor.

5. Any soil where drainage is poor.

6. Any soil where drainage is very poor (if very poorly drained soils
are not set for exclusion).
___________
1
P-PE is Precipitation - Potential Evapotranspiration. All values are negative, in millimetres. These
are the approximate equivalent ranges, adopted from the Climatic Moisture Index Map
(Agronomic Working Group, 1992).

3-9
Table 3-2b. Inherent soil quality constraints matrix for available porosity-aeration sub-element

High
None(0) None(0) None(0) None(0) Slight(1) Moderate (2)
Medium
12 1 None(0) None(0) None(0) None(0) Slight(1) High (3)
Low
81 None(0) None(0) None(0) Moderate(2) High(3) High (3)
1
Very Low 4
None(0) Slight(1) Moderate(2) High(3) High(3) High (3)
1
Soil Climatic Class 1 2 3 4 5 6
P-PE (<-350) (-350 to - (-300 to - (> - 200) All poorly All very
300) 200)
Soils Brown D.Brown Black, All others drained poorly
D. Gray, drained 3
G.Br. Luvisol

1
Crop specific threshold values for the Porosity (cm of air) within the rooting zone can be
set within the CROP.LOG section of the ISQ_PRO program. Default values for the
NATIONAL LOG are shown here.
2
Soil Climatic classes are defined in Table 3-2a.
3
This class is used only where the program has been set to rate very poorly drained soils
(CROP.LOG).

The total aeration porosity values are then added for all valid soil layers within the surface 20 cm.
A minimum threshold value of 1 cm of air within the surface layer (5% of the surface soil volume)
must be reached; if this does not occur, the soil is considered as excluded. The total aeration
porosity within the total rooting depth is then calculated and assigned to one of four porosity
classes based on threshold values specified by the user. The soil climatic class is also determined,
based on the criteria in Table 3-2a. The final rating for the ISQ aeration porosity sub-element is
then assigned, based on the aeration porosity and climatic classes matrix in Table 3-2b.

3-10
Table 3-2c. Inherent soil quality constraints matrix for available porosity - water capacity
sub-element

Exclude(7) Exclude(7) Exclude(7) Moderate(2) None(0) None(0)

Very Low 5 1 — Exclude(7) Exclude(7) High(3) Moderate(2) None(0) None(0)

A Low 61 — Exclude(7) High(3) Moderate(2) Slight(1) None(0) None(0)

W Medium 7 1 — High(3) Moderate(2) Slight(1) None(0) None(0) None(0)

C High 121 — Moderate(2) Slight(1) None(0) None(0) None(0) None(0)

T Very High 151 — Moderate(2) Slight(1) None(0) None(0) None(0) None(0)

1 2 3 4 5 6
Soil Climatic Class2 (<-350) (-350 to -300) (-300 to -200) (> - 200) All poorly All very
P-PE Brown D.Brown Black, All others drained poorly
Soils D.Gray, drained3
G.Br. Luvisol

1
Crop specific threshold values for available water holding capacity (AWHC), in cm of water within the
rooting zone. Default values for NATIONAL LOG are shown here.
2
Soil Climatic classes are defined in Table 3-2a.
3
This class is used only where the program has been set to rate very poorly drained soils (CROP.LOG).

The available water holding capacity (AWHC) is calculated for each soil from the total silt, clay
and very fine sand values for each soil layer in the Soil Layer File. This is multiplied by the
horizon thickness and a total value is accumulated to either the maximum rooting depth, or some
restricting layer above the maximum rooting depth. Values for AWHC are calculated as shown in
Table 3-2d.

The total AWHC value was then assigned to one of six classes, based on threshold values specified
by the user.

3-11
Table 3-2d. Relationship between available water-holding capacity (AWHC) and very fine
sand (VFS), silt and clay. Approximate USDA texture classes are provided for
reference. Adopted, with modifications, from Table 4-1 (Agronomic Working
Group, 1992)

Texture1 0.5*(%VFS)+%Silt+%Clay2 AWHC (mm/m)

Organic 3 0 700
S > 0 to 10 40
LS > l0 to 20 60
SL > 21 to 40 100
L, VFSL > 40 to 60 150
Si, SiL > 60 to 70 170
SiL > 70 to 75 180
SiCL, CL > 75 to 80 190
SiCL > 80 to 85 200
SiC, C > 85 to 100 225
__________
1
USDA soil texture classes indicated here are only approximate. Several classes may overlap due to
varying percentages of silt and clay.
2
For soils with more than 0% by volume of course fragments, AWHC estimates are reduced by the
volume percentage of coarse material, a Soil Layer File attribute.
3
Organic soils are recognized by sand+silt+clay = 0 in the Soil Layer File data.

The soil climatic class is also determined, based on the criteria in Table 3-2a. The final ISQ rating
for the Available Porosity - Water-Holding Capacity sub-element must then be determined based
on the 6 soil climatic (Aridity) classes, as shown in Table 3-2c.

Nutrient retention

Nutrient retention was identified as the second soil quality element for the crop production
function in recognition of the important role played by the soil in accepting, retaining and
releasing nutrients for plant growth. This ISQ element is estimated by calculating the average
cation exchange capacity (CEC) over the surface 20 cm. The form of the calculation for each
horizon is:

CEC (meq/100g) X Bd (g cm-3) X Base Fraction (%) X Depth (cm)

3-12
These values are averaged over the surface 20 cm to give a number with final dimensions of (Meq
100 cm-2 cm-1). The rating used for this element is summarized in Table 3-3.

Table 3-3. Rating class limits for ISQ element-nutrient retention

Classes Cumulative CEC

No Constraint 25
Slight Constraint 19
Moderate Constraint
12

Physical rooting conditions

Rooting conditions are based on the depth to root impeding layers. Research is under way (e.g. the
soil structure and non-limiting water range studies described later in this report), to determine the
physical conditions which restrict or impede root penetration. New results will be incorporated into
the procedures as they become available but in its current implementation, roots are restricted by
the presence of an: ortstein layer, fragipan layer, duric layer, placic layer, lithic layer, most
restricting layers caused by high bulk density or coarse fragments in excess of 10%. The rating is
based on the depth at which the restricting condition occurs, as summarized in Table 3-4.

Table 3-4. Depth class limits for ISQ element-physical rooting conditions

Classes Depth to Impeding Layer

No Constraint > 80 cm
Slight Constraint > 55 cm
Moderate Constraint > 30 cm
High Constraint 20-30 cm

Chemical rooting conditions

Chemical conditions are defined in terms of an adequate depth of soil with pH and salinity within
tolerable limits. Salinity, as estimated by electrical conductivity, and pH are each evaluated
independently and the worst rating of the two is then assigned as the overall ISQ chemical rooting
conditions element rating. Both conditions are calculated in a similar fashion.

The depth weighted average pH for generic crops is computed from the values of all soil horizons
within the surface (0-20 cm) and the subsurface (20-80 cm) zones. The average surface layer pH,
weighted by horizon thickness, is then compared to the threshold values. If this average pH is

3-13
Fig. 3-2. Dominant soils in SLC polygon of southern Manitoba.

3-14
Fig. 3-3. Inherent soil quality (ISQ) elements map of southern Manitoba.

3-15
within the range of 6.0 to 7.3, the soil component is given a "no" constraints rating. If this value
extends beyond this range but does not fall below 5.5 or exceed 7.7, the soil is given a "slight"
constraints rating. Similarly, between 5.0 and 8.1 the rating is "moderate" and between 4.0 and 9.5
the rating is "high".

In the subsurface, pH values between 5.5 and 7.7 are considered to have no constraints while those
outside this range and extending to 5.0 and 8.1 have a slight constraint and those extending to 4.0
and 8.8 have moderate constraints.

Overall ratings are assigned as a combination of surface and subsurface ratings, as indicated in
Table 3-5. The overall pH rating is biased towards the surface rating.

Salinity is evaluated in the same fashion as pH. The weighted average of the electrical conductivity
is calculated for all soil layers within the surface and subsurface zones. These are assigned ratings
based on the threshold EC values. The overall salinity rating for the soil is calculated by comparing
these two ratings, as shown in Table 3-5.

Table 3-5. Surface/subsurface constraints matrix for pH and salinity

SURFACE RATING (CODE)

None Slight Moderate High
SUBSURFACE None, Slight None Slight Moderate High
RATING Moderate Slight Slight Moderate High
High Moderate Moderate Moderate High

Soil Quality Susceptibility to Change (SQS) Assessment Procedures

Soil-modifying processes such as erosion by wind or water, loss or accumulation of organic matter,
and soil salinization exert their effect either individually or in combinations by removing material
from the soil or adding to it. The result is a change to one or more soil attributes and this change
frequently impacts on the capacity of the soil to function. Studies described elsewhere in this report
deal with the development of capabilities to estimate or predict various processes which modify
soil quality. As these capabilities become fully developed, they will be incorporated into the GIS
system capability for assesing soil quality. In the interim, attention is focused on assessments using
a more qualitative approach to estimate soil quality susceptibility. It consisted of identifying
individual soil characteristics and conditions of land management which are likely to make a soil
area more susceptible to quality change than the surrounding soils.

The specific management and soil characteristics which were selected to indicate susceptibility to
change were based in part on the soil-modifying processes and in part on the kinds of information
available. For the process of erosion, measures of soil thickness, slope, and land management
characteristics related to extent of crop and residue cover were chosen, while susceptibility to soil
quality change from organic matter change was characterized by thickness of the surface

3-16
Fig. 3-4. Land cover map of southern Manitoba.

3-17
horizon and the initial value of organic matter in the soil. Cropping and management practices
which include more intensive cultivation and exposure of the soil were considered likely to result
in decline in organic carbon. Each of these factors was considered to be relatively independent; a
threshold level was established for each and areas which exceeded the threshold were mapped.
Table 3-6 provides details of the factors used and the specific threshold levels.

Table 3-6. Criteria and limiting (threshold) values used to indicate SQS

Data Sources Characteristics/ Limiting Values Modifying Processes

Attributes Affected
Biophysical
Attributes shallow topsoil A horizon thickness <=15 water erosion and
cm organic matter decline
Soil Landscapes of low organic carbon organic carbon of A
organic matter decline
Canada (SLC), content of topsoil horizon < 1%
Canadian Soil Carbon
shallow effective depth depth to impenetrable water erosion and
Database, Soil Layer
layer <= 60 cm organic matter decline
File (SLF) and Soil
Name File (SNF) etc. steep surface slope slope steepness > 9% water erosion
high intensity of land use area under crop > 75% of
farmland and **row crops organic matter decline
> 60% of cropped land
Land Use and
Management
Attributes

CENSUS of Agricul- low level of conservation
ture 1986, 1991), tillage and no-till or use of
AVHRR land cover summerfallow
data (1989) increasing intensity of land
use
**land prepared for water erosion and
seeding by conservation organic matter decline
and < 20% no-till > 20%
**increase of cropped land organic matter decline
and row crop > 10%

* applicable to prairie area

** applicable to southern Ontario

The susceptibility to change was estimated by adding the factors together. The level of susceptibility
of soil quality to change from water erosion or loss of soil organic matter was estimated as highest
where the greatest number of conditions occurred on the same parcel of land. Maps of these
conditions can be used to identify study areas where soil quality change is most likely.

3-18
 RESULTS AND DISCUSSION

Inherent Soil Quality (ISQ)

ISQ has been estimated for the four elements of the crop production function; available porosity,
nutrient retention, physical rooting conditions, and chemical rooting conditions. For southern
Manitoba, these estimates were calculated using data for the dominant soil from the Soil Landscapes
of Canada map and the associated soil names and soil layer file. Fig. 3-2 provides an indication of
the extent of coverage which has been achieved by restricting the analysis to only the dominant soil.
The results of the ISQ assessment are illustrated in Fig. 3-3. At this stage in soil quality assessment,
each of the elements of ISQ is mapped separately; ultimately, they will be combined into an
integrated assessment of the inherent soil quality for the crop production function.

It should be noted that the ISQ assessment procedures are at a prototype stage of development. They
will undoubtedly go through a series of revisions and refinements before the final ISQ assessments
are available. Currently, work is in progress to review these prototype ISQ assessments with local
experts to verify that the results agree with their personal experience in the region. In addition, we
are conducting a sensitivity analysis of the algorithms to identify the particular soil attributes which
have the greatest influence on the results. This effort is intended to assess both the algorithm and
also the quality of the data used.

The best assessments of soil quality which can be expected at national and broad regional scales
(e.g. 1:1,000,000) will provide highly generalized estimates of soil quality and indicate areas where
more study is required. Areas within Manitoba have been identified for more detailed study. For
these areas, large scale (1:20,000 and 1:50,000) land resource data will be used with the ISQ
procedures as finally developed to estimate ISQ elements at these larger scales. The comparison of
these estimates to the broad regional assessments will provide another validation of the procedures.
This assessment will be similar to the study carried out in Ontario (MacDonald et al., 1993) which
showed a good general level of agreement between the detailed and generalized assessment but also
the possibility for significant divergence within individual soil landscape polygons.

Soil Quality Susceptibility to Change (SQS)

Soil quality susceptibility to change (SQS) has been assessed by establishing threshold values for
individual soil and landscape properties to characterize the biophysical susceptibility to change
(Table 3-6a). The susceptibility to change, which is mediated by land use and management, has been
estimated by establishing relatively arbitrary thresholds (summarized in Table 3-6b) for information
collected in the Census of Agriculture. Both the criteria and the threshold values are tentative and
are currently being compared with analyses being undertaken in related projects, particularly the
sustainable land management analysis being reported by Cann et al. in a later chapter of this report.

The Census data are summarized on the basis of consolidated census subdivisions (CCSD) which
correspond to townships in Ontario and to Rural Municipalities in Manitoba. This means that the
spatial reporting unit for the land use and management information was quite different from the soil

3-19
landscape polygons on which the land resource data were compiled. The GIS provided the essential
analytical capability necessary to combine the data from these two sources. AVHRR data (Fig. 3-4)
provided broad classes of land use but did not show detail of land management.

For Manitoba, the Census data were available for 1986 and, for Ontario, Census data were obtained
for both 1986 and 1991. Because of the availability of data for two Census years and also due to
the greater range of land use intensity and management, the Census data for Ontario provided a
greater variety of SQS conditions influenced by land use and management. Soil quality susceptibility
to change estimates for southern Ontario (Fig. 3.5) are based on (a) biophysical attributes; and, (b)
Census of Agriculture land use and management attributes.

Integration of Inherent Soil Quality and Soil Quality Susceptibility to Change

While the individual elements of soil quality and the biophysical and anthropogenic assessments
of susceptibility of soil quality to change provide information about the quality and sustainability
of the resource base, it is the combination and subsequent interpretation of these results that has the
potential to provide insights into the current status and trends in soil quality. To demonstrate some
of the aspects which might be considered, four polygons have been selected from the agricultural
region of southern Ontario (the specific polygons are identified on Fig. 3-5). The ISQ rating of these
polygons is high (Table 3-7a,b); each element showing either no or slight constraint. Two of the
polygons (27 and 64) have soil and landscape characteristics which appear to be susceptible to
change from decline in organic matter and the other two (82 and 86) have properties of slope and
horizon thickness which suggest that they are susceptible to the process of erosion by water and that,
because the surface horizon is relatively shallow, the impact of the erosion process on soil quality
will be substantial.

The area of each census subdivision which intersects each polygon is determined by an intersection
using the GIS and the proportion of each soil landscape polygon in each CCSD is determined. The
CCSD information is then summarized (Table 3-7a,b) to indicate the land use and management
factors present which make the soil susceptible to soil quality change. Clearly, these areas have
relatively uniform biophysical characteristics but show a range of land use practices which will make
the areas more or less susceptible to changes in soil quality.

One of the difficulties associated with the interpretation of the data is the difference in boundaries
between the natural units (soil landscape polygons) and the administrative areas (CCSD's) on which
the land use and management data are collected. Work is in progress to regroup the basic (farm
level) census data and summarize them on the basis of natural units. In the interim, however,
because the data are stored and managed within a GIS, it is a straightforward task to reassess the
census information with criteria which provide a closer spatial fit. Table 3-8 provides this
information based on CCSD's which are 50% or more within the soil landscape polygon. This
criteria appeared to encompass approximately the same proportion of the polygon as does the
dominant soil class. In addition, because the data were stored by individual CCSD, the actual values
were calculated for the attributes of soil management and land use. The results are all expressed on
the basis of the 1991 census of agriculture.

3-20
Fig. 3-5. Soil quality susceptibility to change maps of southern Ontario for
biophysical and land use and management attributes.

3-21
Table 3-7a. SQS to organic matter decline of selected SLC polygons in southern Ontario

SLC Combined ISQ Soil and landscape

Land use and management attributes Implications
polygon rating characteristics
number
Occurred Area Occurred Area (%)
(%)
0027 Slight constraint A horizon thickness <= 15 cropped area > 75% & row crops > 60% i) Both polygons are more
cm - organic carbon of A - conservation tillage and no tillage < 20% susceptible to organic matter
horizon 70 - increase of cropped land > 10% decline
< 1% - increase of row crop > 10% 11.4

c ropped area > 75% & row crops > 60% 5.1 ii) Polygon # 27 has
- conservation tillage and no tillage < 20% relatively higher
- increase of cropped land > 10% susceptibility to organic
matter decline
cropped area > 75% - row crops > 60%
- conservation tillage and no tillage < 20% 80.3
0064 Slight constraint A horizon thickness <= 15 cropped area > 75% - row crops > 60% 3.2
cm organic carbon of A
horizon 70
< 1%
cropped area > 75% & row crops > 60%
- conservation tillage and no tillage < 20%
-increase of cropped land > 10% 6.0

cropped area > 75% - row crops > 60% 25.6

- conservation tillage and no tillage < 20%
14.3
cropped area > 75% - row crops > 60%
- increase of row crops > 10 % 13.0

cropped area > 75% - row crops > 60% 39.1

conservation tillage and no tillage < 20% 2.0

SQS not indicated

CCSD - Census Consolidate Sub-division

SLC - Soil Landscape of Canada (maps)

3-22
Table 3-7b. SQS to water erosion of selected SLC polygons in southern Ontario

SLC
Polygon Combined ISQ Soil and landscape Land use and management attributes Implications
Number rating characteristics

Area
Occurred Area (%) Occurred
(%)

0082 No constraint A horizon thickness <= 15 cm conservation tillage and no tillage < 20%
8.7 i) Both polygons are more
-slope steepness > 9% - increase of cropped land > 10%
susceptible to water erosion
70 conservation tillage and no tillage < 20% 67.8

SQS not indicated 23.5 ii) Polygon #86 has

relatively higher suscepti-
bility to water erosion than
polygon #82.
0086 Slight constraint A horizon thickness <= 15 cm conservation tillage and no tillage < 20%
- slope steepness > 9% - increase of cropped land > 10%
60 34.3
conservation tillage and no tillage < 20% 44.6

SQS not indicated 21.1

SLC - Soil Landscape of Canada maps

CCSD - Census Consolidate Sub-division

3-23
Table 3-8. Susceptibility to Soil Quality Change - Land use and management factors
[based on data from Census subdivisions (CCSD) which are greater than 50% within the SLC polygon]

Soil Landscape (SLC) Polygon Number 27 64 82 86

Polygon area (Ha)
71,1731 215,445 73,854 64,557
Number of CCSD's intersected
11 30 10 7
Number of CCSD's > 50 % within SLC
polygon (CCSD>50) 3 10 2 2

% of SLC polygon covered by CCSD>50

64 74 54 60
Cropped land in CCSD>50 in 1991
42,691 152,532 10,902 13,301
Cropped area as a % of farmland in CCSD>50
81 83 45 45
Tame hay as a % of cropland in CCSD>50
8 19 65 70
Row crops as a % of cropland in CCSD>50
80 53 4 4
Conservation tillage as a % of cropland in
CCSD>50 12 14 18 16

Rented land % in CCSD>50

34 23 30 26
1986 cropland as a % of 1991 in CCSD>50
98 99 113 99
1986 row crop area as a % of 1991 in
CCSD>50 79 83 area too small area too small

3-24
These calculations show that both polygons 27 and 64 are intensively cropped but that row crops
are a larger component of the cropping program in polygon 27, while tame hay makes up more of
the land in polygon 64. The results also suggest that annual crops and, in particular, row crops, are
not so important in polygons 82 and 86. From the standpoint of land use and management, the two
polygons are quite similar and both have hay as a major component of the land use. Analysis of this
sort can be used to separate areas which are relatively susceptible from the more stable areas and
target areas for more detailed study.

SUMMARY

The current prototype systems for assessment of inherent soil quality and soil quality susceptibility
to change were designed to use available databases and capabilities of existing GIS systems. The
systems were designed to be applied to national level assessments. The preliminary national
assessments for Ontario and Manitoba were in agreement with subjective estimates by experienced
soil researchers and surveyors.

The results of the assessments may be combined with other environmental information: i) to assess
sustainability of current soil and land management systems; ii) to target current and potential
'problem areas' of soil quality change and land degradation for monitoring and detailed studies; iii)
for State of Environment (SOE) reporting.

Technically, the systems were designed to be flexible enough to:

i) incorporate additional data layers, e.g. topography and land use,
ii) adjust the rating criteria and algorithm for specific crops or other land use requirements,
iii) be used at more detail levels, e.g. regional, watershed and farm levels,
iv) combine with other analytical tools or models, e.g. USLE to assess the potential effects of
soil water erosion to soil quality at national and regional level. Results and models
developed in other studies reported herein may provide capabilities to enhance the ISQ and
SQS assessment procedures.

More precise estimates require detailed, large scale data as well as modelling procedures to
characterize soil modifying processes. However, the sequential assessment of soil quality and its
spatial representation is operational at national and regional scales using existing databases.

REFERENCES

Agronomic Working Group. 1992. Land suitability rating system for spring seeded small grains,
1992. Working Document. Centre for Land and Biological Resources Research, Agriculture
Canada, Ottawa, ON.

Allen, T.F.H., O'Neill, R.V. and Hoekstra, T.W. 1984. Interlevel relations in ecological research
and management: Some working principles from heirarchy theory. USDA Forest Service
General Teechnical Report RM-110. Rocky Mountain Forest and Range Experiment Station,
Fort Collins, CO. 11pp.

3-25
Brklacich, M. and MacDonald, K.B. 1992. Prototype agricultural land evaluation systems for
Canada: II. Selected applications and prospects. Soil Use Manage. 8: 8-15.

Hudson, B.D. 1990. Concepts of soil mapping and interpretation. Pages 63-72 in Soil Survey
Horizon, Fall.

Larson, W.E. and Pierce, F.J. 1992. Conservation and enhancement of soil quality. In Evaluation
for Sustainable Land Management in the Developing World. Volume II. Technical Papers.
Bankok, Thailand: International Board for Soil Research and Management, 1991. IBSRAM
Proceedings No. 12.

MacDonald, K.B. and Brklacich, M. 1992. Prototype agricultural land evaluation systems for
Canada: I. Overview of systems development. Soil Use Manage. 8: 1-8.

MacDonald, K.B., Wang, F., Jarvis, I. and Fraser, W. 1993. GIS procedures to assess soil quality
at regional and national scales. CLBRR contribution No. 92-213. Pages 325-334 in GIS '93
Symposium. Vancouver, BC.

Shields, J.A. and Sly, W.K. 1984. Aridity indices derived from soil and climatic parameters.
Technical Bulletin 1984-14E. Research Branch, Agriculture Canada, Ottawa, ON.

Wang, C, Coote, D.R., and Acton, D.F. 1991. A proposed mineral soil quality classification system
for arable land. Pages 54-62 in S.P. Mathur and C. Wang, Eds. Soil quality in the Canadian
context-1988 discussion papers. Tech. Bull. 1991-1E. Research Branch, Agriculture Canada,
Ottawa, ON.

3-26
CHAPTER 4
A LAND USE ANALYSIS AND MONITORING SYSTEM FOR SOIL
QUALITY ASSESSMENT

J.C. Hiley1 and E.C. Huffman2

1
Agriculture and Agri-Food Canada, Research Branch, CLBRR Alberta Land Resource Unit,
Edmonton, Alberta
2
Agriculture and Agri-Food Canada, Research Branch, Centre for Land and Biological
Resources Research, Ottawa, Ontario

INTRODUCTION

Over the past decade, Canadian concern for the state of the environment and the prospects for
sustainable development has focused considerable attention on the affect of agricultural activities
on soil quality. It is recognized that some agricultural land use and management practices accelerate
soil degradation processes (Standing Committee, 1984; Alberta Agriculture, 1985) and it appears
that some adjustment of land use may be required in order to maintain the long-term health of our
agricultural soils and production industry. In order to document the relationship between land use
and soils, however, and to develop strategies for reducing the stress on the environment, an efficient
methodology for analysing and monitoring current land use over large areas is required.

Soil scientists, geographers and specialists in land evaluation, land degradation and rural planning
have been promoting the need for current and accurate land use and management data in order to
support our efforts to manage resources (Halstead and Dumanski, 1977; University of Guelph, 1978;
Coote et al., 1981; Huffman and Dumanski, 1985; Hiley and Wehrhahn, 1991; Jeck et al., 1990a).
As a result, a number of different land use inventory techniques based on airphotos, field survey,
census data and satellite imagery have been developed to address the issue.

Current efforts to analyze and monitor agricultural land use activities, however, find these
techniques lacking in several ways. For example, the interpretation of airphotos is accurate and
detailed, but it is also slow, expensive and oriented to land cover, with little opportunity for
interpretation of socioeconomic or farm structural characteristics. Dedicated field surveys and
farmer interviews can provide the necessary holistic view but, again, the expense and time required
restricts their use to small areas and infrequent time intervals. Classification of satellite imagery is
a versatile approach, with a variety of radiometric and spatial resolutions, regular and repeat
coverage and suitability for digital manipulation in Geographic Information Systems (GIS), but it
suffers from a lack of socioeconomic interpretability and is therefore low in analytical potential. The
national Census of Agriculture, with its coverage of all farms every 5 years and wide variety of
variables has tremendous potential for analytical studies over large areas, but it is restricted by a lack
of locational accuracy. Also, census data has traditionally been available only on the basis of
politically defined areal units such as Enumeration Areas, Census Subdivisions or Crop Reporting
Districts, and is thus difficult to interpret in a natural or biophysical framework.

4-1
It was within this setting that a system for land use analysis and monitoring was developed as a
component of a soil quality evaluation program. Clearly, the approach chosen to report on the state
of soil quality in Canada, described in the first chapter of this report, required a land use database
and a methodology of manipulation and analysis that would serve a variety of needs from provincial
and national evaluations to local and site specific modelling exercises. Recognizing that any one
method would not likely serve all needs, it was necessary to capitalize on the strengths of different
techniques and, by focusing on information needs at different scales of analysis, provide relevant
data in an efficient and timely fashion. The development of the capability to analyze and monitor
agricultural land use activities and management practices within a landscape framework, using the
two approaches of census data interpretation and remote sensing, is addressed in this chapter.

GOALS AND OBJECTIVES

The objective of this study was to develop a land use analysis and monitoring system for the
agricultural regions of Canada. The orientation of the research was toward supplying the specific
land use data requirements for soil quality assessment, but the requirement to develop databases,
methods of analyses and capabilities for integrated agricultural evaluations was also recognized.
Since soil quality assessment requirements relate to defining and measuring soil quality and its
susceptibility to change, the emphasis in the development of the land use and analysis system was
on acquiring data that reflected the type and intensity of farmland use. Developing and testing
procedures for the acquisition, manipulation, analysis and presentation of land use data, and
developing expertise and technical capabilities for its integration with biophysical and climatic
information were the principal activities. A wide range of land use characteristics that were
considered to have an effect on soil quality, including farm type and structure, economic and
financial conditions, crop type and rotations and land management features were included.

Specific goals of this study were to develop a system that:

i) is applicable in all parts of the country,

ii) is oriented toward land management, farming activities and socioeconomic conditions rather
than land cover,
iii) is applicable at a variety of scales from local (1:25k) to national (1:5m),
iv) incorporates a time dimension to facilitate monitoring, and
v) is functional within a biophysical (landscape) framework.

SYSTEM REQUIREMENTS

System requirements relate to the type and source of data to be monitored and the spatial
representation of databases within the system. These needs were identified through consultation with
potential clients of a land use analysis and monitoring system, a process that involved a
questionnaire survey of 22 agencies and 12 participants in the Soil Quality Evaluation Program, as
well as the formation of a 14-member national Advisory Committee.

4-2
All potential clients used the concept of 'study scale' to identify the characteristics of data required
from the system (Table 4-1). Indications were that national and regional land evaluation, land use
monitoring and general soil quality/land use studies required a range of crop, farm type and
socioeconomic information that could be updated approximately every 5 years. High spatial accuracy
was not required, and it was agreed that such data could be acquired from secondary sources such
as the Census of Agriculture. However, respondents indicated that the calibration of degradation
process models and the measurement of changes in soil quality over small areas required
observations with a high degree of spatial accuracy, in many cases repeated several times within a
single growing season or in consecutive seasons. It was agreed that remote sensing could provide
primary source land use information appropriate for some of those needs.

The client survey, supported by committee input, showed that the preferred method of organizing
land use data was also affected by study scale. Studies over large areas would be served best by
information that refers to homogeneous landscape units in order to hold physical conditions as
constant as possible in analytical and descriptive procedures. For detailed process-model work, the
area of concern tends to be a site or a field and land use data should correspond to that area. It was
the opinion of the potential clients and the advisory committee that the Soil Landscapes of Canada
(Shields et al., 1991), which are relatively uniform soil-based units of approximately 50-60,000
hectares, would be an appropriate base for organization of national level land use data. These
landscape units are also nested within larger 'Agricultural' or 'Land Resource Areas' (ARA's or
LRA's) which allows the aggregation of data for studies at a coarser resolution. For site related
research, the pixel resolution of the appropriate satellite imagery (eg. 10, 20, 30 or 80 metres) serves
as the spatial unit.

Table 4-1. Characteristics of data from the system

Broad Area Assessments Small Area Assessments

Principal use of data Land Evaluation Models Calibration of Process Models
Scale 1:5 Million to 1:250,000 1:250,000 to 1:5,000
Scope National to Regional Regional to Local
Spatial Resolution Landscape Units Sites, Fields
1. Cropping Systems 1. Crop Rotations
2. Land Management Variables 2. Tillage Systems
Type of Data
3. Socioeconomic Variables 3. Production Practices
4. Farm Structure Variables 4. Conservation Practices
Type of Data Descriptions, Comparisons of Large Descriptions, Statistical Correlations,
Analysis Areas, Long-term Change Variability
Time Interval to Update 5 years 3 Times in Growing Season
Source of Data Secondary Primary

4-3
SYSTEM DESIGN

A review of Canadian literature showed various methods of summarizing data from the national
Census of Agriculture on the basis of biophysical landscape units (Kraft, 1980; Huffman, 1988).
Consultation with Statistics Canada personnel indicated that a standard method for all regions of
Canada was not likely to be appropriate. The fundamental problem, not likely to be resolved in the
near future, relates to the integration of different land survey approaches and geographic coordinate
systems used in database management.

Similarly, different methods to obtain land use data over small areas have been documented. One
method, combining airphoto interpretation and site inspection, has been used extensively in regional
studies, and an adaptation of the approach to provide land cover, socioeconomic and farm activity
information was successfully demonstrated at the field level in several regions of Canada (Huffman
and Dumanski, 1985). Another approach, using satellite imagery to identify land cover, has also
been employed successfully throughout the agricultural regions of Canada (Ryerson et al., 1979;
Wilson, 1986; Energy, Mines and Resources Canada, 1987). In the context of a land use analysis
and monitoring system with national capability, a significant limitation in the form of prohibitive
costs renders the approach of airphoto interpretation and field survey impractical except for specific
projects with requirements for highly accurate data. On the other hand, the repetitive, large area
coverage and relatively timely aspects of satellite imagery classification make it an appropriate
approach.

A system design containing two components was adopted. The first component addressed broad area
assessments and was based on the linkage of the 1991 Census of Agriculture to biophysical
landscape units using the most appropriate method available within a particular region. The second
component dealt with small area assessments and focused on the development of a satellite image
analysis capability within Agriculture and Agri-Food Canada. It involved the purchase of image
analysis hardware and software and the training of staff.

METHOD

The system uses three different methods to link land use and management data to biophysical
landscape units in different parts of the country. A brief discussion of these regional approaches is
presented here, followed by a description of the agricultural data modules.

Regional Approaches

In the Prairie provinces, match-merging of farm headquarters locations with biophysical units was
the chosen method, and Soil Landscapes of Canada (SLC) maps provided the spatial stratification.
Digital files of SLC polygons were combined with a Dominion Land Survey (DLS) file in a GIS and
the extent of each polygon was defined on the basis of square-mile 'sections'. A file containing the
DLS location of the headquarters of each farm in the Census (maintained by Statistics Canada) was
merged with the polygon file, thereby assigning farm headquarters to SLC polygons. The merge
produced a link file that was used to summarize farm-level data by SLC polygons. The procedure
is summarized in Table 4-2.

4-4
In central Canada (Ontario and Quebec), the DLS is an irregular design based on Counties,
Townships, Lots and Concessions, with "Gores" and "Broken Fronts" due to surveys originating
along natural watercourses. Due to this irregularity, the capture of farm headquarters location is not
consistent, thus making the linking of headquarters to polygons impractical. In this area,
match-merging of Census Enumeration Areas (the smallest area at which census data is aggregated
for reporting, and covering generally from 30 to 75 farms) within composite biophysical/land cover
units was the approach used. An SLC map in a GIS is overlain with a summary land cover map
interpreted from Advanced Very High Resolution Radiometer (AVHRR) satellite data to produce
a data layer showing extensively cultivated SLC polygons, which is then merged with a file
containing Enumeration Area (EA) centroids. This merge produces a link file assigning each EA and
its associated data to a composite landscape unit.

System development in eastern Canada involved match-merging of EA's to Land Resource Areas
(LRA's). The latter are groups of SLC polygons within common agroclimatic zones. The LRA file
is maintained as a coverage in a GIS and is overlain with an EA centroids file to produce a
preliminary link file. Local agricultural specialists then reassign some selected EA's to ensure that
there is at least 70% overlap between EA's and LRA's and, thus, that local land use patterns are
appropriately represented.

Table 4-2. System capabilities for broad area assessments in Canada

Region Prairies Central Eastern

Scale 1:1 Million 1:1 Million 1:2 Million
Soil Landscapes of Soil Landscapes of Land Resource
Biophysical Spatial Unit
Canada (SLC) Canada (SLC) Areas (LRA)
Additional Bases None Land Cover Map None
Census Unit Farm Headquarters Enumeration Area Enumeration Area
Compilation Technique Match records GIS overlay GIS overlay
Additional Decision 70% overlap, refined by
None None
Criteria expert opinion
Land Use Classification 1991 Census 1991 Census 1991 Census

Data Modules

The Census provides a wealth of information pertaining to agricultural production in Canada. As

such, it was possible to include more than 100 variables for each landscape unit within the land use
analysis and monitoring system. Within the scope of this chapter, only a general discussion of the
data modules is possible. In that each regional method uses standard census variables, the following
discussion is applicable to all regions.

Based on the results of the questionnaire survey and committee review, five data modules were
selected. They include: farm structure, crops and land use, land management, livestock and
economics. The land use, management and economic modules are directly relevant to assessments
of soil quality. The crop/land use module addresses the need for information on production

4-5
characteristics and almost 50 variables are organized into the categories of summary land use, major
field crops and major specialty crops. Tillage systems can be described from data in the land
management module, which summarizes 20 variables covering chemical inputs, tillage practices,
conservation practices and land degradation concerns. The economic module contains summaries
of capital value, cash flow and labour inputs and includes approximately 30 variables.

RESULTS

Considerable work on the development of the land use analysis and monitoring system has been
completed, and the following section presents a summary of progress and research findings.

Component 1 (Broad Area Assessments)

This component is nearing completion for the Prairie provinces, with methods finalized and link
files prepared. Custom processing of the census data is nearing completion and a trial run of the data
stratification shows successful output. As an example of the type of results being obtained, Table 4-3
presents a comparison of several variables for three SLC polygons.

Table 4-3. A comparison of three SLC polygons in the Prairie Provinces, Canada

Red River, Vulcan, Assiniboia,

Variable
Manitoba Alberta Saskatchewan

Black Dark Brown Brown

Dominant Soil Type
Chernozemic Chernozemic Chernozemic
Surface form Level Rolling Undulating
Area (ha) 70103 224750 83078
Number of Farms 233 431 217
Farm Size (ha) 292 521 383
Cultivated Area (% of farmland) 95.5 77.5 81.7
Wheat (% of cult. area) 53.8 39.4 52.6
Summerfallow (% of cult. area) 0.4 36.0 44.9
Capital Value * ($/ha) 2317.95 1523.41 1042.00
Gross Margin ** ($/ha) 112.54 72.28 36.93

* total capital value of all farms (land, buildings, machinery, equipment and livestock) divided by
farmland area.
** total of all farm receipts minus total operating expenses divided by farmland area.

4-6
It is apparent that with a list of over 100 farm characterization variables and a spatial stratification
that subdivides Manitoba, Saskatchewan and Alberta into approximately 1000 landscape units, there
is considerable potential for analysis of spatial relationships between soil, soil quality and
agricultural activities. For example, comparisons and correlations can be made between and among
a variety of socioeconomic and biophysical variables in order to identify anomalous situations, which
could then be targeted for more detailed investigation. Similarly, locations with combinations of
characteristics that are known or suspected of being degrading to the environment can be identified
for action. The data could also be accessed in order to tailor policy or extension efforts to specific
concerns. Soil quality issues in each of the areas presented in Table 4-3 above, for example, could
be addressed in different ways, depending on the structure and focus of the particular farming
practices.

In addition to the SLC database, a classification and characterization of farms on the basis of
cropping systems has been developed for additional information on the type and variability of farm
enterprises within a region. This farm typing database is spatially organized on the basis of
Agricultural Resource Areas (ARAs), which are comprised of groups of SLC polygons. The smaller
scale of spatial organization is required in order to obtain sufficient numbers of farms in each type
and landscape unit to adhere to Statistics Canada's confidentiality provisions (minimum of 10 farms
per reporting unit). The ARA/Farm-type database duplicates similar ones developed for the 1981 and
1986 censuses (Kirkwood et al., in press) and thus presents a potential for analysing change in the
structure of agriculture within well defined landscape units.

For central and eastern Canada, development of the biophysically stratified census files has been
held up by the unavailability of EA centroid files. These GIS coverages will be overlain on SLC and
LRA maps to prepare files linking the census to landscape units, but the 1991 coverages are only
now becoming available. Some preliminary work in Ontario has been conducted using census data
on a Census Consolidated Subdivision (CCS) basis, but the size of these spatial units (approximately
equivalent to a Township) restricts their use in detailed spatial analysis.

In addition to the database development work, a generic Spatial Decision Support System (SDSS)
has been developed to assist in the management and presentation of data for the system. ELLY, the
Encyclopedia for Landscape and Land-indexed Inquiry, is a software package with a low-end GIS
functionality (Hiley et al., 1991) designed to run on PC-based computers. The package is currently
used within the project to manage the SLC maps and associated land resource data for the 10
provinces and the Yukon territory. With the completion of the development of this land use and
analysis and monitoring systems capability, the program will also be able to include the agricultural
production data modules referenced by biophysical unit. The software is distributed under private
licence and data developed within this system will be available through a separate request to
Agriculture and Agri-Food Canada.

Component 2 (Small Area Assessments)

The development of a remote sensing capability within Agriculture and Agri-Food Canada has been
achieved with the installation of necessary hardware and software and the training of personnel in
system operation and image analysis. A detailed research study completed in the Minnedosa area

4-7
of Manitoba has shown that the identification and mapping of 3-year crop rotations can be achieved
with 90% or greater accuracy using conventional Landsat (TM) imagery (Huffman, 1992). This work
was performed at a sub-field level of spatial precision and was successful at identifying land cover
features as small as 1 ha in size. It indicates the potential for use of satellite imagery in the
application of process models at the field and farm levels of detail.

Future research in the area of remote sensing will address the feasibility of using this technology to
supply data for specific evaluations and small area assessments, and in linking the results of satellite
imagery classifications with biophysical data. In particular, the integration of this component with
the broad area analysis and monitoring procedures established in Component 1 of the system will
be evaluated. In the initial stages, the census data stratified by SLC and ARA polygons will provide
not only the ability to analyze, monitor and assess soil/land use relationships, but will allow
researchers to "target" areas for more detailed and specific studies. It could also provide a
mechanism for scaling up research findings from a site to a landscape.

REFERENCES

Alberta Agriculture. 1985. Proceedings, soil at risk conference. Alberta Agriculture. Edmonton,
AB. 117 pp.

Coote, D.R., Dumanski, J. and Ramsay, J.F. 1981. An assessment of the degradation of
agricultural lands in Canada. Land Resource Research Institute, Research Branch, Agriculture
Canada. Ottawa, ON. 86 pp.

Energy, Mines and Resources Canada. 1987. Remote sensing for agriculture. User Assistance and
Marketing Unit, Ottawa, ON. 7 pp.

Halstead, R.L. and Dumanski, J., Eds. 1977. Land evaluation and systematic data collection.
Agriculture Canada, Ottawa, ON. 111 pp.

Hiley, J.C. and Wehrhahn, R.L. 1991. Evaluation of the sustainability of extensive annual
cultivation within selected agroecological resource areas of Alberta. Agriculture Canada,
Edmonton, AB. LRRC Contribution No. 91-18. 88 pp.

Hiley, J.C., Tajek, J. Toogood, K.E. and Milinusic, T. 1991. 'ELLY Factsheet'. Pages 240242 in
D.L. Beever and L.C. Marciak, Eds. Proceedings, First Interprovincial Soil and Water
Conservation Planning Workshop. Alberta Agriculture, Edmonton, AB.

Huffman, E. and Dumanski, J. 1985. Agricultural land use systems: An economic approach to
rural land use inventory. J. Soil Water Conser. 40:302-306.

Huffman, E. 1988. A description of physical and economic strategies of farming in the major soil
zones of the Canadian prairies. Pages 17-30 in J. Dumanski and V. Kirkwood, Eds. Crop

4-8
 production risks in the Canadian prairie region in relation to climate and land resources.
Research Branch, Agriculture Canada. Ottawa, ON.

Huffman, E. 1992. Procedures for the identification and mapping of cropping systems using
Landsat TM imagery. Unpubl. Ph.D. Thesis, University of Waterloo. Waterloo, ON. 127 pp.

Jeck, S.C., Moon, D.E. and Selby, C.J. 1990a. The LANDS system user's manual - Vol I -
Procedures (1st ed.). Agriculture Canada. Vancouver, BC. 149 pp.

Jeck, S.C., Moon, D.E. and Selby, C.J. 1990b. The LANDS system user's manual - Vol II -
Appendices (1st ed.). Agriculture Canada. Vancouver, BC. 164 pp.

Kirkwood, V., Bootsma, A., deJong, R., Dumanski, J., Hiley, J.C., Huffman, E.C., Moore, A.,
Onofrei, C., Pettapiece, W.W. and Vigier, B. 1994. Documentation of the database files
associated with the Agroecological Resource Area maps for Alberta, Saskatchewan and
Manitoba. Agriculture Canada. Ottawa, ON. 28 pp.

Kraft, S.E. 1980. The evaluation of rural land in Saskatchewan: a general background to the
evaluation of land and a description of a prototype data base for use in evaluation studies.
University of Saskatchewan. Saskatoon, SK. 110 pp.

Ryerson, R.A., Mosher, P. Waller, V.R. and Stewart, N.E. 1979. Three tests of agricultural
remote sensing for crop inventory in Eastern Canada: results, problems and prospects. Can. J.
Rem. Sens. 5:53-66.

Shields, J.A., Tarnocai, C., Valentine, K.W.G. and MacDonald, K.B. 1991. Soil Landscapes of
Canada - Procedures Manual and User's Handbook. LRRC Contribution No. 88-29. Agriculture
Canada. Ottawa, ON. 74 pp.

Standing Committee on Agriculture, Fisheries and Forestry. 1984. Soil at risk: Canada's eroding
future. The Senate of Canada. Ottawa, ON. 129 pp.

University of Guelph. 1978. Methodology study and development of a data base for rural land
evaluation in Ontario. Centre for Resources Development. Guelph, ON. 60 pp.

Wilson, D.A. 1986. The role of remote sensing in the Canada Land Use Monitoring Program
(CLUMP). Pages 947-956 in Proc. of the 10th Can. Symp. on Rem. Sens. Edmonton, AB.

4-9
CHAPTER 5
BENCHMARK SITES FOR ASSESSING SOIL QUALITY CHANGE

C. Wang1, B.D. Walker2, H.W. Rees3, L.M. Kozak4,

M.C. Nolin5, W. Michalyna6, K.T. Webb7, D.A. Holmstrom8,
D. King9, E.A. Kenney10 and E.F. Woodrow11

1
Agriculture and Agri-Food Canada, Research Branch, Centre for Land and Biological Resources
Research, Ottawa, Ontario
2
Agriculture and Agri-Food Canada, Research Branch, CLBRR Alberta Land Resource Unit, Edmonton,
Alberta
3
Agriculture and Agri-Food Canada, Research Branch, CLBRR New Brunswick Land Resource Unit,
Fredericton, New Brunswick
4
A griculture and Agri-Food Canada, Research Branch, CLBRR Saskatchewan Land Resource Unit,
Saskatoon, Saskatchewan
5
Agriculture and Agri-Food Canada, Research Branch, CLBRR Quebec Land Resource Unit, Ste-Foy,
Quebec
6
Agriculture and Agri-Food Canada, Research Branch, CLBRR Manitoba Land Resource Unit, Winnipeg,
Manitobaã7
7
Agriculture and Agri-Food Canada, Research Branch, CLBRR Nova Scotia Land Resource Unit, Truro,
Nova Scotiaã8
8
Agriculture and Agri-Food Canada, Research Branch, CLBRR Prince Edward Island Land Resource
Unit, Charlottetown, Prince Edward Island
9
Agriculture and Agri-Food Canada, Research Branch, CLBRR Ontario Land Resource Unit, Guelph,
Ontario
10
Agriculture and Agri-Food Canada, Research Branch, CLBRR British Columbia Land Resource Unit,
Vancouver, British Columbia
11
Agriculture and Agri-Food Canada, Research Branch, CLBRR Newfoundland Land Resource Unit, St.
John 's, Newfoundland

INTRODUCTION

Questions about trends in soil quality change in Canada and means of measuring those trends arose
in the late 1980's in response to the sustainable agriculture issue. Soil quality was reported to be
declining in many areas due to degradation processes such as erosion, compaction, salinization, and
acidification. The accelerated rates of deterioration were linked to farming practices.

Many studies relating change in soil properties important to crop production have been conducted
over the past several decades. However, most of these studies were narrow in scope and lacked
sufficient characterization of some soil properties and of some important land use and management
systems to serve as a basis for a comprehensive evaluation of change in soil quality.

5-1
Several approaches to the assessment of soil quality change have been reported. The first approach
(Martin, 1989; Tabi et al., 1990) involves the selection of specific sites where baseline levels of soil
properties considered important to soil quality are determined and soil quality change is evaluated
from periodic re-assessment of these properties. In the second approach, change to soil quality
attributes is predicted by applying relationships established from research sites. These predictions
enable an assessment of the direction and magnitude of change to an attribute based upon
information on the prevailing land management and other conditions of the area in question (Larson
and Pierce, 1991). The study described herein exemplifies the first approach. A network of sites will
be established that will enable a direct comparison of change in soil properties over time in relation
to prevailing agricultural land use and management practices across Canada. This will provide direct
measurement of change as well as an opportunity to verify change to soil quality that has been
predicted in the various other systems capabilities described elsewhere in this report.

The objectives of the national network of benchmark sites were:

i) to provide a baseline data set for assessing change in soil quality and biological productivity (i.e.
yields) of representative farming systems;
ii) to provide a means for testing and validating predictive models of soil degradation and for
evaluating the sustainability of current and proposed agricultural land management; and
iii) to provide a network of well-documented sites at which integrated multidisciplinary research
programs could be developed.

A NETWORK OF BENCHMARK SITES

Site Selection Protocol

- The criteria used to guide the selection of benchmark sites were:

- represent a major soil or climatic zone and/or ecological region;
- represent a typical physiographic region (landscape) and/or broad textural grouping of soils;
represent a major (or potentially major) farming system within a region;
- complement provincial priorities and opportunities;
- provide potential for the evaluation of the impact of a susceptible degradation process; occupy
about 5-10 ha, or a small watershed; and
- be located on cultivated agricultural land, as part of an actual farming system.

Location and General Description of the Sites

A network of 23 benchmark sites have been established across Canada (Fig. 5.1). The sites have
been selected to represent prevailing land use and management practices on representative
soil-landscapes across the country (Table 5-1).

5-2
Fig. 5-1. General location of the benchmark sites.
5-3
Table 5-1. Brief description of the benchmark sites

ECOLOGICAL SETTING PARENT MATERIAL AND

SITE NO. CROPPING SYSTEM TILLAGE SYSTEM
AND SOILS SURFACE FORM
South Coastal Pacific, Medium-textured fluvial.
01-BC Silage corn Tiled drainage, conventional tillage
Humic Gleysols Level
03-AB Parkland-Boreal Transition, Fine-textured lacustrine.
Cereals - canora - forage Conventional tillage
Dark Gray Soils Level
04-AB Northern Parkland, Medium-textured fluvial over till.
Cereals - forage - oilseed Conventional vs no-till
Black soils Undulating
05-AB Prairie-Parkland Transition, Medium-textured fluvial over till.
Canora - wheat - fallow Conventional tillage (cultivators)
Dark Brown soils Hummocky
06-AB Southern Prairie, Medium-textured lacustrine over Wheat (seed) - beans - sugar
Irrigated conventional tillage
Brown soils till. beets
07-SK Southern Prairie, Medium-textured loess over till.
Wheat - fallow Conventional tillage
Brown soils Undulating, dissected
08-SK Mixed-grass Prairie, Fine-textured lacustrine. Extended rotation,
Conventional tillage
Dark Brown soils Undulating mainly wheat
09-SK Prairie-Parkland Transition, Medium-textured till. Extended rotation,
Conventional tillage
Black soils Hummocky continuous cereal
10-SK Southern Boreal, Medium-textured till.
Cereals - canola Conventional tillage
Gray Luvisols Undulating, dissected
11-MB Eastern Parkland, Fine-textured lacustrine.
Continuous cereals No-till
Humic Gleysols Level
12-MB Eastern Parkland, Medium-textured lacustrine.
Wheat - canola Minimum tillage
Black soils Level
13-ON Southern Temperate, Fine-textured lacustrine.
Corn - soybean - wheat Minimum tillage
Gray Brown Luvisols Level and hummocky

5-4
Table 5-1, cont. Brief description of the benchmark sites

SITE NO. ECOLOGICAL SETTING PARENT MATERIAL AND CROPPING SYSTEM TILLAGE SYSTEM
AND SOILS SURFACE FORM
14-ON Mid Temperate, Medium-textured till. Corn - soybean- wheat No-till
Gray Brown Luvisols Rolling
15-QU Northern Temperate, Medium-textured till. Silage corn - forage Conventional tillage
Dystric Brunisols Rolling
16-QU Northern Temperate, Medium-textured till. Silage corn - forage Conventional till
Dystric Brunisols Rolling
17-QU Mid Temperate, Marine clay. Corn - forage Conventional tillage
Humic Gleysols Level
18-QU Mid Temperate, Marine clay. Corn - wheat - soybean - Minimum tillage
Humic Gleysols Level barley
19-NS Atlantic Temperate, Medium-textured till. Corn - forage Moldboard plow, spring disked
Gray Luvisols Undulating
20-NB Mid Temperate, Coarse-textured till. Potato - grain Chisel plow
Humo-Ferric Podzols Rolling
21-PE Atlantic Temperate, Coarse-textured till. Potato- grain - Conventional (plowdown of forage)
Humo-Ferric Podzols Undulating forage-forage
22-NB Mid Temperate, Coarse-textured till. Potato - potato - grain Chisel plow, grassed waterway,
Humo-Ferric Podzols Rolling diversion terraces
24-ON Mid Temperate, Medium-textured fluvial. Corn - soybean - alfalfa Conventional vs no-till
Humic Gleysols Level
25-NF Atlantic Boreal, Medium-textured till. Hay - grain Single furrow plow
Podzolic soils Hummocky

Note: Sites 15-QU and 16-QU, 17-QU and 18-QU are paired sites.

5-5
 MONITORING

The benchmark approach to soil quality monitoring will require a decade or more to reach its full
potential, although measurement of change to some attributes is anticipated at some sites much
sooner than that. It was critical that a true baseline be established at each site at the time monitoring
was initiated. Due to short-term variability, this required repeated measurement of some attributes
several times a year for several years, at the beginning. Other properties, considered to be more
stable, were not measured nearly as frequently.

Trends at each site will, in themselves, indicate whether soil quality is degrading, aggrading or
sustaining under the prevailing farming system. On this basis, it is anticipated that by using regional
soil and climatic information and expert systems, it will be possible to make general statements on
soil quality trends, both regionally and nationally.

Baseline Characterization

Detailed characterization of each benchmark site involves the measurement of various soil,
topographic and land use characteristics. These include sampling and analyses for various chemical,
physical, mineralogical and morphological soil properties and the preparation of detailed soil and
contour maps for each site. Climate stations were installed at sites lacking adequate characterization
from external data sources. More specifically, the types of information collected are outlined below:

• Site history - as far back as possible on cropping and tillage systems and fertilizer and pesticide
use.
• Soil map - about 1:1500 scale.
• Contour map - same scale as soil map with 0.1 to 1 m contour interval, depending on relief.
• Farm operation - kind and type of farm machinery, current cropping and tillage system.
• Pedon descriptions - two representative soil descriptions per site.
• Pedon analyses - soil moisture desorption curves, chemical, physical, mineralogical analyses (as
described in baseline data below) for each horizon of the two selected pedons.

Sampling Designs

• Grid design - used at sites having very gentle or simple slopes. The design involves a 25 x 25 m
grid, with a total of 80 to 100 grid points per site. A loose sample of surface Ap horizon was
taken at every grid point and loose samples of subsurface horizons were taken at randomly
selected grid points. Colour, texture, structure, horizon type and thickness were recorded for
each sample and landscape position for each profile.

5-6
• Transect design - used at sites with hummocky or ridged to undulating terrain and distinct
internal relief. Five or more transects with a total of about 60 sampling points, spaced 10 m apart,
adequately represent such landscapes. Orientation of each transect is perpendicular to the
contour, stretching from the crest of a "hill" to the bottom of an adjacent depression. A loose Ap
sample was collected as in the grid design. Subsurface samples were collected at 25% or more
of the transect points, selected to represent different slope positions. Data recorded at each
sampling point is the same as for the grid design.

Monitoring Frequency
Attributes Monitored (if applicable)
• Soil reaction (pH) 3 to 5 years
• Total organic carbon 3 to 5 years
• Total nitrogen 3 to 5 years
• CaCO3 equivalent 3 to 5 years
• Cation exchange capacity 3 to 5 years
and exchangeable cations
• Total elements (i.e. Al, Ca, Cd, Co N/A*
Cr, Cu, Fe, K, Li, Mg, Mn, Ni, Pb and Zn)
• Extractable Fe and Al (by oxalate, 3 to 5 years
dithionite-citrate and by pyrophosphate)
for Podzolic soils only.
• Available P and K 3 to 5 years
• Soil surface area N/A*
• Particle-size distribution N/A
• Clay mineralogy N/A
• Dry aggregate size distribution 3 to 5 years (Western sites)
• Saturated hydraulic conductivity (in situ) Yearly
• Penetrometer reading and soil moisture Twice yearly
(in situ)
• EM 38 conductivity measurements (in situ) Yearly
• (only for area with potential salinity problem)
• Biopore and root counts (in situ) Yearly
• Earthworm count (in situ) (except Prairies) Yearly
• Crop yields Yearly
• Climatic (in situ) Daily
• Farmer's management diary Yearly

* N/A = not applicable.

REPORTING

A network of 23 benchmark sites have been established for assessing long-term soil quality change.
At least one benchmark site was selected in each of the 10 provinces. Most of the sites are located
in major agricultural regions such as the Prairies and the St. Lawrence Lowland. Each site represents
a typical farming system on a typical soil and landscape.

5-7
A baseline database has been completed for eight benchmark sites and data from an additional 11
sites currently is being added to the database. By mid-1995, the baseline database will be completed
for the entire network of 23 benchmark sites. Baseline data include: chemical analyses, clay
mineralogy, particle-size and surface area, soil moisture retention and bulk density, total elements,
saturated hydraulic conductivity, penetrometer readings, biopore and earthworm counts (for some
sites) and yield. Climatic data from instrumented sites will be added to the database in the future.

Site documents, which describe each site in considerable detail, and include location, maps,
methodology, history, current management practices, and data tables from the database, are nearing
completion for 14 sites. Some of this data is also available for the remaining nine sites.

Beginning in 1997, the database (baseline and monitoring) will be used to validate predictive
models for various soil degradation processes; and recommend suitable environmental degradation
indicators for major agricultural regions. By 2000, analyses and reports of soil quality changes for
the major agricultural regions of Canada will be produced with potential for utilization in the State
of the Environment Report.

REFERENCES

Larson, W.E. and Pierce, F.J. 1991. Conservation and enhancement of soil quality. In: Evaluation
for sustainable land management in the developing world. IBSRAM Proc. No. 12(2), 175-203,
Bangkok, Thailand: International Board for Soil Research and Management, 1991.

Martin, S. 1989. Observatoire de la qualite des sols. Service des Technologies Propres et des
Dechets, Secretariat D'Etat Aupres Du Premier Ministre, Charge De L'Environment.
Neuilly-sur-Seine, France.

Tabi, M., L. Tardif, D. Carrier, G. Laflamme, and M. Rompre. 1990. Inventaire Des Problemes
De Degradation Des Sols Agricoles Du Quebec. Rapport Synthese. Gouvernement du Quebec.
Publication 90-130156. 71p.

5-8
CHAPTER 6
WIND EROSION PREDICTION AND ASSESSMENT

M.S. Bullock1, F.J. Larney1, W. Michalyna2

A.P. Moulin3 and G.A. Padbury4
1
Agriculture and Agri-Food Canada Research Station, Lethbridge, Alberta
2
Agriculture and Agri-Food Canada, Research Branch, CLBRR Manitoba Land Resource Unit,
Winnipeg, Manitoba
3
Agriculture and Agri-Food Canada Research Station, Melfort, Saskatchewan
4
Agriculture and Agri-Food Canada, Research Branch, CLBRR Saskatchewan Land Resource Unit,
Saskatoon, Saskatchewan

INTRODUCTION

In general, the term "soil quality" describes the relative capacity of a soil to perform a particular
function or use. More specifically, it can be thought of as the degree to which the characteristics or
attributes of the soil (texture, pH, slope, etc.) meet the optimum requirements of the particular use
or function. Thus, changes to one or more soil attributes can have a significant and often predictable
effect on soil quality (Larson and Pierce, 1991).

Soil degradation processes such as erosion remove material from the soil and, in so doing, invariably
change one or more of its attributes, a change that is frequently reflected in the capacity of the soil
to perform a particular function. It follows, therefore, that if erosion loss can be accurately and
quantitatively predicted, specifically in terms of a change to specific soil attributes, then it can be
used to predict soil quality change. The development of a wind erosion predictive capability is the
focus of this study.

The Wind Erosion Equation (WEQ) (Woodruff and Siddoway, 1965) is the main erosion prediction
technology currently in use. It has been used throughout the world to make average annual estimates
of soil loss due to wind erosion. In Canada, it has been used to compile wind erosion risk maps, and
as an extension tool for comparing the relative merit of selected management alternatives. It is,
however, an empirical, "black-box" model, and its application is somewhat limited. Wind erosion
is a dynamic process. Wind forces vary daily. Soil surface conditions (aggregates; crusts; loose,
erodible material) and vegetation residues that protect the soil from the forces of the wind change
daily, seasonally, and yearly in response to tillage and other management practices. This complex
and dynamic nature of wind erosion makes erosion prediction using an empirical model particularly
difficult. Moreover, model output, which is expressed in terms of average annual loss, is often
unrealistic, almost impossible to validate, and difficult to comprehend given the often sporadic
nature of wind erosion events.

6-1
To address the above concerns the Agriculture Research Service of the United States Department
of Agriculture (USDA-ARS) initiated the development of a new process-orientated wind erosion
prediction technology which represents, conceptually at least, a significant improvement over the
WEQ. The new "Wind Erosion Prediction System" (WEPS) is a computer system that uses
fundamental wind erosion principles to simulate the temporal changes in factors affecting wind
erosion, and to compute erosion when critical combinations of these factors occur. The actual model
that predicts wind erosion loss is called the Wind Erosion Research Model (WERM).

The objective of the present study was to evaluate alternative procedures for predicting wind erosion
in Canada and, in particular, the Wind Erosion Research Model (WERM) currently being developed
in the U.S.A.

WIND EROSION RESEARCH MODEL (WERM) VALIDATION

The Wind Erosion Research Model is a daily time-step process model that predicts temporal
characteristics such as surface soil conditions and residue levels, and then uses long-term weather
records to estimate the probability and severity of an erosion event. The temporal soil properties,
residue levels and climatic parameters are predicted via the following submodels:

SOIL submodel - predicts surface soil parameters such as aggregate size, density and stability, crust
thickness and stability, surface micro-relief, bulk density and loose erodible material.

WEATHER submodel - this submodel is, in essence, a weather file containing a series of daily
weather parameters required by the overall erosion model and several of the submodels.

HYDROLOGY submodel - predicts near-surface soil moisture conditions. The effects of

freeze-thaw and freeze-drying processes are also addressed in this model.

MANAGEMENT submodel - predicts changes to both temporal soil surface characteristics and
residues due to management practices such as tillage.

CROP submodel - predicts initial residues, and also considers the effect the crop may have in
reducing erosion during the early growth stages.

DECOMPOSITION submodel - predicts the amount of residue on a daily basis as a function of

initial residue, tillage operations, climatic conditions, and decomposition rates.

A WERM validation site was established near Lethbridge on a clay loam soil which had been under
zero-tillage for 6 years. The site, itself, consisted of a circular plot, 200 m in diameter, which was
tilled to enhance wind erosion events. A meteorological tower was erected in the centre of the plot
and wind speed and direction, relative humidity, precipitation, soil and air temperatures, solar
radiation, and vapour pressure were measured. A wind erosion sensor (SENSIT) was installed to
record when wind erosion began (threshold conditions) and ended. Dust collectors (BSNE) were
located in a circular pattern within the plot. Since the WERM Validation Site is one of a network
(the rest being in the United States), it was characterized in a manner consistent with the overall
program (Larney et al., 1992).

6-2
The site was established in November, 1990 and was operated until May, 1992. After each erosion
event, major precipitation event, or tillage operation, surface samples (0-2.5 cm) were taken for
aggregate size distribution analysis. Soil cores were taken to determine soil moisture content and
bulk density which are temporal surface properties affecting soil erodibility (Bullock et al., 1991).
Sixteen wind erosion events were monitored. Soil erosion losses are given in Table 6-1.

Table 6-1. Erosion losses at WERM validation site

EROSION LOSS (Mg ha-1)

1991 1992
Apr. 4 6.1 Apr.3 30.4
Apr.8 2.4 Apr.4 6.1
Apr. 25 0.3 Apr.5 5.7
Dec.6 22.5 Apr. 9 1.2
Dec.9 20.3 Apr. 13 2.2
Dec. 10 13.6 Apr. 18 12.3
Dec. 13 13.9 Apr. 27 0.5
Dec. 16 6.1 May 11 0.8
TOTAL 85.2 TOTAL 59.2

The total loss for the 16 events was about 144 Mg ha-1 or the equivalent of 14 mm of topsoil over
a 3.14 ha area. These losses illustrate the fragility of the soil, considering that it had just been in
zero-tillage for 6 years. The so-called T-value (tolerable level of soil loss due to erosion) is about
11 Mg ha-1 annually (Larney et al., 1992).

Although not directly related to the development and validation of WERM, two other studies were
carried out at the site. Dust samples from the April erosion events were sent to the Regina Research
Station for herbicide analysis. The off-farm impact of dust, which may contain herbicide residues,
may pose a potential environmental problem in terms of water and air quality. The samples were
analyzed for Dicamba ("Banvel"), MCPA, Bromoxynil ("Hoegrass II"), 2,4-D, Trifluralin
("Treflan"), Triallate ("Avadex"), and Diclofop ("Hoegrass") residues. Residues of 2,4-D, trifluralin
and treflan were found to be the most abundant with lesser amounts of Diclofop, traces of
Bromoxynil and Dicamba and no MCPA. The highest residue concentration found was one sample
with 52 ppb of 2,4-D. Another WERM validation site was established at the Regina Research Station
in the fall of 1991 largely to monitor pesticide residues in dust (Allan Cessna, personal
communication). A few minor erosion events were monitored, despite the extremely moist
conditions.

6-3
The WERM site was seeded to spring wheat in 1992, which provided an opportunity to quantify the
effects of wind erosion on crop productivity. Plant density, dry matter production and grain yield
were measured to assess these effects under natural conditions. The 200 m diameter circle was
divided into 20 transects, from west (closest to protected surface with least erosion) to east (furthest
from protected surface with most erosion). Each transect was 10 m wide. There was a significant
decline in crop yield from west to east represented by the equation:

y = 3106 - 3.6x (r2 = 0.36**)

where y = grain yield in kg/ha and x = distance from protected surface in m. For every 10 m
increase in the fetch distance from the protected surface, grain yield decreased by 36 kg ha-1.

SUBMODEL VALIDATION

Submodel validation studies were designed in cooperation with USDA-ARS scientists to extend the
development and validation of the submodels for Canadian conditions.

SOIL submodel

The SOIL submodel predicts surface soil parameters such as aggregate size, density and stability,
crust thickness and stability, surface micro-relief, bulk density and loose erodible material.
Dependent temporal soil variables in the SOIL submodel are calculated from "equivalent" values
for independent variables such as precipitation or wind erosion. The SOIL submodel was originally
designed to incorporate empirical equations primarily based on cumulative rainfall although new
approaches based on rainfall energy were recently proposed. Approximations have been proposed
for freeze-thaw and freeze-dry effects on aggregate size distribution though they may be changed
with further analysis of the data.

The research protocol developed by the USDA-ARS was used to evaluate the SOIL submodel. Field
studies of soil variables were conducted in 1991 and 1992 at two sites, a silty clay Chernozemic soil
at the Melfort Research Station and a sandy loam Luvisolic soil under fallow east of Gronlid,
Saskatchewan. The treatments at the Melfort site were conventional tillage, fallow with herbicides
and tillage, and fallow with herbicides only. Thirteen variables related to soil erodibility, including
aggregate size distribution, dry aggregate stability, and dry crushing energy, were measured five
times during the field season at the Melfort site, and eight times at the Luvisolic site (Moulin,
1993a).

Evaluation of long-term and short-term aggregate size data from the chemical fallow study at the
Melfort Research Station revealed two trends. The long-term data showed a trend which persisted
over several years and was unrelated to the tillage system. Analysis of aggregate size data taken
over the 1991 fallow season indicated that the proportion of small aggregates increased with time
(Moulin, 1993a).

A freeze-thaw study was carried out on a clay loan soil at the Fairfield Farm neat Lethbridge. The
soil was cultivated with a chisel cultivator in September 1992, and samples for aggregate size
distribution were taken at intervals throughout the winter to ascertain the effect of overwinter

6-4
climatic forces on erodibility. Thermocouples were used to measure the depth of freezing and the
number of freeze/thaw cycles. Time Domain Reflectometry (TDR) probes were used to monitor soil
moisture. Surface roughness was measured over the winter period using the chain method on six 1
m transects. Three transects were parallel and three were perpendicular to the direction of tillage.
Wet aggregate stability was determined on 1-2 mm aggregates from samples collected over the
winter period.

Another phenomenon related to frequent freeze-thaw cycles prevalent in the chinook belt of southern
Alberta, is the influence of blowing snow on the wind erosion process. Observations at the WERM
validation site showed that the abrasive capacity of blowing snow was similar to wind-transported
soil particles, although it occurred under different meteorological conditions. Differential snow
cover caused by random drifting exerts a significant effect on the soil surface. Areas with little or
no snow dry quickly and become erodible, whereas meltwater from deeper snow cover saturates
the soil surface causing it to slake and crust which offers some resistance to soil movement by wind
(Bullock et al., 1992).

A study was carried out at two sites in southern Manitoba to assess the importance of clay, organic
carbon, and carbohydrate content on the size and stability of aggregates. At the Carman site, a
loamy Black Chernozemic soil, the clay content of the aggregates varied in relation to the aggregate
size. The clay content was lowest in the < 0.84 mm size aggregates. The clay content, on the other
hand, was higher in the 0.84 to 1.0 and 1.0 to 2.3 mm size aggregates; the increase in clay content
compared to the < 0.84 mm size varied from 4 to 12%. The clay content of aggregates greater than
2.3 mm was 1 to 3% above the clay content of the < 0.84 aggregate size. The clay contents of
aggregates from the Osborne clay at Brunkild were variable (54 to 65%) and showed no relationship
to aggregate size (Michalyna, 1993).

The organic carbon content of the aggregates on the loam soils at the Carman site showed a pattern
similar to that obtained for clay content. In most cases, the organic carbon content was lowest in the
< 0.84 mm size aggregates. The highest organic carbon levels occurred in the 0.84 to 1.0 and 1.0
to 2.3 mm size aggregates. On the clay soils at Brunkild, the organic carbon was generally greater
than on the soils at Carman; but the differences among the various-sized aggregates was generally
low.

Carbohydrate content of the aggregates on the loam soils at Carman showed a similar trend to the
organic carbon content. In most cases, the lowest carbohydrate levels occurred in the < 0.84 mm size
aggregates; mean values ranged from 0.25 to 0.37%. A few lower levels occurred in the > 38 mm
and 20 to 38 mm aggregate sizes. The highest carbohydrate levels occurred in the 1.0 to 2.3 mm
followed by the 2.3 to 6.3 mm, and 0.84 to 1.0 mm sized aggregates; mean values ranged from 0.51
to 0.92%, 0.42 to 0.56%, and 0.37 to 0.54%, respectively.

On the clay soils at Brunkild, the carbohydrate content was generally greater than at Carman with
values commonly between 0.45 to 0.60%. For the soils at both the Carman and Brunkild sites, the
mean organic carbon/carbohydrate ratio was 10.5; with generally no relationship to aggregate size
(Michalyna, 1993).

6-5
MANAGEMENT (TILLAGE) Submodel

The MANAGEMENT submodel, in essence, predicts changes to both temporal soil surface char-
acteristics and residues due to management practices such as tillage. The submodel is incomplete
as some equations remain to be developed, although an equation based on Proctor density, which
simulates the mixing of soil from lower depths with aggregates at the surface has been developed.
Field studies at the Melfort Research Station indicate that tillage operations with a double disk
cultivator increase the proportion of large aggregates at the surface. Subsequent cultivation with a
field cultivator appears to have little effect (Moulin, 1993a).

Two sites in southern Manitoba were investigated to assess the effect of cultural practices on bulk
density. Since the surface layer (Ap) was variably disturbed, the underlying Ah at a depth below 12
cm was used as an intrinsic parameter. At the Carman site, a loamy textured soil, the average bulk
density of the Ah horizon was 1.32 and ranged from 1.24 to 1.45. Since any disturbance of the
surface would likely reduce the bulk density, the soil condition in the fall (following harvest) was
considered the most stable and most similar to the intrinsic Ah values. Throughout the growing
season, the surface bulk density ranged from an average near-stable value of 1.29 to about 0.83
depending on the seeding or cultivation equipment used. The higher values (1.45 and above) were
assumed to be due to compaction by equipment tires. On the clay soil at Brunkild, the bulk density
in the fall was 0.91; average cultivated bulk density values were 0.76 and 0.68 in the fall and in the
spring, respectively (Michalyna, 1993).

A comparison of the bulk density with the % aggregate size diameter (ASD) greater than 0.84 mm
and the geometric mean diameter (GMD) was conducted. At Carman, on very find sandy loam
textured soil, the stubble treatments (wheat and canola) had a higher bulk density and a slightly
larger GMD than the cultivated treatments. In general, for a 25% decrease in bulk density (1.29 to
0.97) due to cultivation, there was an 8% decrease in ASD > 0.84 mm (72 to 64) and a decrease in
GMD (8.9 to 3.3). At Brunkild on a clay soil, the stubble treatments had the higher bulk density but
had a lower ASD > 0.84 and GMD than the cultivated treatments. For a 20% decrease in bulk
density (1.00 to 0.80) due to cultivation, there was an 8% increase (80 to 88) in ASD > 0.84 mm and
a considerable increase in GMD (7.2 to 16.0) (Michalyna, 1993).

A study was carried out near Lethbridge to investigate the effect of moisture content at time of
tillage and the tillage implement on soil erodibility factors such as aggregate size distribution and
bulk density. The site was in fallow in 1991. The main treatment was soil moisture content at time
of tillage; wet (23%), moist (18%), and dry (14%). The optimum soil moisture content for
compaction (standard Proctor test) was 16.8% and which yielded a density of 1.7 Mg/m3. The
sub-treatment was tillage implement: Noble blade, tandem disc, and powered rotary cultivator. Prior
to the tillage operations, surface relief measurements were taken for surface roughness
determination, and soil samples were taken for soil moisture, bulk density, dry aggregate size
distribution, and dry aggregate stability determinations. Immediately following the tillage operations
the same measurements and sampling were repeated, the objective being to determine parameter
changes as affected by moisture content at time of tillage, and tillage implement (Bullock et al.
1993; Larney and Bullock, 1993). The results showed that winter breakdown of aggregates was
more prevalent on tandem disc and rototiller treatments compared with blade cultivated treatments.

6-6
The tandem disc and rototiller brought unstable aggregates to the soil surface which were less
resilient to the freezing/thawing and wetting/drying cycles over the winter of 1991-92. The blade
cultivator has a non-mixing action which leaves unstable aggregates at depth and stable aggregates
on the soil surface.

DECOMPOSITION Submodel

Residue decomposition in WERM is based on a weighted-time variable calculated from functions

of temperature and moisture. Optimum moisture and temperature conditions result in the
accumulation of 1 decomposition day for each day of the simulation. There is a significant
correlation between cumulative degree-days and decomposition of residue at Melfort (Moulin,
1993b).

In another Saskatchewan study, photographs of stubble residue were analyzed using image analysis
software in order to estimate the percentage of residue cover, and to track the reduction in residue
cover with time and tillage practices. Crops included spring wheat, durum wheat, barley, canary
seed, lentils, peas and flax.

CROP Submodel

The CROP submodel is a modified version of the crop growth submodel in the Erosion Productivity
Impact Calculator (EPIC). Crop growth is calculated on the basis of accumulated heat units. Leaf
area index, biomass and partitioning of biomass between roots and above ground biomass are
similar to those in EPIC. Evaluation of EPIC with data from rotations at the Melfort Research
Station indicates that the crop growth model simulates long-term means of yield which are similar
to actual yields for spring wheat. Evaluation of soil nitrate nitrogen generated by EPIC indicates that
the simulated values are greater than those measured in the field (Moulin, 1992).

MONITORING AND CHARACTERIZATION SITES

Ten wind erosion monitoring sites were established across the prairies in cooperation with agencies
such as Alberta Agriculture, Manitoba Agriculture, Saskatchewan Save our Soils Program, and the
Prairie Farm Rehabilitation Administration (PFRA). The purpose was to measure or estimate actual
wind erosion loss in the field in order to compare with those values predicted by the WEPS.

Each site was instrumented with four clusters of soil collectors (BSNE) with each cluster consisting
of four collectors at heights of 10 cm, 25 cm, 50 cm, and 100 cm above the soil surface. The
collectors were placed on the leeward side of the erodible area. Wind speed data along with
information on temporal soil parameters such as aggregate size distribution and residue levels were
collected at selected sites. Chemical and physical analysis of both the soil and dust samples were
carried out. The sites were set up in late March and taken down in early June.

6-7
A total of 20 wind erosion events were recorded over a two year period, with measured erosion
losses ranging from 0.3 to almost 20 tonnes per hectare, depending upon the soil type, the condition
of the soil surface in terms of aggregates, surface roughness, and residue levels, as well as the wind
speed and the duration of the wind erosion event. In most cases, about 70% of the material collected
was attributed to creep-saltation (< 50 cm height) and about 30% to suspension, although on one
clay soil almost all of the losses were attributed to creep-saltation. A number of properties of the
dust particles from the various collectors were compared. In general, the organic carbon content, the
clay content, and the moisture content increased slightly from the lowest (10 cm) to the highest
collector (100 cm).

Soil depositions along field boundaries can serve as a reference to the amount of erosion that has
occurred from adjacent fields in the past, and are indicative to a degree of future susceptibility. In
a Manitoba study, cross-section elevations were conducted through a number of depositional areas
adjacent to field sites where dust collectors were set up. Based on the cross-section area and the
length of the banks, the amount of soil in the banks was calculated at between 17,880 to 21,500
tonnes. Assuming that about 30% of the eroded material was carried away in suspension, the actual
erosion loss would be greater by a factor of 1.4 making the total estimated loss at between 25,000
and 30,000 tonnes, which is equivalent to about 5 cm of topsoil. Depending upon the time interval
since the first erosion event, the estimated loss would be between 12 Mg ha-1 yr-1 (over 60 years) and
24 Mg ha-1 yr-1 (over 30 years) (Michalyna, 1993).

Wind Erosion Prediction in Relation to Soil Quality Monitoring

Since the development of the WERM model has not progressed as was anticipated when the project
was initiated, the original objectives of validating the WERM model for Canadian conditions and
developing a new operational wind erosion prediction capability for soil quality evaluations has not
yet been accomplished.

In terms of wind erosion prediction it means that current estimates will have to be based on the Wind
Erosion Equation (WEQ) or perhaps the Revised Wind Erosion Equation (RWEQ). The latter is an
interim technology which has been developed to address some of the concerns raised with WEQ
until the new system is operational. Some of the concepts and data from the WERM development
have been incorporated into the new RWEQ technology.

As mentioned earlier, the assessment of soil quality or of soil quality change requires a quantitative
evaluation of soil modifying processes such as wind erosion. Simply put, how much topsoil has been
lost? What does this mean in terms of individual soil attributes such as organic matter? And, how
does a change in one or more of these attributes, in turn, affect the soil's ability to perform a
particular function? Although WEQ can and has been used to provide quantitative estimates of wind
erosion loss, the reliability of these estimates has often been questioned.

The Wind Erosion Equation is an empirical model which predicts erosion loss as a function of
climate, soil and management parameters by calculating the relative significance of each factor.
From previous work in estimating the relative wind erosion risk on the Prairies, the equation
appeared to give reasonable results. In other words, the relative ranking of land in terms of its

6-8
inherent erosion risk using the wind erosion equation appeared reasonable. This suggests that while
WEQ may not be sophisticated enough to provide a quantitative estimate of erosion loss for the
purpose of soil quality evaluation, it could be used to assess the relative risk of wind erosion, or,
perhaps more importantly, to assess the change in the erosion risk over time as a result of changing
management practices.

ACKNOWLEDGEMENTS

The assistance of John Timmermans, Alberta Agriculture; Dennis Haak, Michelle Harland, Jamshed
Merchant, Dale Regnier, and Dean Smith, PFRA; Larry Slevinsky, Manitoba Agriculture; and Dale
Fyke, Saskatchewan, Save Our Soils Program was invaluable in establishing the monitoring and
characterization sites. We also acknowledge the co-operation of Sean McGinn, Agriculture and
Agri-Food Canada and D.W. Fryrear, ARS-USDA in the establishment and instrumentation of the
validation site as well as Ike Lanier on whose land the validation site was located. Allan Cessna,
Agriculture and Agri-Food Canada, carried out the herbicide analysis on the dust samples.

REFERENCES

Bullock, M.S., Larney, F.J., McGinn S.M., and Fryrear, D.W. 1991. Wind erosion processes in
the chinook belt of southern Alberta. 83rd Annual Meeting, American Society of Agronomy,
Denver, Colorado. Agronomy Abstracts, American Society of Agronomy, Madison, WI.

Bullock, M.S., Larney, F.J., McGinn S.M., and Olson, B.M. 1992. Influence of snow on wind
erosion processes in the Chinook belt of southern Alberta. Proc. Saskatchewan Soils and
Crops Workshop, Saskatoon, SK.

Bullock, M.S., Larney, F.J., and Lindwall, C.W. 1993. Tillage management effects on aggregate
size fractions influencing wind erosion. 30th Ann. Alberta Soil Science Workshop Proc.,
Edmonton, AB.

Faculty of Agriculture, U of Manitoba. 1991. Contribution to issues - erosion section; CH 4-Land

in "Sustainability of Canada's agri-food-system: A prairie perspective". Faculty of
Agriculture and Food Science, University of Manitoba, Winnipeg, MB.

Larney, F.J., Bullock, M.S., and McGinn, S.M. 1992. The wind erosion prediction process
(WEPS) and wind erosion processes in southern Alberta. Proc. 29th. Annual Soil Science
Workshop, Lethbridge, AB.

Larney, F.J., Bullock, M.S., McGinn, S.M., and Fryrear, D.W. 1992. Characterizing wind
erosion processes on summer-fallow in southern Alberta. Proc. Erosion: Causes to Cures
Conf., Regina, SK.

6-9
Larney, F.J. and Bullock, M.S. 1993. Influence of soil wetness at time of tillage and tillage
implement on soil properties affecting wind erosion. Soil and Tillage Res. (accepted).

Larson, W.E. and Pierce, F.J. 1991. Conservation and enhancement of soil quality. In: Evaluation
for sustainable land management in the developing world. IBSRAM Proc. No. 12(2),
175-203, Bangkok, Thailand: International Board for Soil Research and Management, 1991.

Michalyna, W. 1993. Temporal soil properties and soil loss during wind erosion events. 36th
Annual Manitoba Society of Soil Science Meeting, Publications Branch, Province of
Manitoba, Winnipeg, MB.

Moulin, A. 1992. Evaluation of the CERES and EPIC models for predicting spring wheat grain
yield over time. Can. J. Plant Sci. (in press).

Moulin, A. 1993a. Temporal trends in soil erodibility during the fallow season in northeastern
Saskatchewan. Proc. Saskatchewan Soils and Crops Workshop, Saskatoon, SK.

Moulin, A. 1993b. Decomposition of crop residue in northeastern Saskatchewan. Proc.

Saskatchewan Soils and Crops Workshop, Saskatoon, SK.

Padbury, G.A. and Acton, D.F. 1992. New initiatives in wind erosion modeling and prediction
in western Canada. Proc. Saskatchewan Soils and Crops Workshop, Saskatoon, SK.

Woodruff, N.P. and Siddoway, F.H. 1965. A wind erosion equation. Soil Sci. Soc. Am. Proc. 29:
602-608.

6-10
CHAPTER 7
WATER EROSION PREDICTION AND ASSESSMENT

G.J. Wall1, E.A. Pringle1 and D.R. Coote2

1
Agriculture and Agri-Food Canada, Research Branch, CLBRR Ontario Land Resource Unit,
Guelph, Ontario
2
Agriculture and Agri-Food Canada, Research Branch, Centre for Land and Biological
Resources Research, Ottawa, Ontario

INTRODUCTION

Soil quality has been described elsewhere in this report as the capacity of the soil to provide for
sustainable crop production and still maintain the quality of the environment. It is generally
understood that processes such as water erosion often impact negatively on the capacity of soils to
perform these critical functions. However, until better estimates of the current status and trends in
the extent and severity of water erosion are available, it is difficult to assess the regional and
national significance of this form of soil degradation. Further, such estimates and associated
capabilities are required to predict the vulnerability of soil to water erosion under various land
management practices so that the sustainability of current and emerging land management practices
may be more appropriately assessed.

Measurement of soil erosion rates in the field is both time consuming and expensive since it is
necessary to monitor runoff events throughout the year over several years to collect data from the
range of both high and low return-period storms that result in soil loss. It is also necessary to
measure soil loss under various crops and tillage management systems and for various soils and
slopes.

The Universal Soil Loss Equation (USLE) is an empirical model that was developed in the United
States in 1965 to predict long-term average annual soil loss due to water erosion. Although it has
been used to predict rates of soil erosion for agricultural landscapes across Canada, it has not been
possible to validate the predicted rates of soil loss since appropriate monitoring data have not been
available. The USLE has also been used in farm planning to select alternative cropping practices that
lower soil erosion losses. This application of the USLE has been limited since predicted soil loss
rates do not distinguish between sediment deposition in the field and sediment transport from the
field. Therefore, it is not appropriate to employ the USLE to study the off-site or environmental
impacts of soil loss on water quality.

Recent findings from a study at the University of Guelph compound the difficulties of using the
USLE for farm planning purposes. This research has indicated that soil transport by tillage
equipment can result in significant soil loss on convex slope positions. It appears that tillage erosion
(or tillage translocation) is an important process that should be included in soil loss prediction
equations that are used for conservation farm planning. Remedial measures to control a tillage
erosion problem may be markedly different from those aimed at reducing soil erosion by water.

7-1
As Canadian farmers move toward more sustainable crop production systems, there is a need for
improved information on soil erosion risks and impacts on sustainable crop production. Transfer
of current soil and water conservation technologies is required for the implementation of
environmental farm planning programs of the future. Such a process would benefit greatly from an
accurate capability to predict water erosion loss for various soils and landscapes and for any given
set of management practices.

Efforts are ongoing in the USA to revise and improve the USLE for use in farm planning for
sustainable production. The revised universal soil loss equation (RUSLE) is now available but has
not been tested or modified for use in Canada. Concurrent to the revision of the USLE, a new soil
erosion model called the Water Erosion Prediction Project (WEPP) is being developed and tested
by the United States Department of Agriculture. The WEPP model is process based and more
sensitive to soil and climatic conditions. Canadian soil scientists are participating in the validation
of this model and the testing of winter runoff routines. The testing of both the RUSLE and WEPP
models in Canada is limited by the lack of soil loss data for agricultural landscapes on which to
calibrate and validate soil erosion model results.

In consideration of the requirement for an improved predictive technology and the potential of
technologies that are emerging in the USA, this water erosion study was established to ascertain the
validity of these emerging capabilities for water erosion prediction for soils and farming systems
in Canada and to collaborate with scientists across Canada and the USA in the development of
improved predictive capabilities.

OBJECTIVES

A water erosion monitoring and prediction study was developed to evaluate the effect of water
erosion on soil quality and provide guidelines for the implementation of soil conservation practices
by:

i) conducting baseline monitoring of soil erosion rates on agricultural landscapes across

Canada,
ii) validating water erosion prediction methods,
iii) using water erosion models to predict past and future changes in soil erosion losses on
agricultural landscapes, and
iv) compiling a report on soil erosion risk and a handbook for estimating soil loss from water
erosion for conservation planning.

7-2
 METHODS

Baseline Erosion Monitoring

137
Cs was deposited on the earth's surface in the early 1960's as a result of nuclear testing. Since it
is a radioisotope which is held strongly to clay particles and therefore moves with the soil, it acts
as a tracer for investigating soil movement. Soil quality benchmark sites (see Chapter 5, Figure 5-1)
across Canada have been sampled for 137Cs analysis on a detailed grid basis or transects to study soil
redistribution on agricultural landscapes. Soil redistribution is the movement of soil by any process
from one landscape position to another or off the field. In addition to these benchmark sites, 27 slope
profiles (3 soils, 3 cropping histories, 3 replicates) have been sampled for 137Cs in PE. Erosion plots
also have been sampled for 137Cs in BC and NB. Using these data, a national data base of soil
redistribution has been initiated. In several years time, these sites can be resampled to further
investigate the redistribution of soil on the landscape.

Plots for measuring soil erosion by water have been established in ON, BC, AB, MB, and NB. At
these locations, soil loss is measured for all runoff events throughout the year. Year-round
monitoring is critical given that significant percentages of annual erosion can occur in very short
periods of time, often during the spring thaw period.

Model Validation

Several versions of the WEPP model are planned for release in the near future (1995). The hillslope
version (for which working versions currently exist) has been designed to model soil and water loss
at the field scale. The watershed version allows for all the processes modelled with the hillslope
version but permits the movement of runoff and eroded sediment into grassed waterways,
impoundments, ephemeral gullies etc. Modelling soil and water losses for a larger area is permitted
with the grid version of WEPP. The grid version requires breaking up the landscape to be modelled
into square grid cells. Water and soil is then routed from one cell to another. This version permits
GIS linkages.

Work has been done cooperatively with the WEPP core team from the USA to validate model
routines in the preliminary model releases and perform some initial calibration exercises. The
WEPP model utilizes measured weather data or a climatic data generator known as CLIGEN for
creating the necessary climatic file inputs to WEPP. The CLIGEN model is being used to fill in
missing data for selected Canadian climate stations. Testing of the CLIGEN model was also initiated
to establish whether the model generates a realistic distribution of extreme events that are so critical
for soil erosion analysis.

Climatic data on freeze-thaw cycles, rainfall on frozen ground, and snowmelt were also compared
with winter soil loss data to better understand and predict winter soil erosion losses.

Tillage erosion is defined as the variable redistribution of soil through tillage translocation resulting
in net losses and net accumulations of soil along the path of tillage transecting a landscape. The
action of tillage is dynamic in that the impulse applied to the soil varies in time and space in
response to the spatial complexity of the landscape. A study was conducted at the University of

7-3
Guelph to model tillage translocation and tillage erosion. The purpose of this exercise was to come
closer to quantifying the contribution of tillage erosion, alone, on redistribution of soil in the
landscape. Data from two sites in southern Ontario were used to validate the proposed model.

Changes in Water Erosion Losses

As conservation management practices are implemented across the country, there is a need to track
improvements in soil erosion rates and highlight areas where high erosion potential still exist. To
analyze the trend in erosion rates on a national basis, estimates of soil erosion rates over time must
be calculated. It was proposed that the WEPP model be utilized to calculate trends in water erosion
using farm census data from 1981 and 1991. Since the WEPP model is still in the developmental
stage, the USLE was determined to be the best technology currently available for this state of
environment reporting.

Technology Transfer of Water Erosion Information

Identification of areas in Canada that are at risk to water erosion was seen as an important client
need. Consequently, the soil landscape maps for Canada were used as the basis for establishing
water erosion risk maps. The risk of water erosion for all of Canada was estimated using the USLE
for each polygon on the soil landscape map.

Another important area of technology transfer involves providing assistance to planners, consultants,
farm managers and conservation specialists in predicting soil losses from water erosion. A soil
erosion manual which is easy to use was prepared to meet this need. The purpose of the handbook
is to provide soil conservation planners with current information for USLE or RUSLE soil loss
predictions. The information in the handbook will improve the reliability of erosion predictions
made with these existing methods and will provide a compilation of the best methods for predicting
soil erosion by water for Canadian conditions.

RESULTS AND DISCUSSION

Baseline Monitoring

To study soil redistribution on agricultural landscapes, soil samples from field-scale plots in BC,
AB, SK, MB, ON, QU, and NB were analyzed for 137Cs. Watershed-scale plots were sampled for
137
Cs analysis in SK and PE. The data for QU, NB and PE have been analyzed and reported. Cao
et al. (1993) reported that soil redistribution at 2 QU sites averaged 24.3 and 12.9 t ha-1 yr-1; well
above the often proposed tolerance level of 5 t ha-1 yr-1. Soil loss differences were related to slope
and cropping history variation. The average soil movement per year of tillage was almost the same
at each site. The most eroded landscape positions at the 2 sites lost 97 and 64 t ha-1 yr-1. Cao et al.
(1994) reported soil movement from continuous potato production in NB. Average soil loss over the
last 30 years was 53 t ha-1 yr-1 with a maximum value of 190 t ha-1 yr-1. Potato yield data showed a
strong relationship with soil movement. About 0.15% loss of yield was associated with each 1 t
ha-1 yr-1 soil loss. Kachanoski (1992) reported soil loss rates under different soil and cropping
practices from small watersheds in PE. The average losses for all slope positions were 20 t ha-1 yr-1

7-4
and 40 t ha-1 yr-1 for a less intensive and more intensive management system, respectively. Some
soils exhibited soil loss rates of 70 t ha-1 yr-1. The findings from all of these monitoring studies
indicate extremely high soil losses which are unlikely a result of water erosion alone but illustrate
the severe effect of tillage translocation of soil on some landscapes.

Model Validation

Climatic data files required to run the WEPP model have being compiled for 33 locations across
Canada for use in erosion prediction equations. A detailed list of WEPP model parameter
requirements has been sent to study co-operators collecting soil loss data to ensure that the best
possible data bases were created. Modelling protocols have been developed for project modellers
to ensure compatible data across the country. The protocols include details on data management,
methods for calibration and preferred modes of operation. Information on troubleshooting and
information dissemination are also included.

Preliminary versions of the WEPP hillslope model have been evaluated on microplot-scale data in
ON and with plot-scale data in BC. Sensitivity analysis of important input parameters has indicated
that saturated hydraulic conductivity and rill and interrill erodibility are the most critical input
parameters.

Previous versions of the hillslope model did not allow for seasonal variation of saturated hydraulic
conductivity (Ksat). Validation work by Cao et al. (1993), using a data set collected from three
Ontario soils for a one-year period, indicated a large variation of the calibrated Ksat parameter over
the year (Fig. 7-1). These data indicate a marked decline in Ksat from June through to March. Note
that these soils were fallowed and untilled for the entire year eliminating tillage as a cause for the
temporal variation. As a result of these findings, the final version of the model will permit temporal
variation of the parameter. The calibrated interrill erodibility values (Ksat) for the same three soils
are shown in Fig. 7-2. Again, the results clearly show a temporal variability of this parameter, with
erodibilities of all soils increasing in the fall and reaching their highest values in March.

Additional WEPP validation studies in Ontario using another data set also on the microplot scale
and in event mode, have indicated that, under notill conditions, the model tends to overpredict
surface runoff volume. Alteration of the hydraulic conductivity parameter to values almost twice the
observed were required to force the model's predicted runoff to match observed runoff. For the
conventionally-tilled plots, field-measured Ksat values were the best estimates of measured runoff.

Validation work in BC at the plot scale in event mode has shown that WEPP underpredicts soil and
water losses. Running in continuous mode, the model underpredicted snowmelt and overpredicted
rainfall runoff. The model was also unable to correctly rank the monitoring plots in terms of soil and
water losses (van Vliet, 1993).

7-5
 Fig. 7-1. Temporal variability of calibrated Ksat for three soils.

Fig. 7-2. Temporal variability of calibrated Ki for three soils.

7-6
Hayhoe et al. (1992a, 1992b) reported information on freeze-thaw cycles, rainfall on frozen ground,
and snowmelt over a six-year period for selected locations in Canada. Further work by Hayhoe et
al. (1993) indicated that current models (WEPP and the versatile soil moisture budget - VB) do not
adequately predict winter and spring runoff. Specific problems include the tendency for WEPP to
overpredict snow depth (Fig. 7-3), and to show no relationship of predicted to observed frost depth
(Fig. 7-4). Seasonal freeze/thaw cycles were also not well modelled by WEPP (Fig. 7-5). These
results highlighted the need to modify the winter components of WEPP through continued
communication and cooperation with WEPP developers in the USA.

Fig. 7-3. Estimated and observed snow depth using the versatile
soil moisture budget (VB) and WEPP for the 1982-83
winter season.

Lobb and Kachanoski (1993) developed a tillage erosion model at the University of Guelph. The
redistribution of soil within the landscape is modelled as a simple input/output system using
estimated grid point values for soil and tillage parameters, and forward, backward and central
differences for topographic parameters. The tillage erosion model will be used to identify landscapes
where tillage erosion processes are dominant. A working model demonstrating the relationship
between landscape position, tillage system, and soil loss has been developed and tested on two sites
with complex topography in Ontario. The new model developed to estimate tillage translocation as
a function of slope gradient and slope curvature as given by Lobb and Kachanoski (1993) is:

7-7
 TN = (" + $2 + J N)

where: TN = net tillage translocation (kg m-1)

" = translocation by tillage unaffected by slope gradient or curvature (kg m')
2 = slope gradient (%)
N = slope curvature (% m-1)
$ = coefficient to describe the additional translocation resulting from slope
gradient (kg m-1)
J = coefficient to describe the additional translocation resulting from slope
curvature (kg %-1)

To calculate net tillage translocation or tillage erosion, the equation is:

TN = (" + $ 2o + J No) - (" + $ 21 + J N1)

where 2o, No represent forward (output) movement of soil from the point and 21 + N1 represent
inputs.

Fig. 7-4. Relationship between observed and estimated frost depth

using WEPP for 1982 through 1984.

7-8
 Fig. 7-5. Relationship between observed and estimated seasonal freeze-thaw
cycles using WEPP

Changes in Water Erosion

Trend analysis of water erosion is in progress using census data from 1981 to 1991. The USLE is
being used to estimate water erosion for each province for each year that census data are available.
The provinces of Ontario and Manitoba are being used to test the methodology initially. C factors
will be developed for each of the soil landscape map polygons based on census data. Other USLE
factors used in creating the erosion risk maps will be used to create an estimate of soil loss rates. The
results of this study will highlight areas where changes in management practices have resulted in
significant reductions in soil erosion by water on a national scale.

Technology Transfer of Water Erosion Information

Maps of water erosion risk are currently available for all regions of the country (except BC and the
maritime provinces) at a scale of 1:1 million (remaining provincial maps to be released in 1995).
These maps are one of a series of interpretive soil degradation maps based on the Soil Landscapes
of Canada series. The risk class for water erosion occurring on bare, unprotected mineral soils is
provided in the map symbol for the dominant soil and the sub-dominant soil (if present) for each
map polygon. There are 5 water erosion risk classes that are useful to agriculturists, planners, soil
conservationists, and others concerned about soil quality at the regional level. The maps are of

7-9
limited use for site specific evaluations. Findings from the water erosion trend analysis will also be
available in tabular form and, in the future, in map form.

The soil erosion manual or handbook has been completed. Until improved technology is developed
for application in Canada, the manual focuses on the use of the USLE for soil and water
conservation planning. Besides utilizing the information published by Wischmeier and Smith
(1978), the manual contains recommendations for determining USLE factors for Canadian
conditions. Where appropriate, the manual incorporates the RUSLE technology as modified for
Canadian conditions. As well, a wide variety of C-factors for Canadian crops are available in table
format. Details of special cases as well as limitations are also presented in the document.

SUMMARY

A water erosion project was initiated to determine the potential for water erosion to impact on soil
quality. The project included plot-scale erosion monitoring to measure actual soil loss as well as soil
loss and redistribution based on 137Cs analysis for agricultural landscapes. The soil loss and
redistribution data were used for validation of soil erosion prediction equations for Canadian
conditions. Technology transfer publications were prepared to assess erosion risk at a regional scale
and to assist erosion prediction at the field level for soil conservation planning.

FUTURE WORK

Validation efforts of the RUSLE and WEPP models have been impeded as a result of delays in their
release. The RUSLE (Revised Universal Soil Loss Equation) model has only recently been
published (Soil and Water Conservation Society 1994) The model has been tested under USA
conditions but, to date, has not undergone any validation with Canadian data. This validation process
is very important given the importance of freeze/thaw cycles on water erosion in Canada. Recently
collected soil erosion data in BC and NB will be used to validate the model under Canadian climatic
conditions.

Preliminary releases of the WEPP model are being validated with erosion plot data from across
Canada as part of the cooperative effort with the USDA scientists. Winter erosion data from BC,
ON and MB will be used to further evaluate the winter runoff and soil loss routines of the newer
version of the WEPP model.

Once the WEPP model has been validated for Canadian conditions, the appropriate version will be
used to predict changes in soil erosion rates that may have resulted from the implementation of soil
and water conservation practices on agricultural land over recent years. In the interim, the soil
erosion trend analysis is continuing using the USLE.

The WEPP model will be used in conjunction with the 137Cs data to compare actual (as measured
with 137Cs) vs predicted soil redistribution on a landscape. This will highlight regions of the country
where tillage erosion is the dominant process.

7-10
 REFERENCES

Cao, Y.Z., Dickinson, W.T., Rudra, R. and Wall, G.J. 1993. Calibration and sensitivity analysis
of the WEPP hillslope model using simulated event data from southern Ontario. University of
Guelph, School of Engineering. Unpublished report.

Cao, Y.Z., Coote, D.R., Rees, H.W., Wang, C. and Chow, T.L. 1994. Effects of intensive potato
production on soil quality and yield at a benchmark site in New Brunswick. Soil and Tillage
Research 29:23-24.

Cao, Y. Z., Coote, D.R., Nolin, M.C. and Wang, C. 1993. Using 137Cs to investigate soil erosion
at two benchmark sites in Quebec. Can. J. Soil Sci. 73:515-526.

Hayhoe, H.N., Pelletier, R.G. and van Vliet, L.P.J. 1993. Estimation of snowmelt runoff in
the Peace River region using a soil moisture budget. Can. J. Soil Sci. 73:489-501.

Hayhoe, H.N., Coote, D.R. and Pelletier, R.G. 1992a. Freeze-thaw cycles, rainfall and snowmelt
on frozen soil in Canada. Climatological Bulletin 26:2-15.

Hayhoe H.N., Pelletier, R.G. and Moggridge, S. 1992b. Analysis of freeze-thaw cycles and
rainfall on frozen soil at seven Canadian locations. Can. Ag. Eng. 34:135-142.

Kachanoski, R.G. 1992. Evaluation of soil loss rates under different soil and cropping practices
in Prince Edward Island. Department of Land Resource Science, University of Guelph. Guelph,
ON. (Unpublished report).

Lobb, D.A. and Kachanoski, R.G. 1993. The relationship between landscape position, tillage
practices and soil loss: model development. Report to the National Soil Conservation Program.
Agriculture Canada. Harrow, ON.

Soil and Water Conservation Society. 1994. The Revised Universal Soil Loss Equation (RUSLE).
Soil and Water Conservation Society. Ankeny, IA.

van Vliet, L.J.P. 1993. Preliminary testing of the WEPP hillslope profile version model with
erosion plot data from Beaverlodge, Alberta. Agriculture Canada. British Columbia Land
Resource Unit. Vancouver, BC. Unpublished report.

Wischmeier, W.H. and Smith, D.D. 1978. Predicting rainfall erosion losses -a guide to
conservation planning. US Department of Agriculture, Agric. Handbook No. 537. US
Government Printing Office. Washington, D.C. 58 pp.

7-11
CHAPTER 8
EVALUATING CHANGES IN SOIL ORGANIC MATTER

E.G. Gregorich1, C.M. Monreal1, B.H. Ellert2,

D.A. Angers1 and M.R. Carter1
1
Agriculture and Agri-Food Canada, Research Branch, Centre for Land and Biological Resources
Research, Ottawa, Ontario
2
Agriculture and Agri-Food Canada, Research Branch, Lethbridge Research Station, Lethbridge,
Alberta
3
Agriculture and Agri-Food Canada, Research Branch, Sainte-Foy Research Station, Sainte-Foy,
Quebec
4
Agriculture and Agri-Food Canada, Research Branch, Charlottetown Research Station,
Charlottetown, Prince Edward Island

INTRODUCTION

Soil organic matter plays a key role in soil quality. It is a source of and a sink for plant nutrients in
soils and is important in maintaining soil tilth, aiding the infiltration of air and water, promoting
water retention, reducing erosion, and controlling the efficacy and fate of applied pesticides.
Because of its involvement in the different aspects of soil quality, Larsen and Pierce (1991)
suggested that soil organic matter is the single most important indicator of soil quality and
productivity.

Land use and agricultural management practices directly affect the amount, distribution, and
decomposition of organic matter in soil. Changes in soil organic matter may be evaluated using
qualitative, quantitative and predictive methods. Assessment of the effect of management practices
on the amount of organic matter in soils can be made at the landscape level and in smaller research
plots. However, the quantity of organic nutrients alone may not be the best indicator of soil quality.
Actively cycling fractions or pools of organic matter are sensitive to changes in soil management,
soil perturbations, and inputs in the soil. Therefore, evaluation of these properties may be important
in providing an early indication of longer-term trends in soil quality.

The concept of sustainable agriculture implies that a soil must sustain its ability to produce crops
over an extended period of time. In this context, the prediction of changes in soil organic matter
using simulation models becomes an important tool in assessing the impact of management practices
on soil quality.

This study was initiated to evaluate changes in soil organic matter in Canada. The objectives of this
study were to test and assess a simulation model capable of predicting changes in organic matter in
different soils/crop rotations in Canada, to develop criteria for evaluating soil organic matter quality,
and to assess the actual changes in soil organic matter associated with the adoption of agricultural
systems in Eastern Canada.

8-1
 THE CENTURY MODEL

A detailed description of the CENTURY simulation model is presented in Parton et al. (1987). In
the model, surface and root litter material are considered separately, with incoming plant material
divided into structural and metabolic material. Surface microbes are modeled separately and soil
organic matter (SOM) comprises three pools with different turnover times. The active pool consists
of live soil microbes and microbial products, the slow pool includes plant and microbial products
that are relatively resistant to decomposition and the passive pool is chemically recalcitrant or
physically protected. Decomposition of SOM is considered as a function of monthly soil temperature
and moisture. Soil texture controls the amount of stabilized and oxidized organic matter. The split
of the decomposing plant residue between structural and metabolic components is considered to be
controlled by the initial lignin to nitrogen (N) ratio of the residue. In addition CENTURY has input
files to assess the effects of management practices (e.g., fertilization, tillage, harvest, addition of
organic amendments) and of events, such as grazing and erosion, on biomass production and SOM
levels.

A sensitivity analysis of the model was conducted to determine the effects of changing specific input
parameters and variables on model output, i.e., soil organic carbon (C). Control simulation runs
involved scenarios with and without amendments of organic matter. The model was sensitive to the
rate of soil erosion, texture, rates of CO2 flow from soil organic matter pools, plant biomass
production, and removal of plant residues. Rates of fertilizer application and the lignin/N ratio had
smaller effects on model output (Monreal et al., 1992; Gregorich, unpublished information).

Simulating Soil Organic Matter Dynamics in Long-Term Rotations of Western Canada

The dynamics of soil organic matter were examined in long-term crop rotations established across
a climosequence of soils in western Canada. Organic C and total N in the 0-30 cm depth were
measured over periods of up to 80 y in soils under wheat-fallow (WF), wheat-wheat-fallow (WWF),
continuous wheat (CW) and barley (CB) rotations. These rotations were managed under
conventional tillage and no-till.

Management practices, type of rotation, and rate of soil erosion influenced the level of SOM across
all sites. On average, for a Chernozemic Brown soil cropped for 23 y with fertilization based on soil
tests, the amount of N decreased by 3% under WF, increased by 11% under CW, and was sustained
under WWF. The dynamics of C paralleled those of N. Cropping a Chernozemic Dark Brown soil
at Lethbridge for 80 y without fertilizer additions resulted in a decrease in the amount of C by 24%
under WF and WWF and 16% under CW. The dynamics of N followed that of C (Monreal and
Janzen, 1993). Between 1910 and 1990, the C/N ratio of soil under WF and WWF remained fairly
constant (9-10), however, it steadily increased under CW to a value of 11. Continuous removal of
N by grain was associated with faster N depletion under CW. The amount of C and N was sustained
under ten years of CB under no-till in a Chernozemic Black soil. In general, greater depletion of
SOM was observed in plots with lower inputs of plant residues and high rates of soil erosion (> 20
t ha-1 y-1.

8-2
Measurements of 137Cs activity indicated that physical removal of soil by erosion may outweigh the
effects of biochemical oxidation on SOM. Sixty-nine percent of soil organic matter losses were
associated with soil erosion in the Chernozemic Brown and Dark Brown soil. No erosion losses
were detected in the Chernozemic Black soil under zero-tillage.

The CENTURY model mimicked the SOM dynamics in the Brown, Dark Brown, and Black
Chernozemic soils. On average, for all sites, crop rotations, and years, CENTURY underpredicted
changes of C by up to 5% of measured values, and either over- or underpredicted N by up to 2% of
measurements (Figs. 8-1, 8-2). The greatest variation from observed values occurred for some
specific sites, crop rotations, and years, where CENTURY underpredicted C by up to 13% and
overpredicted N change by up to 20%. CENTURY is not appropriate for modeling soil organic
matter dynamics in Solonetzic soils because the effects of unique hydro-geochemical processes on
the C and N cycles are not represented in the model (Monreal et al., 1992).

CENTURY accurately predicted changes in soil organic C and N as influenced by soil fertility
treatments in a relatively short time under continuous corn in southwestern Quebec (Table 8-1). The
positive effects of cattle and poultry manure applications on soil organic matter levels in Ontario
and Quebec also were predicted with CENTURY. Simulated values of C were within 5% of
measured values in soils that had begun receiving manure applications after 10 years of declining
C levels due to continuous corn and high rates of erosion (Fig. 8-3).

IMPACT OF MANAGEMENT ON THE QUANTITY AND COMPOSITION

OF SOIL ORGANIC MATTER

A study was initiated in which soils under agricultural crops and adjacent forests were compared
to evaluate the impact of agriculture on the quantity and composition of soil organic matter. Soils
under different management regimes were sampled at 22 sites in eastern Canada. The sites were
carefully selected to provide a detailed history of soil management and to minimize the effects of
soil redistribution from erosion and tillage. Each site included a native forest and one or more field
crops or pasture; four profiles were sampled under each management type at each site.

The mass of C in cultivated soils was about 35% less than in forested soils. Losses of C exceeded
the losses of N at all sites, consequently, the mass of N in cultivated soils was about 20% less than
in forested soils. Several cultivated soils had greater levels of C and N than adjacent forested soils.
The decline in C levels that occurred with deforestation and cultivation was greatest on light-textured
soils. In a poorly drained soil, C levels were significantly higher in areas where tile drains had been
installed. A comparison between fertilized and unfertilized soils showed that annual application of
fertilizers over 35 years increased soil C levels, primarily as a result of greater amounts of residues
returned to the soil. The light fraction (< 1.8 g cm-3), isolated from soils by flotation on dense liquids
(Gregorich and Ellert, 1993), accounted for a larger proportion of the soil mass and whole-soil C
in forest soils than in cultivated soils. The light fractions were enriched in C relative to the whole
soils and had a C/N ratio of 20 to 30, suggesting that the carbon-rich

8-3
 Fig. 8-1. Predicted change in organic C in a Chernozemic Dark Brown soil.
Solid lines = observed, dotted lines = predicted.

Fig. 8-2. Predicted change in total N in Chernozemic and Solonetzic soils.

Solid lines = observed, dotted lines = predicted.

8-4
 Fig. 8-3. Observed and predicted levels of organic C in a Luvisolic
soil. Solid line = observed, dotted line = predicted.

Table 8-1. Predicted soil organic C (OC) and N (ON) by the CENTURY model and percentage
differences (PD) between predicted and measured values

Normal Fertilizer Rate ‡ High Fertilizer Rate

Year
OC PD ON PD OC PD ON PD
(g m-2) (%) (g m-2) (%) (g m -2) (%) (g m-2) (%)
1984 4000 - 461 - 4074 - 479 -
1985 4320 - 420 - 4479 - 433 -
1986 4355 - 411 - 4564 - 426 -
1987 4376 4.0 410 0.7 4634 6.0 429 -1.2
1988 4421 - 413 - 4720 - 435 -
1989 4423 - 412 - 4755 - 437 -
1990 4455 -4.5 413 -1.9 4822 -1.7 441 -0.9
1991 4456 - 414 - 4852 - 444 -
1992 4466 - 414 - 4889 - 446 -
1992 † 4490 0.7 415 3.5 4938 -1.6 450 3.7

† Values for the fall of 1992.

‡ Normal=170-100-170 kg N-P2O5-K2O ha-1, and High=400-300-400 kg N-P2O5-K2O ha-1.

8-5
material in the light fraction is a potential source of energy for the soil microbial biomass. The C
enrichment (% C in light fraction/% C in whole soil) in the light fraction of cultivated soils was
generally greater than in the light fraction of forest soils because cultivation caused greater decreases
in the C concentration of whole soils than in the light fractions.

Natural 13C abundance was measured in a forest and in adjacent cultivated soil (under continuous
corn for more than 25 y) to estimate the quantity of carbon derived from corn, as well as the turnover
of SOM (Gregorich et al., 1992). About one-third of the organic matter in the plow layer had turned
over since the start of continuous corn cropping. The turnover of organic matter was most rapid in
the sand fraction and slowest in the coarse silt faction.

The effects of crop rotation and management on the quantity and quality SOM were also studied
by cross-polarization magic angle spinning (CP/MAS), solid 13C nuclear magnetic resonance (NMR),
and pyrolysis field ionization mass spectrometry (Py-FIMS). CP/MAS 13C NMR showed that 70 and
23% of the soil organic C consisted of aliphatic and aromatic structures, respectively. Excessive
tillage decreased SOM and most chemical structures were lost from the sand and silt fractions of
a Gleysol. Removal of organic matter with cultivation did not discriminate against the major types
of organic structures. On average, aliphatic and aromatic compounds of SOM showed similar
turnover rates. Therefore, biological recalcitrance may not be a consequence of an increased content
of resistant aromatic structures, as previously believed (Monreal et al., 1993a).

Studies conducted with aggregates showed that SOM is protected within macro-aggregates and the
disruption of these units by tillage results in significant deterioration of soil quality (Monreal et al.,
1993b). Py-FIMS permitted the identification of almost 60% of the macromolecules in SOM of
Gleysolic and Chernozemic soils. An aggrading Chernozemic Brown soil under CW had a greater
amount and variety of soil organic compounds than a degrading WF rotation. Carbohydrates and
phenolic plus lignin monomers were the most abundant (< 10-15%) classes of identified SOM
compounds. Sterols, lignin dimers, and other long-chain fatty acids were found at lower
concentrations (< 4%) but were responsible for the stability of macro-aggregates (> 250 fm).

BIOLOGICAL PROPERTIES USED TO ASSESS SOIL ORGANIC MATTER

QUALITY IN AGRICULTURAL SOILS

Several biological properties that are indicators of a soil's ability to accept, store, and release energy
can be used to assess SOM quality, each providing a specific characterization or facet of SOM
quality. To characterize how soil quality changes over time, these properties must be sensitive to
changes in soil management, soil perturbations, and inputs into the soil system.

Soil organic matter quality may be evaluated by measuring the following biological properties: total
organic C and N, light fraction, mineralizable C and N, microbial biomass, and soil enzymes.

8-6
Total Organic Carbon and Nitrogen

Organic carbon and nitrogen contents in soil are a result of a complex interaction between additions
of C and N through fertilizers and plant and animal residues and losses of C and N through
microbial decomposition, mineralization, and erosion. Changes in inputs, such as fertilizers and
returned residues, and changes in mineralization rates and microbial activity will ultimately be
reflected in the total C and N content. Temperature and moisture are the most important factors
influencing mineralization rates in soil, and the impact of management practices on C and N levels
varies with soil climate. Management practices, such as the type of tillage, cropping frequency, type
of crop, and the addition of inorganic and organic amendments, affect C and N levels (Campbell
et al., 1990).

Light Fraction

The light fraction of soil contains residues derived from plants, animals, and microorganisms in
various stages of decomposition. It serves as a readily decomposable substrate for microorganisms
in soil and is an important reservoir of plant nutrients. The light fraction usually represents a small
portion of the soil by weight (0.1 to 3% of the total weight of cultivated soils) but yields up to a
tenfold concentration of organic materials, compared with the whole soil. Thus, the proportion of
organic matter in the light fraction may respond more quickly and provide an earlier indication of
changes in organic matter than the total level of soil organic matter.

Mineralizable Carbon and Nitrogen

Mineralizable C and N are derived from the fraction of soil organic matter that receives inputs of
fresh plant residues and that yields outputs of CO2 and inorganic N when the residues are
decomposed. More than 75% of soil organic matter exists as compounds that are only slowly
decomposable, whereas less than 25% is present as readily decomposable or mineralizable
compounds. Mineralizable soil organic matter is a critical indicator of organic matter quality because
it determines nutrient dynamics within a single growing season, organic matter content in soils under
contrasting management regimes, and C sequestration over more extended periods.

Microbial Biomass

The soil microbial biomass functions both as an agent for the transformation and cycling of organic
matter and plant nutrients within the soil and as a sink or source of labile nutrients. In the latter role,
the microbial biomass accounts for 2 to 5% and 2 to 6% of soil organic C and N, respectively. Thus,
it functions within the soil as a store of labile organic matter. Due to its dynamic nature, microbial
biomass quickly responds to changes in soil management and soil perturbations and is sensitive to
various toxicities in soil. Microbial biomass has also been shown to be related to various soil
structure indices (Carter, 1992).

8-7
Soil Enzymes

Soil enzymes are synthesized by plants and soil organisms that transform simple and complex
substrates to satisfy their metabolic needs. These proteins are found in living organisms (biotic
enzymes) or in dead cells of microbial and plant tissues (abiotic enzymes). Soil enzymes represent
molecular subsystems of the microbial biomass and can be used as indicators of soil quality if their
activities are sensitive to the effects caused by pollutants and farming practices, such as cultivation,
summerfallowing, fertilization, and pesticide application.

Carbohydrates

Soil carbohydrates originate from plants, animals, and microorganisms. They account for 5 to 20%
of the total organic C in soil and represent one of the largest chemically identifiable fractions of soil
organic matter. In the free state they are readily degradable and are primarily used metabolically as
sources of energy by microorganisms, but their persistence in soil has been attributed to their
involvement in soil aggregation processes and their chemical structure, which limits metabolism.
Soil carbohydrates have been studied primarily in relation to their role in soil aggregation. Good
correlations between carbohydrate content and soil aggregate stability (Angers et al., 1993) suggest
that the labile fraction of the carbohydrate pool could be used as a sensitive indicator of changes in
organic matter quality.

ACKNOWLEDGEMENTS

We thank Drs. C.V. Cole, W.J. Parton, and A.K. Metherell for providing us with updated versions
of CENTURY and for detailed discussions on the theory and application of its use. We also thank
Drs. P.H. Groenevelt, A.F. MacKenzie, and B.C. Liang for providing data on organic matter content
and erosion from their research plots for testing the model. We express our sincere thanks the
scientists from the Agriculture and Agri-Food Canada Research Stations at Lethbridge, Swift
Current, and Vegreville, as well as the scientists from the Departments of Soil Science at the
Universities of Alberta and Saskatchewan; without their help and cooperation this study would not
have been possible.

REFERENCES

Angers, D.A., Bissonnette, N., Legere, A. and Samson, N. 1993. Early changes in water-stable
aggregation induced by rotation and tillage in a soil under barley production. Can. J. Soil
Sci. 73: 51-59.

Campbell, C.A., Zentner, R.P., Janzen, H.H. and Bowren, K.E. 1990. Crop rotation studies on
the Canadian prairies. Research Branch, Agriculture Canada, Publication 1841/E.

8-8
Carter, M.R. 1992. Influence of reduced tillage systems on organic matter, microbial biomass,
macro-aggregate distribution and structural stability of the surface soil in a humid climate.
Soil Tillage Res. 23: 361-372.

Gregorich, E.G. and Ellert, B.H. 1993. Light fraction and macroorganic matter in mineral soils.
Pages 397-407 in M.R. Carter, Ed. Soil sampling and methods of analysis. Lewis Publishers
of CRC Press, Inc., Boca Raton, FL.

Gregorich, E.G., Monreal, C.M., and Ellert, B.H. 1992. Natural carbon-13 distribution in forested
and cultivated Gleysols. Pages 158-166 in Management of Agricultural Science. Proceedings
of the Saskatchewan Soils and Crops Workshop '92. Saskatoon, SK. 20-21 Feb.

Larsen, W.E. and Pierce, F.J. 1991. Conservation and enhancement of soil quality. Pages 175-203
In: Evaluation for sustainable land management in the developing world. Vol. 2: Technical
Papers. Bangkok, Thailand. International Board for Soil Research and Management
(IBSRAM) proceedings No. 12.

Monreal, C.M., Zentner, R.P, Campbell, C.A., Nyborg, Greer, K., and Gregorich, E.G. and
Anderson, D.W. 1992. Simulating the dynamics of soil organic matter in long-term rotation
plots of Saskatchewan and Alberta. Pages 206-236 in Management of Agricultural Science.
Proceedings of the Saskatchewan Soils and Crops Workshop '92. Saskatoon, SK. 20-21 Feb.

Monreal, C.M. and Janzen, H.H. 1993. Soil organic dynamics after eighty years of cropping a
Dark Brown Chernozem. Can. J. Soil Sci. 73: 133-136.

Monreal, C.M., Gregorich, E.G., Schnitzer, M. and Anderson, D.W. 1993a. The quality of soil
organic matter as characterized by CP/MAS 13C NMR and Py-FIMS. In P.M. Huang, Ed.
Environmental impacts of soil components interactions. Soil Quality, organic components.
Lewis Publishers of CRC Press, Inc. Boca Raton, FL. (In press.)

Monreal, C.M., Campbell, C.A., Anderson, D.W. and Schnitzer, M. 1993b. The effect of organic
structures on the water stability of macro-aggregates. Pages 55-66 in Crop quality. What does
it mean? Proceedings Soils and Crops Workshop, University of Saskatchewan, Saskatoon,
SK.

Parton, W.J., Schimel, D.S., Cole, C.V. and Ojima, D.S. 1987. Analysis of factors controlling soil
organic matter levels in Great Plains grasslands. Soil Sci.Soc. Am. J. 51: 1173-1179.

8-9
CHAPTER 9
ASSESSING AND MONITORING SOIL SALINIZATION

R.G. Eilers1, W.D. Eilers2, W.J. Stolte3,

M. Trudell4, R. Stein4, W. Read5 and H. Vander Pluym5
1
Agriculture and Agri-Food Canada, Research Branch, CLBRR Manitoba Land Resource Unit,
Winnipeg, Manitoba
2
Agriculture and Agri-Food Canada, Research Branch, CLBRR Saskatchewan Land Resource Unit,
Saskatoon, Saskatchewan
3
University of Saskatchewan, Department of Civil Engineering, Saskatoon, Saskatchewan
4
Alberta Research Council, Edmonton, Alberta
5
Alberta Agriculture, Food and Rural Development, Lethbridge and Edmonton, Alberta, respectively.

INTRODUCTION

Soil salinization is a major soil quality issue in many agricultural landscapes on the Canadian
Prairies. The extent of salinized lands has been estimated at 2.2 million hectares or about 6.0 per cent
of improved agricultural land. This translates into an estimated annual financial loss of $104 to $257
million to Canadian prairie agriculture (Dumanski et al. 1986).

The process of soil salinization is primarily related to long-term hydrologic and climatic conditions.
High water tables, high evaporation rates, and the presence of soluble salts in the soil system are the
prime requisites for soil salinization.

The majority of saline soils on the prairies were present long before the advent of cultivated
agriculture. Nevertheless, there is a widely held perception that agricultural practices have
contributed to a significant and widespread increase in the extent of soil salinity.

Many agricultural practises have changed the local hydrology of the landscape by changing the
vegetation, rate of infiltration and surface runoff. These changes have resulted in a redistribution
of water in the landscape and thus have changed the primary factors controlling the process of soil
salinization. Information about the impact that common agricultural management practices have on
changing the status of soil salinity is needed for designing sustainable land management practices.
Methodologies and tools (techniques) are required for developing effective methods of controlling
the salinization process, mitigating the salinity problem if possible, and/or developing sustainable
alternative land use scenarios.

Within the overall context of soil quality assessment and monitoring, salinity must be assessed in
terms of its deleterious impact on productivity. Crop yields decline at various rates related to the

9-1
inherent salt tolerance of the plants. Increases in either the extent or degree of salinity results in
decreased soil quality and hence decreased crop yield.

Soil salinity can adversely affect soil quality - productivity in three ways. Firstly, high concentrations
of soluble salts restricts the ability of plants to utilize available soil water by creating an unfavorable
osmotic differential between the plant root and the soil solution. Secondly, the dispersive effects of
abundant sodium in the soil can cause a deterioration of soil structure, resulting in restriction of root
penetration and the physical flow of water and nutrients. Thirdly, toxic influences of certain anions
or cations present in saline soil solution may adversely affect the nutritional balance of plants. The
end result is a serious reduction in the capacity of the soil to produce food and fibre crops.

USER REQUIREMENT AND STUDY OBJECTIVES

The development of management practices for salinized soils must be designed with consideration
for the long-term impact to control, mitigate, sustain or ameliorate the quality of the soil. This
requires improved capabilities to monitor and predict change in soil salinity as a function of applied
management practices. The specific goals of this study were:

i) to assess the extent and severity of soil salinity at selected benchmark sites representative
of typical agricultural landscapes and hydrologic settings on the prairies.

ii) to utilize simulation process models (SEEP/W and CTRAN/W) and field monitoring
procedures to assess the impact of agricultural management practices on the status of soil
salinity, and

iii) to determine the type and magnitude of the change in soil salinity under known conditions
of weather and land use through annual and seasonal monitoring of benchmark sites.

Description of Approach and Rationale

Salinity is fundamentally a water management problem as controlled by the surface morphology of

the landscape, the type of surface cover and the stratigraphy and hydraulic gradients below ground
level. The long-term evaporative loss of water from the soil surface eventually results in the
accumulation and concentration of soluble salts on or near the soil surface. On this premise, a
multi-disciplined study team consisting of pedologists, hydrologists, agriculturalists, and civil
engineers was established to develop a monitoring program, to monitor the seasonal and annual
salinity status and to investigate, through simulation process modelling, the parameters controlling
salinity in several selected landscapes on the Canadian Prairies.

9-2
 VALIDATION OF SALINITY CHANGE ON THE CANADIAN PRAIRIES

Documentation of soil quality change as a function of soluble salts requires a series of reference
points in representative agricultural landscapes

Benchmark Sites, Distribution and Characteristics

Seven sites distributed throughout the Prairies were selected to serve as long-term benchmark sites
for investigating and monitoring salinity change. These sites as listed below, will serve as baseline
reference points to which future analysis and assessments will be compared. These are the first
monitoring sites to be established on the Canadian Prairies with the specific purpose to assess the
dynamics of soil salinity change with time, climate and land use.

The monitoring sites are also intended to serve an important extension function for promoting soil
salinity awareness, education extension, and discussing soil quality as affected by salinization in the
landscape. Farmers, conservationists, scientists, and students will benefit from visiting these sites
and from the information being gathered.

Province Site Name Investigator (Agency)

Manitoba Warren Site R.G. Eilers (CLBRR)1

Saskatchewan Cory (Dalmeny) Site W.D. Eilers (CLBRR)

St. Denis Site W.D. Eilers (CLBRR)
Prairie View Site W.D. Eilers (CLBRR)
Alberta Lunty (Forestburg) Site M.R. Trudell (ARC)2
Crossfield Site B. Read (ADA)3
Blackspring Ridge Site R. Stein (ARC)

1
Centre for Land and Biological Resources Research,
2
Alberta Research Council,
3
Alberta Department of Agriculture, Food and Rural Development.

The distribution of these sites is shown in Fig. 9-1 and the general characteristics of each site are
presented in Table 9-l.

9-3
 Fig. 9-1. Location of salinity monitoring sites on the Canadian
Prairies.

Table 9-1. General characteristics of the salinity monitoring sites

ALBERTA

Forestburg
Blackspring Ridge Crossfields
(Lunty Site)
Soil Zone: Black Dark Brown Black
Genetic Material: Morainal Lacustrine Morainal
Textural Class: Clay loam Clay loam Loam
Surface Expression: Hummocky Level Undulating
Slope Class: 4 (5.0 - 9.0%) 2 (0.5 - 2.0%) 3 (2.0 - 5.0%)
Purpose: (Original) Ph.D. Res. Site ARC Study Site DSIS1
(Current) Monitoring Monitoring Monitoring
Location: SW4, SE5-41-15-W4 SW12-13-23-W4 SE26-28-28-W4

- Dryland Salinity Investigation Service

9-4
 SASKATCHEWAN

Cory St. Denis Prairie View

Soil Zone: Dark Brown Dark Brown Brown

Genetic Material: Lacustrine Morainal Morainal
Textural Class: Clay loam Loam Loam
Surface Expression: Undulating Hummocky Undulating
Slope Class: 2 (0.5 - 2.0%) 5 (10.0 - 15.0%) 3 (3.0 - 5.0%)
1 2
Purpose: (Original) Cropland CWS -NHRI Res. PFRA3
(Current) Salt monitor Salt monitor Salt monitor
Location: SW 32-37-6W3 SW 29-37-1W3 SW 18-18-11W3
1
Canadian Wildlife Service
2
National Hydrology Research Institute
3
Prairie Farm Rehabilitation Administration

MANITOBA

Warren Hamiota

Soil Zone: Black Black

Genetic Material: Lacustrine Morainal
Textural Class: Clay Clay loam
Surface Expression: Level Hummocky
Slope Class: 1 (0.0 - 5.0%) 3 (2.0 - 5.0%)
Purpose: (Original) GDA1 Crop Tol. M.Sc./Ph.D. Research Site
(Current) Monitoring Model validation
Location: a) SW 35-13-2W1 (King Subsite) SW11-14-23-W1
b) NW 26-13-2W1 (Gallant Subsite)
1
General Development Agreement, Canada - Manitoba

9-5
 SOIL SALINITY MONITORING, MODELLING AND ASSESSMENT

Monitoring

The monitored parameters at each of the benchmark sites include: weather, soil salinity, soil climate,
groundwater table levels and groundwater chemistry. Only the soil salinity data will be discussed
here.

The soil salinity monitoring activities have focused on the seasonal and annual dynamics of salt
movement in the soil. Spring, summer and fall grid surveys were conducted using electromagnetic
induction (EM38) instruments on each of the seven benchmark sites. Soil samples were taken at
permanent soil sampling sites for monitoring salt movement within the profile and calibration of
the EM38 data. Based on the first three years of data it is not apparent that the extent of salinity is
increasing. Annual fluctuations occur in salt concentration with depth in the profile (Fig. 9-2), but
there does not appear to be any significant expansion of the area affected by the different levels of
salinity (Fitzgerald and Eilers, 1993).

The EM38 data was utilized to evaluate the local variation in salinity and to evaluate the timing and
frequency of monitoring activities. This monitoring will be continued for the next few years to
document trends in the status of salinity at these sites.

In the long term, this data will be used to assess the impact that agricultural management as well as
climate variations may have on the level of salt in the soil. This information has already been used
as a guide and as a check for validation of the mathematical modelling activities at each site.

Modelling

Simulation modelling is a major component of this salinity assessment and monitoring study. The
first goal of the mathematical modelling was to facilitate an understanding of the environmental and
physical factors affecting the processes of salinization. Two commercial modelling programs,
SEEP/W and CTRAN/W, were adapted to five of the monitoring sites and one test site. Data input
for the models was obtained from site data or data for similar soils and conditions in the near
vicinity. The models were run on each study site to predict the occurrence and severity of salinity
within each landscape. Through an iterative process, incorporating data on long-term average
climate, hydrology, geology, and salinity, the models have been conditioned to accumulate water
flow and salts in the appropriate positions of the landscape as verified by field inspections (Fig. 9-3,
Stolte et al., 1992).

9-6
Fig. 9-2. Annual variation in salinity as measured by
EM38 at the King subsite near Warren,
Manitoba.

9-7
Fig. 9-3. Measured and simulated distribution of water and salt
along a hillslope; 9.3a. Measured bulk electrical
conductivity (EM38 data); 9.3b. Head contours and
flow vectors for the end of August (Aquifer Head =
16m, PET over lower 25 m); 9.3c. Predicted salinity
distribution along hillslope for the end of August
(Aquifer Head = 16m, PET over lower 25m) (After
Stolte et al. 1992).

9-8
Mathematical modelling of the salinization processes in each of the soil landscapes indicates that the
major factor influencing the accumulation of soluble salts is the evaporative flux at the soil surface.
A preliminary sensitivity analysis of a test site at Hamiota, Manitoba, shows that varying the
evapotranspiration (ET) rates along the length of a hillslope causes water from the slough to be
drawn upslope through the more permeable subsoil. With time, soluble salts gradually accumulate
in the upslope position causing salinity zones around sloughs (Stolte et al. 1992). Changing the depth
at which ET occurs had little effect, while changing the hydraulic conductivity of the top soil
affected the amount of water in the sloughs. Reducing the hydraulic conductivity reduces the amount
of flow upslope out of the slough and, consequently, reduces the amount of salt that is transported
upslope. However, changing the surface slope steepness and the hydraulic head in the aquifer had
significant effects on the position of occurrence of salt in the landscape. Increasing the steepness of
slope caused the salts to accumulate in lower slope positions.

Utilization of this modelling process has allowed elucidation of the various factors that influence and
control the salinization process and has given insight into the magnitude of response of soil salinity
to changes in these various parameters.

Assessment and Prediction of Soil Quality Change Due to Salinization

The third component of this study, an overall assessment of salinity status, has not been completed
as yet. Determining the status of soil salinity on the Prairies at a given point in time is a complex
undertaking. Information and techniques learned from this study indicate that simulation modelling
utilizing landscape, climate and land use data will be an indispensable component of this assessment.
Modelling combined with the use of surrogate indicators of salinity status will facilitate extrapolation
of site information to the broader scale. For example, changes in land management practices to a
more efficient water-use production system could indicate a potential for reduced risk of salinization.
This analysis could be done after each agricultural census which would provide land use information
on a landscape basis. Assessment of the change in land use practices between census years could be
incorporated in the model for a new assessment of salinity. The landscape units would be those
utilized in the last prairie region assessment of extent and severity of soil salinity which was
completed in 1989 (LRRC, 1:1m scale maps) using existing soil maps and data bases. The
utilization of cropping trends as a surrogate indicator of the status of salinity change is a feasible and
practical undertaking since it is often considered to be an indicator of soil quality.

Simulation modelling will facilitate identification of those landscapes where there is likely to be
change in salt status under conditions of climate or land use practice change over time. A series of
transects distributed according to landscape map units in areas where sufficient land use data is
available could be used for model verification.

9-9
 EVALUATION OF THE IMPACT OF MANAGEMENT PRACTICES
ON SOIL SALINIZATION

Assessing the impact of common land use scenarios on the distribution and accumulation of salinity
in the landscape is required to develop effective and efficient remediation practices. Sensitivity
analysis to evaluate the response of soil salinity to altered hydrologic (climatic, biologic and
geologic) parameters has provided a new insight to alternative practices.

At this point in the study, the modelling has concentrated on simulating the accumulation of salts
in the observed locations rather than trying to actually derive the measured concentrations or
chemistry of the salt present. Currently, field monitoring methods are being utilized to determine
annual and seasonal changes. It is known that salinity is the net result of external climatic factors,
however, seasonal and annual changes maybe a function of short-term climatic conditions combined
with land use practices. These relationships will be examined in a future study to design and
implement remedial measures.

Future modelling activities will also utilize simulation techniques to design and evaluate the efficacy
of a series of management practices prior to their implementation. Modelling will thus facilitate an
evaluation of alternative mitigation practices and provide an indication of the probable time
requirement for changing soil salinity. Modelling techniques represent a viable management tool for
use in designing sustainable management practices for salinized lands. The two-dimensional models
used in this study have proven to be an indispensible part of the study and have lead to a greater
understanding of the possible mechanisms affecting soil salinization in each of the landscapes
modelled to date.

Adequacy of Benchmark Sites and Additional Capabilities

The benchmark sites were chosen to represent a wide range of soil, landscape, geological and
climatic conditions throughout the prairie region. The information gained on the short- and longterm
dynamics of salts in these landscapes will be applicable to similar landscapes in the vicinity of these
sites. However, direct extrapolation of the salinity status of these sites to the entire prairie region is
not possible. In order to assess the salinity status of the entire region, information gained on the
controlling processes at these sites must be combined with other sources of information. Provincial
soil testing data, rural land assessment, crop insurance and soil conservation salinity investigation
and characterization sites (PFRA and provincial), Canada Census, geology and hydrology data sets
and climatic data sets all represent potential supplemental data sets which could provide information
on long-term trends on a much broader scale.

The documentation of soil quality as determined by salinity can be examined from two points of
view. Firstly, at the regional level, a relative assessment of risk of change could be undertaken.
Secondly, additional sites or landscapes could be investigated and monitored to verify the risk
analysis. Future monitoring capabilities should incorporate some or all of the following activities to
facilitate assessment of soil quality change due to salinity.

9-10
1. Use of satellite sensing technology to track land use changes in landscapes as a function of
salinity. Using the soil landscapes map as a base, those landscapes (polygons) with a high ratio
of fallow land could be identified as highest risk for soil quality change due to salinity, while
those landscapes with continuous permanent vegetation cover would represent the lowest risk
units. Landscapes with combinations of fallow, annual crops and permanent cover would be
grouped to reflect some relative intermediate risk for soil quality change due to salinization
processes.

2. Use of generalized simulation modelling tools to evaluate landscape sensitivity to salinity

change.

3. A census of soil salinity management and remediation practices adopted in specific landscapes
to provide another surrogate indicator of change in salinity status.

4. A statistical review of the crop production trends in selected landscapes based on the inherent
tolerance of the crop to salinity. Note, this may simply mean that a more sustainable
management has been adopted for these soils rather than a change in management due to
increasing salinity.

5. Establishment of a network of transects through designated salinized landscapes and collection

of EM38 salinity data. Transects would be geo-referenced and would be monitored on a
seasonal/annual basis depending on site and available resources. Data would be calibrated
with standard temperature and moisture conditions, and utilized for validation of simulation
modelling analysis.

6. Continued EM38 grid surveys of salinity benchmark monitoring sites established under NSCP.

Items 1 and 2 can be considered techniques of assessing risk to soil quality and a means of
extrapolating site information (items 5 and 6) to the broader scale. Items 3 and 4 could be
considered a kind of change verification process.

CONCLUSIONS

Grid and transect monitoring using EM38 instruments appears to be an efficient and effective
technique for monitoring seasonal and annual soil salinity.

Mathematical modelling has been an indispensable part of this salinity investigation. The models
SEEP/W and CTRAN/W have been easily adapted to simulate observed hydrologic and saline
conditions in the field. Sensitivity analysis has shown that one of the most significant factors
affecting salinity is the magnitude of moisture deficit (precipitation-potential evapotranspiration,
P-PET) and where it occurs in the landscape. In this regard, land use i.e. cropping practice is a major
influence on the water regimes and thus on salinity. There is a very close relationship between the

9-11
type of vegetation cover, available and/or excess water in the soil profile and the concentration of
soluble salts in the root zone.

Preliminary results of this study indicate that much of the localized salinity in the landscapes are
locally derived from transient surface ponding and shallow subsurface groundwaters. Therefore, to
evaluate salinity change over long periods of time will require monitoring of salinity and land use
coupled with a dynamic modelling capability.

Through information gained in this salinity monitoring and assessment study it is hoped that
changes in short term climate and land use on landscapes can be related and used to predict the
change in salinity status with in those landscapes and hence the change in soil quality due to
salinization.

ACKNOWLEDGEMENTS

The authors are grateful to S.L. Barbour, D. Swanson, D. Willems, and S. Kumar for the application
and running of the modelling portion of this study; to R. Hahn, O. Beisel, V. Machacek, G. Lunty,
M. King, D. Gallant, and the Canadian Wildlife Service for providing access to their land; and to
D.F. Acton and W. Harron for continued encouragement and support of this monitoring and
prediction study.

REFERENCES

Dumanski, J., Coote, D.R., Luciuk, G.M., and Lok, C. 1986. Soil conservation in Canada. J. Soil
and Water Conserv. 41: 204-210.

Fitzgerald, M.M. and Eilers, R.G. 1993. Utilization of the EM-38 to monitor changes in soil
salinity. Pages 64-75 in Proc. of 36th Annual Manitoba Society of Soil Science Meetings.
Manitoba Agriculture.

Stolte, W.J., Barbour, S.L. and Eilers, R.G. 1992. A study of the mechanisms influencing salinity
development around prairie sloughs. Trans. ASAE 35: 795-800.

9-12
CHAPTER 10
ASSESSING CHANGE IN SOIL PHYSICAL QUALITY

R.A. McBride1, D.W. Veenhof1, G.C. Topp2,

Y.T. Galganov2 and J.L.B. Culley2
1
Department of Land Resource Science, University of Guelph, Guelph, Ontario
2
Agriculture and Agri-Food Canada, Research Branch, Centre for Land and Biological Resources
Research, Land Resource Division, Ottawa, Ontario

THE SCOPE OF SOIL PHYSICAL QUALITY ASSESSMENT USED IN THIS STUDY

Soil physical quality affects the soil's ability to function for the production of food and fibre, and
also as a porous medium for the partitioning of water and gaseous exchange and for environmental
buffering. In this chapter, we will consider the physical conditions within the upper soil plant root
zone or solum, knowingly excluding those very important physical conditions at the immediate soil
surface as well as other important aspects including soil structural stability. Soil pore space is of
critical importance for water, air and nutrient supply to plant roots. In addition, soil strength
determines or limits the growth and distribution of plant roots and influences the load bearing
capacity and trafficability of the soil. The studies reported in this chapter were aimed at developing
or adapting procedures to assess selected aspects of soil physical quality, such as the structural
condition of agricultural soils. The results presented have provided an indication of the current status
of soil structural quality for a range of agricultural soils in Canada, and the methods introduced can
be applied at differing spatial scales and at differing levels of refinement.

The suitability of physical rooting conditions in a soil is a function of complex interactions involving
soil strength and the supply of air and water to plant roots. An attempt has been made to integrate
this complex interaction into a single parameter, the Non-Limiting Water Range (NLWR) (Letey,
1985). Water contents that lie within this range are those where the water available for root uptake
is not limited by lack of aeration for root respiration, by excessive mechanical resistance to root
growth or by water too tightly sorbed by the soil. Methods are described which were used to
determine the NLWR for the upper profile of soils from various locations across Canada. The results
of using these methods on a variety of soils subjected to varying levels of intensity of cultivation and
management serve to illustrate the potentials and limitations of the NLWR parameter. The NLWR
concept was used in a relative sense at each site to assess the degree of soil physical degradation
from long-term cultivation. Some limitations of the NLWR concept have been identified at both
ends of the water availability range. Identification of the soil strength level which effectively
restricts root elongation and the uptake of soil water remains elusive and difficult to measure, thus
limiting the application of the NLWR parameter as a general indicator of soil structural quality.

10-1
The NLWR is a dynamic soil parameter and can be influenced by various forms of soil degradation.
For example, soil compaction would have the effect of narrowing the NLWR by reducing the
plant-available water-holding capacity of the soil, by increasing the water content at which the
threshold penetration resistance is reached (i.e., 2 MPa), and by decreasing the water content at
which the threshold air-filled porosity is reached (i.e., 10% m3 m-3). A method of assessing the
compaction risk from farm traffic based on soil mechanical properties and their interaction with
climate has been developed and applied in the Regional Municipality of Haldimand-Norfolk in
Ontario. Some of the same concepts were evaluated further through application in other
municipalities across southwestern Ontario. Knowledge of soil consistency limits and other basic
soil properties in combination with knowledge of agricultural cropping practices and other relevant
agronomic information show some promise for estimation of soil compaction risk at regional scales.
This chapter summarizes findings from McBride (1993) and Topp et. al (1994) undertaken
concurrently to address the assessment of soil structure as part of the Soil Quality Evaluation
Program.

OBJECTIVES

Very little is known in absolute terms about either the Non-Limiting Water Range or the current
state of compaction of agricultural soils in Canada. A recent survey commissioned by Agriculture
and Agri-Food Canada showed that agricultural soil compaction was the soil and water conservation
problem identified most frequently by corn (maize) producers in Ontario as a concern on their farms
(Deloitte and Touche, 1991). It has further been estimated that 50 to 70 percent of the fine-textured
soils of southwestern Ontario have been adversely affected by soil compaction, with 3/4 of this
affected land area rated as moderately compacted and 1/4 as severely compacted (Can-Ag
Enterprises, 1988). A conservative estimate places the cost of this form of soil degradation to the
Ontario agricultural economy at about $21M annually (Science Council of Canada, 1986).

In view of this deficiency in basic knowledge of soil physical quality, the following broad objectives
were set out under this study:

i) to develop procedures to measure aspects of the soil physical quality of agricultural

soils and the current state of these physical qualities.
ii) to assess the influence of current agricultural land management practices (traffic,
tillage and cropping patterns) on these physical qualities.

Procedural development under the first objective concentrated on well-established measures in-
cluding soil strength (penetration resistance, stress-strain relationships), soil hydraulic properties
(water retention, infiltration) and soil plasticity. The assessment of soil compactibility and its
agronomic significance can be viewed as a sequence of three complementary stages involving: i)
the estimation or measurement of susceptibility to total soil porosity change, ii) soil compaction
probability or risk assessment, and iii) the interpretation of soil physical property changes with
respect to their agronomic impact (e.g. NLWR) (Voorhees, 1987). All of these stages are represented
in this chapter in addressing the second objective stated above.

10-2
 MEASUREMENT OF SOIL PHYSICAL QUALITY

Measuring Stress-Strain Relationships for Agricultural Soils

Static, Uniaxial Compression of Structurally-Intact Soils

Twelve field sites were selected in the spring of 1991 within the Regional Municipality of
Haldimand-Norfolk. This part of southwestern Ontario was selected due to the wide textural range
of the soils, the existence of recent soil inventory mapping at both detailed (1:25,000) and
reconnaissance (1:100,000) scales, and the availability of considerable soil characterization data
gathered during the inventory, including long-term soil hydrological information. These sampling
locations were representative of the more important soil associations and landscapes in the region
that are currently being used for corn production. Most of these locations were near sites where
groundwater table depths had been monitored for more than a decade (Hohner and Presant, 1985).
All sites were cropped in corn in the year of sampling (1991).

One pedon-sized sampling area (about 1 m2) was selected between corn rows at each field site.
Structurally intact soil cores and bulk soil samples (about 2 kg) were taken from two levels in the
plow layer (0-7.5 cm, 7.5-15 cm) and from one level in the underlying horizon (generally 1530 cm).
A split-sleeve coring device fitted with aluminum sampling rings (8.25 cm diam. x 3.0 cm high) was
used for core sampling. The bulk samples were collected for the purpose of routine physical and
chemical analysis and for remoulded (slurry) compression testing. Ten replicate soil cores were
collected at each sampling depth. Excess soil was left on the samples so that the cores could be more
carefully trimmed to the proper size in the laboratory. Cores were wrapped in plastic to limit water
loss and were placed in 4ºC storage in the laboratory until they were to be prepared for compression
testing. Bulk samples were air dried and sieved to 0.425 mm. Routine soil physical and chemical
analyses consisted of particle-size distribution using the pipette method (Gee and Bauder, 1986),
carbonate content by the gravimetric technique (Nelson, 1982), and organic carbon content by
LECO induction furnace (Nelson and Sommers, 1982). Particle density (Blake and Hartge, 1986)
was determined from trimmings of excess soil taken from intact cores prior to compression testing.
Compression tests were carried out on the soil core samples in either a saturated condition or after
desorption to prescribed matric potentials (-10, -100, -1500 kPa) in a pressure plate apparatus. Each
of the four treatments (one saturated, three unsaturated) consisted of two replicate cores randomly
chosen from the ten sampled from the same depth. Core samples were trimmed to the desired sample
height (3.0 cm) prior to saturation in distilled and de-aired water. After equilibration on the pressure
plate, the unsaturated treatment cores were weighed and the circumference was trimmed to fit the
Rowe cell sample chamber by using a custom-made sample cutter (7.2 cm inside diam.). The excess
soil from core trimming was oven-dried at 105ºC for 48 h for gravimetric water content
determination.

Static, uniaxial (drained) compression tests were carried out using two ELE Rowe type consolidation
cells (sample chamber 7.6 diam. x 3.0 cm high) equipped with a regulated air pressure system.
"Fast" compression testing (Bradford and Gupta, 1986) was used to generate the
density-moisture-stress relationships, with increasing loads of 0.02, 0.05, and 0.1 to 1.0 MPa in 0.1
MPa increments. Loading times of 30 min were used throughout for each consecutive stress level.
The effective cell diameter was reduced to 7.2 cm because of the porous plastic liner required for

10-3
sample drainage during "drained" tests. Volume displacement was measured with calibrated MPE
electronic displacement transducers connected to a Campbell Scientific 21X datalogger. Soil
pore-water pressure was also measured electronically using a Sensotec A-10 pressure transducer.
Volume displacement readings were recorded at 30 s intervals, while pore-water measurements took
place every 10 s for the saturated core tests and every 60 s for the unsaturated tests. At the end of
an individual compression test, the datalogger was immediately downloaded to a microcomputer.
Void ratios were calculated and the compression data were plotted in void ratio (e) vs. logarithm
of total stress (F) co-ordinates.

A simplex non-linear regression procedure (CoHort, 1990) was used to fit the compression data to
the equation of Bailey et al. (1986). The preconsolidation stress (FcN) was calculated using a
computer model developed by McBride (1988) which numerically determines the first and second
derivatives of the Bailey et al. (1986) equation from the fitted parameter coefficients. The model
constructs the tangent to the point of maximum curvature and the bisector of the virgin compression
line (VCL) from the derivatives of the non-linear regression equation to derive FcN as in the
Casagrande method. The preconsolidation stress is defined as the maximum vertical stress (in kPa)
that a soil has been subjected to in its past, and is thus a useful indicator of a soil's prestress history.
All other statistical analyses were carried out using the PC-SAS statistical analysis package (SAS
Institute Inc., 1985).

Remoulded (Slurry) Consolidation

Consolidation of soil slurries was carried out using a modified version of the instrument described
in McBride and Baumgartner (1992). The modified consolidometer is based on the same principles
but it allows for higher total pressures up to 500 kPa. Instead of having free drainage towards the
cell periphery, it allows water drainage through a porous plastic rod situated in the centre of the
sample chamber. The applied load is transmitted to the soil slurry via a triaxial latex membrane
which is placed around the inside of the cell body. Pressurized air enters the aluminum cell body
which is lined with porous plastic to aid in distributing pressure equally to the latex membrane. The
latex membrane applies pressure radially inward (similar to a triaxial cell) and displaces water
towards the drainage rod and onward to the open drainage port. Primary consolidation is complete
when the air pressure measured in the pore-water pressure sensor tubes drops to zero.

The 36 bulk soil samples taken from the 12 soil profiles (three sampling depths) in the R.M. of
Haldimand-Norfolk, referred to in the preceding section, were supplemented with the same 20
agricultural soil horizon samples from Middlesex County which were reported on by McBride and
Baumgartner (1992). All 56 soil samples were air dried and sieved to 0.425 mm in preparation for
slurry consolidation testing using similar procedures to McBride and Baumgartner (1992). For the
Haldimand-Norfolk soils, six to ten points were measured within a stress range of 10 to 300 kPa to
enable accurate construction of the "normal consolidation line" (NCL). The Middlesex soils had
been tested previously (McBride and Baumgartner, 1992), so these materials were only consolidated
at stresses of 30, 100 and 300 kPa. Drop cone penetration tests (BS 1377:1975 Test 2A) as well as
liquid limit (wL) and plastic limit (wp) Atterberg tests (A.S.T.M. D423-72 and D424-71,
respectively) were routinely carried out on soil samples recovered from the slurry consolidometer
upon completion of each overnight consolidation test. Values were calculated from the w(log FN)
functions, or NCL, for the "effective stress at the liquid limit" (FNwL) and "effective stress at the
plastic limit" (FNwP). Values for the latter which were beyond FN of 300 kPa were determined by

10-4
extrapolation of the NCL to the measured wp test indices. All regression analyses were carried out
using the General Linear Models procedure of PC-SAS (SAS Institute Inc., 1985).

Measuring Non-Limiting Water Range (NLWR)

The measurement of NLWR included both a field and laboratory component where the upper part
of the soil-water desorption curve was determined from field measurements and the full range of
the desorption curve was measured on soil cores taken to the laboratory. The limiting conditions for
aeration were estimated from the shape of the upper part of the desorption curve from both the field
and laboratory measurements over this range of desorption. The upper limit of the NLWR was
determined by choosing the lower of either the water content value at which there would be at least
10% air-filled pore space or the water content when the soil-water potential had reached -10 kPa.
The lower end of the NLWR was the higher of either the water content when the penetrometer
resistance reached 2 MPa or when the soil-water potential reached -1500 kPa (permanent wilting
point). The strength of the soil at the different soil-water potentials was measured as a penetrometer
resistance of the soil in the cores.

The Soil Sites and Treatments

The sites for making these measurements were chosen on the basis that the sites were already used
as a well-documented research or farm site on which there was additional information on the
performance of the soils for crop production. We also attempted to investigate a wide range of
cultivation practices and history in order to assess, in some sense, the sensitivity of the NLWR to
differing cultural practices. A total of eight sites from different climatic regions across Canada were
chosen for this study, but only data from four of these sites will be discussed in the results to follow.
In general, three infiltration rings were installed as replicates in each treatment. The 8 sites and soils
were Lethbridge AB clay loam, Oxbow loam (Termuende, SK), Indian Head SK heavy clay,
Osborne clay (Brunkild, MB), Brady sandy loam (Clinton, ON), Huron clay (Clinton, ON),
Conestogo/London loam (Elora, ON) and Charlottetown PE sandy loam. The treatments discussed
in this report are given in Table 10-1.

Infiltration and Redistribution of Water in the Field

The upper part of the soil-water desorption curve was measured at depths of 10, 20 and 30 cm in
the field using both tensiometers to record the soil-water potential and time-domain reflectometry
(TDR) to determine the water content. The field experiments were necessary to assure that field
conditions were approximated by the laboratory measurements, particularly in relation to aeration
limits (60x cone x 3.6 mm diam.) . The soil was initially wetted to a field-saturated condition by
infiltration into a 40 cm diameter ring from an applied head of 10 cm until increases in water content
and potential were no longer recorded at the 30 cm depth. At this point, the supply of infiltration
water was shut off and the water content and tensiometer potential were recorded periodically
during the ensuing time period of 24 to 48 h. At the end of this time period the infiltration ring,
the tensiometers and the TDR probes were removed from the soil.

10-5
Intact soil cores (7.6 cm diam. by 7.6 cm high) were taken from soil which was not disturbed by
tensiometers or TDR probes but which had been wetted during the infiltration. The cores were taken
vertically and centred at 10, 20 and 30 cm depths. The cores were sealed in plastic bags for
transport to the laboratory.

The tensiometers and the TDR probes were installed before any wetting of the soil had taken place.
Six ceramic tensiometer cups, 2 cm diam. by 5 cm length, with "O"-ring seals to varying length
extension tubes were installed vertically within the 40 cm diameter ring to be wetted. Pilot holes
were drilled with a 1.9 cm diam. spiral wood auger bit powered by a portable electric drill. Soil in
very dry and crumbly condition was prewetted with a spray bottle to prevent collapse of the pilot
hole. The prewetted tensiometer was carefully pushed into the drilled pilot hole to the
predetermined depth. At the soil surface, direct infiltration down the tensiometer tube was
prevented by 5 cm diam. rings around each tensiometer inserted 4 cm into the soil. The tensiometer
potentials were recorded by a CR21 datalogger (Campbell Scientific) from pressure transducers
(Sens-Sym) connected to each tensiometer.

The 40 cm diam. infiltration ring was hammered 5 cm into the surface soil to prevent leakage of the
infiltrating water beyond the perimeter of the ring. A perforated metal sheet with 1.5 mm holes
giving approximately 30% opening coverage was suspended 5 mm above the soil surface to
distribute the water and dissipate its erosive energy. A 10 cm head of water was maintained during
infiltration using a float valve (livestock watering control). The supply of water to the float valve
was from a 20 litre bottle whose weight was being recorded by a digital bathroom scale connected
to the CR21 datalogger giving a continuous record of the infiltration rate.

In two diametrically opposite locations, pits were excavated in the soil to allow the horizontal
installation of the TDR probes from the outer edge of the 40 cm diameter ring. The soil pits were
about 20 cm wide by 35 cm deep and extending 50 cm out from the edge of the ring. Two pronged
TDR probes, 15 cm in length and consisting of 6.35 mm diam. rods at 50 mm spacing were inserted
on a horizontal plane at depths of 10, 20 and 30 cm in each of the two pits. The TDR measured
water content was determined by visual interpretation of the TDR pulse travel time directly from
the display of a 1502C Tektronix cable tester.

Friction Sleeve Penetrometer in Soil Cores at Selected Soil-Water Potentials

In the laboratory, the soil-water desorption curves were measured according to the procedure
described by Topp et al. (1993). This procedure included saturation of the cores by immersion in
water for one day followed by desorption on glass bead and aluminum oxide tension media to give
the upper portion of the desorption curve (0 to -33 kPa). The lower part of the curve was
determined from a pressure plate apparatus used on the intact cores. The seven selected matric
potentials were -0.35, -6, -10, -33, -100, -400, -1500 kPa. When the soil had reached each of the
potentials the penetrometer resistance was measured at two locations in each core using the friction
sleeve penetrometer (60 x cone x 3.6 mm diam.) as described by Bradford (1986). Thus, a total of
14 penetrometer measurements were made within each core. Two penetrations were made over the
two depth ranges at the centre of each core, and over one depth range at 12 locations on a circle
whose radius was 1.9 cm. This allowed each penetration to be separated from a neighbouring
penetration or the cylinder wall by 1.9 cm in the horizontal direction. The penetrometer resistance
was taken as a visual average of the measurements recorded at each mm increment over the depth

10-6
of 10 to 30 mm or 40 to 60 mm. After measurements were completed at -1500 kPa the soil was
oven-dried at 105ºC and weighed for determination of the bulk density of the soil as originally
sampled.

OVERVIEW OF SOIL QUALITY ASSESSMENT RESULTS

Compressive and Consolidation Behaviour of Agricultural Soils from the R.M. of

Haldimand-Norfolk

Soil Compression Results (Structurally-Intact Soils)

Static, uniaxial compression lines for 285 soil cores (saturated and unsaturated) were successfully
fitted to the equation of Bailey et al. (1986) with coefficients of determination (r2) greater than 0.98
in all cases (Fig. 10-1). The non-linear regression parameter coefficients were used to calculate the
"compression index", or slope of the "virgin compression line" (VCL), and the preconsolidation
stress for each structurally intact soil sample. Using data on the amounts of clay and organic matter,
degree of saturation and initial void ratio, a regression equation was formulated that allowed
prediction of the compression index and the preconsolidation stress of the soils used in this study.

These compression data also showed that one of the two "state" variables of a soil, the initial void
ratio, exerts a dominant influence on the configuration of compression lines in e(log a') coordinates
for structurally intact soils. Larson et al. (1980) showed that the other "state" variable, initial soil
water content, played a large part in determining the compression characteristics of unsaturated
soils. These compression experiments, however, were conducted on prepared aggregate beds in
which the initial void ratio was kept relatively uniform for a given soil. Compression data are shown
in Fig. 10-1 for the sandy clay loam Colwood Ap horizon (7-15 cm depth) that are very typical for
most of the soils investigated in this study. The compressive behaviour of this topsoil under
saturated and unsaturated conditions shows that the compression index increases with increasing
initial total soil porosity and the degree of saturation has only a limited influence on the relative
positions of the compression lines. Overall, the compression index showed a strong correlation with
the initial void ratio, the initial water content and the previous stress history for both plastic and
non-plastic soil types in this region of Ontario. An important conclusion drawn from this study is
that the widely adopted model of Larson et. al (1980) does not perform well on structurally intact
Ontario soils due to the overriding influence that the initial void ratio has on the configuration of
the compression line, even on treatment replicate soil cores.

The compression data also showed that the preconsolidation stress of the soils investigated was
curvilinearly correlated to a "degree of compactness" index similar to the one proposed by
Häkansson (1990). In this case, the index (Dv) is simply the quotient of initial void ratio (e0) in situ
and a reference void ratio determined by static uniaxial compression to 0.2 MPa. This index was
also found to be sensitive to changes in soil strength with variation in soil water content (i.e.,
strength increases with desaturation). A threshold index value of Dv = 1.3, representing a somewhat
arbitrary division between soils that are susceptible to significant degrees of soil compaction from
those that are not, was determined based on the relationship between Dv, the measured
preconsolidation stress and the soil water content at the plastic limit.

10-7
Fig. 10-1. Saturated and unsaturated compression lines in e(log FN) co-ordinates for a sandy
clay loam Colwood Ap horizon (7-15 cm depth).

10-8
Soil Consolidation Results (Slurries)

Additional data were generated in this study that could be used to validate the slurry consolidation
method of estimating the Atterberg soil consistency limits for agricultural soils in southwestern
Ontario (McBride and Baumgartner, 1992). The positions of the A.S.T.M. wL and wP test indices
in w(log FN) co-ordinates were determined for a wide range of agricultural soils (56 soil horizons)
sampled from Middlesex County and the R.M. of Haldimand-Norfolk. It was found that the
logarithm of the organic carbon content was the soil variable that most influenced the value of the
effective stress at the liquid limit (FN wL) as interpolated from the NCL. Knowledge of the organic
carbon content of a soil prior to slurry consolidation testing would yield precise estimates of wL for
these soils (s.e.e. = 0.88% kg kg-1). The position of FNwP on the NCL, however, was found to be
significantly correlated to two soil variables, the amount of clay present and the interaction
between clay content and the logarithm of the organic carbon content.

Estimates of w, using predicted FNwP values were less reliable (s.e.e. = 2.24% kg kg-1). It was also
found that wP could be estimated as precisely using multiple regression with clay, silt and organic
carbon contents as independent variables (s.e.e.= 2.16% kg kg-1). Soils which had slopes of the NCL
greater than -5.00% kg kg-1 and/or had clay contents less than about 13% kg kg-1 were considered
non-plastic. It was thus reaffirmed that the slurry consolidation method has potential as an unified
procedural alternative for the estimation of the Atterberg limits of soils in southwestern Ontario.

Comparison of Stress-Strain Relationships (Remoulded vs. Structured)

By mechanically dispersing primary soil particles and aggregate binding agents in the process of
creating a slurry, it is possible to obtain a relative indication of soil structural condition (aggregate
strength), the degree of soil overconsolidation and the nature of the soil fabric of the same soil
when in a structurally intact condition by comparing their respective consolidation characteristics.
The comparison of virgin compression lines (saturated, structurally intact) with their remoulded
counterparts for the Haldimand-Norfolk soils indicated that the soil horizon depth (i.e., overburden
pressure), the degree of past mechanical disturbance (i.e., tillage), the mode of soil deposition and
soil plasticity are all important determinants of the consolidation behaviour of agricultural soils. The
graphs shown in Fig. 10-2 are from the same profile of the Kelvin soil series (loam textured till)
and illustrate the full range of possible mechanical behaviour. In a well-aggregated soil, the NCL
would be expected to plot below the VCL under saturated conditions (i.e., at lower void ratios for
a given stress load) due to the structural bonds (organic and mineral) that would impart added soil
strength to the structurally intact soil; bonds that would be disrupted during compression. This is
true only in the case of the upper portion of the Kelvin series plow layer (Fig. 10-2a). At the base
of the Ap horizon, this relative displacement between the two compression lines is lost (Fig. 10-2b),
indicating a more overconsolidated condition in situ and the likelihood of increased orientation of
phyllosilicates due perhaps to high vertical stresses from wheel loading and/or shear from soil
engaging implements or rotating tires/tracks. In the horizon lying beneath the plow layer, a high
degree of overconsolidation is indicated in the field sample since the VCL plots well below the
NCL (Fig. 10-2c). This effect is exaggerated to some extent by a relative increase in clay content
in the subsoil, but the highly compacted state of this underlying horizon is confirmed by
considering that the saturated gravimetric water content (particularly of intact replicate 2) translates
into a "liquidity index" of approximately zero.

10-9
Fig. 10-2. Compression lines (structurally intact and remoulded) measured under saturated conditions for the
Kelvin soil series profile sampled at three depths a) 0-7.5 cm, b) 7.5-15 cm, and c) 15-30 cm.

10-10
In other words, when this subsoil horizon is field-saturated, it is barely within its "plastic"
consistency range and will behave more like a semi-solid.

It is, therefore, possible to segregate the soils of the R.M. of Haldimand-Norfolk into three
categories based on the relative positions of these stress-strain functions under saturated conditions.
Furthermore, the study results suggest that this relationship between the VCL and NCL for a given
soil is a more sensitive measure of the degree of soil overconsolidation than the more traditional
preconsolidation stress variable.

Changes in the NLWR of Selected Soils

Overall Comparison of NLWR vs. Traditional Available Water Capacity (AWC)

The NLWR was calculated, as described in a previous section, from the laboratory measured
desorption curves. The more traditional plant available water capacity of the soil (AWC) was
calculated as the difference between the water content at -10 kPa and that at -1500 kPa. In almost
all cases NLWR is less than the traditional AWC. For the soils under discussion, the generally high
penetrometer resistances indicate that soil strength has possibly restricted root growth more than has
lack of aeration. Only about 20% of the horizons showed inadequate aeration porosity, whereas over
90% of the horizons showed > 2MPa penetrometer resistance at potentials above -1500 kPa.

The NLWR was compared to the AWC in the form of the percentage reduction in plant available
water if NLWR is used in place of AWC (Table 10-1). The cultivation treatments have been
arranged in the table from the most intensive at the top of the table to the least intensive at the
bottom. Considering each soil separately, one can see a downward trend in overall reduction of the
NLWR:AWC ratio from top to bottom of the table. Cultivation can enhance the NLWR in the tilled
layer as shown for the conventional tilled Clinton soil and 15 years of conventional tillage at
Termuende. By contrast, however, the action of cultivation decreases the NLWR in the layer
immediately below the cultivation as shown particularly for the Termuende, Winnipeg and Clinton
soils. The great improvement in NLWR at the 20 cm depth from 15 years of no-till in the Clinton
soil can be attributed to earthworm activity and less soil compaction effects from traffic.

In Fig. 10-3, the left profile is the field-measured initial water content. The other labelled lines
represent values measured from the soil core data. The heavy-dotted line (not labelled) is the field
measured water content profile at -10 kPa. In the soil tilled since 1910, the -10 kPa profile is down
near a water content of 0.2 m3 m-3, close to the value found for the cores when the penetration limit
of 2 MPa was achieved. Whereas the cores, wetted by submersion for 16 h, gave a water content
profile near 0.3 m3 m-3 after desorption to -10 kPa. With the other two treatments, there was good
agreement between the field-measured and core-measured water contents at -10 kPa. This is an
indication that the longer duration cultivation has resulted in the soil becoming resistant to wetting.
Water repellency was also found in the Lethbridge soil under continuous wheat since 1912.

10-11
Table 10-1. The percentage reduction in water available to plants when Non-Limiting Water Range
(NLWR) is used in place of the usual available water capacity (AWC)

Lethbridge AB Termuende SK Winnipeg MB Clinton ON

Loamy Loamy Clayey Sandy

depth treatment % treatment % treatment % treatment %

10 cont. 35 conv. 38 conv. 21 conv. 12

20 wheat 20 till 48 till 34 till 66

30 for 80y 40 for 81y 46 21 99

total 34 44 24 61

10 wheat 46 conv. 8 no 32

20 fallow 25 till 41 till 0

30 for 80y 17 for 15y 21 for 15y 75

total 29 20 39

10 native 0 native 19 grass 30

20 grass 45 grass 22 23

30 35 30 32

total 27 23 29

The initial water content profiles in Fig. 10-3 indicate, generally, that the growing wheat crop, which
was harvested prior to the measurement, or the native grass had extracted water to lower than -1500
kPa and much lower than the 2 MPa penetration limit. This and similar findings for the Lethbridge
site are in agreement with the findings of Cutforth et al. (1991). It is worth noting that NLWR is
based on allowing water uptake essentially without restriction to plant growth. The maturation of
a wheat crop, by contrast with most other growth stages, is less severely restricted by reduced water
uptake. Irrespective of this, the 2 MPa penetration limit may be too low and needs to be reassessed
in relation to crop performance in the field. Until a more critical assessment of this soil strength
criterion has been undertaken, we have used the ratio of NLWR:AWC for the assessment presented
above (Table 10-1). Adopting the assumption that the -1500 kPa water content is primarily
determined by soil constituents and affected only slightly by soil management, we reasoned that
changes in the ratio of NLWR:AWC within a soil are indications of soil structural
degradation/agradation even when the limits for NLWR are not fully rationalized.

10-12
Fig. 10-3. Water content profiles for the soils at the Termuende Farm showing how the
Non-Limiting Water Range is related to various profiles.

10-13
 ESTIMATING THE CHANGES IN SOIL PHYSICAL QUALITY

Estimation of the Susceptibility of Soils to Compaction and the Current Status of the Problem
in Southwestern Ontario

The Risk of Soil Compaction in the R.M. of Haldimand-Norfolk

An evaluation of a soil water balance simulation model (SWATRE) was conducted for some of the
major soil associations and landscapes where corn was grown in the Regional Municipality of
Haldimand-Norfolk during the 1991 cropping season. The SWATRE model requires soil physical
properties and crop characteristics along with climatic data as input in order to estimate soil water
content profiles on a daily time step. Soil input data included a combination of measured and
predicted soil water retention curves in conjunction with measured values of field-saturated
hydraulic conductivity (Kfs). The predicted soil water characteristic curves were validated with
actual water retention data measured from structurally intact cores. It was found that there was very
good correspondence between measured and predicted data, with the exception of the soil water
content at a soil water potential of -1500 kPa. A predictive equation was developed for the
gravimetric soil water content at -1500 kPa for structurally intact, rather than disrupted, soils with
a standard error of estimate of 1.69% kg kg-1. Overall, the SWATRE model adequately simulated
the soil water content and soil water extraction and replenishment patterns for the texturally
disparate sites located in this regional municipality and under the climatic and cropping conditions
found in this part of Ontario.

The SWATRE model was used in conjunction with the "degree of compactness" index (Dv) and
climatic data for a 32 year period (1960-1991) to calculate probability distributions of exceeding a
compaction susceptibility threshold value of Dv equal to 1.3. Six sites representative of major soil
associations in the Regional Municipality of Haldimand-Norfolk were modelled at four depths (7.5,
12.5, 22.5 and 32.5 cm). The D, index was calculated for each site from the daily soil water contents
modelled from April 1 to November 30 for each of 32 years in order to determine the probability
distributions. Probability distributions for a loam textured plow layer at two depths are shown in Fig.
10-4. The results from this stochastic analysis suggest that the soils which are the most susceptible
to a significant loss of total soil porosity are the medium-textured soils (i.e., clay content of 20 to
35% kg kg-1) that are imperfectly to poorly drained. The fine-textured soil (clay content >39% kg
kg-1) that was modelled in this way was found to be susceptible to compaction at only the 7.5 cm
depth, while the lower depths were already sufficiently overconsolidated that they did not show any
significant vulnerability to further compaction at any point during the growing or harvesting season.
A coarse-textured (non-plastic) soil that was sampled from a no-till cropping system showed
susceptibility to compaction only at the 22.5 and 32.5 cm depths, while the plow layer (7.5 and
12.5 cm) showed sufficient structural strength that it was not susceptible to any significant porosity
loss at the 0.2 MPa stress load that was assumed throughout this analysis.

Use of Digital Image Analysis and GIS to Determine the Relation Between Agricultural
Cropping Patterns and Soil Associations in the R.M. of Haldimand-Norfolk

Current information was required on the distribution of agricultural crops (primarily corn and other
row crops) in relation to the major soil associations mapped in the regional municipality.

10-14
Fig. 10-4. Probability distributions for exceeding a compaction
susceptibility threshold of Dv = 1.3 for the Kelvin soil
series Ap horizon at two depths a) 7.5 cm and b) 12.5 cm.

10-15
Existing agricultural land use mapping was outdated (early 1980's) and was based on farm systems
rather than field-level crop classification. Landsat 5 Thematic Mapper imagery was obtained for this
region for two dates during the middle of the growing season when the study area was relatively free
of cloud cover (1 August 1990 and 3 July 1991). This study began in January, 1991, so visual
assessments were made in the field in the spring of 1991 of land areas cropped in corn during the
1990 crop year (i.e., corn stubble on plowed and unplowed fields). A more comprehensive survey
of crop cover for the 1991 crop year was carried out in the summer of 1991 by ground inspection.
Three large blocks were delineated within the regional municipality where the location and type of
common field crops were noted on orthophoto base maps. This information served as "training areas"
for digital image analysis with EASI-PACE and allowed statistical confidence levels to be
calculated.

The 1991 crop cover results showed that the total area of the major crop groups (hectares) within
three of the larger local municipalities as classified from the satellite imagery (Fig. 10-5)
corresponded well to total areas obtained from the 1991 Census of Agriculture (Table 10-2). Field
level verification of the 1991 crop classification further showed that the major row crop groups were
correctly identified at between 82% (soybeans) and 100% (tobacco) of the time. Corn was identified
correctly 88% of the time.

The generalized soil map (scale 1:100,000) for the regional municipality was not yet available in
digital form from Agriculture and Agri-Food Canada, so the map was digitized on the ARC/INFO
geographical information system and imported into the SPANS GIS. This scale of soil mapping was
chosen over the more detailed 1:25,000 inventory available for the region since the effective scale
of Landsat 5 imagery (30 m x 30 m pixels) is about 1:50,000. Overlay analysis of the crop cover (Fig.
10-5) and soil association spatial information was carried out with the SPANS geographical
information system. This analysis assisted in identifying soils that are the most susceptible to
traffic-induced soil compaction during planting and harvesting periods given a) their present use for
row crop production (including monoculturing), and b) the soil compression behaviour and soil
compaction risk assessment results for these soil associations as presented in preceding sections of
this chapter. These vulnerable soil associations included about 3300 ha of mainly lacustrine silty
clays (Brantford, Beverly and Toledo series), about 1700 ha of mainly silty clay loam tills (Muriel,
Gobles and Kelvin series) and about 2400 ha of mainly lacustrine silt loams (Brant, Tuscola and
Colwood series) (Table 10-2).

A Pedotransfer Function For Estimating Preconsolidation Stress And Its Application To

Agricultural Soils In Southwestern Ontario

In the section pertaining to comparison of stress-strain relationships on remoulded vs structured soils,

the hypothesis was tested and corroborated that the compressive behaviour (static, uniaxial) of
structured subsoils in the Regional Municipality of Haldimand-Norfolk can be predicted reasonably
well from the consolidation behaviour of those same soils when in a remoulded (slurried) condition,
and hence from their Atterberg consistency limits (McBride and Baumgartner, 1992). The exception
to this are soils that are sufficiently overconsolidated that their void ratio in the field translates into
a liquidity index at saturation of less than about 0.25. In view of this finding, a soil survey
interpretive procedure (or "pedotransfer function") was developed with the purpose of assisting in
characterizing the degree of overconsolidation of subsoils without the need for an extensive and
costly soil compression testing program. This pedotransfer function estimates a "first approximation

10-16
Fig. 10-5. Crop cover map for the Regional Municipality of Haldimand-Norfolk in 1991 as derived from digital image
analysis of Landsat 5 imagery.
10-17
Table 10.2 SPANS cross-tabulation results of map unit areas (hectares) from an overlay of crop cover and the generalized soil inventory
for the R.M. of Haldimand-Norfolk.

SOIL ASSOCIATIONS (Scale 1:100,000)* CORN AREA IN 1991 BY MUNICIPALITY (ha)

MODE OF SOIL CORRRESPONDING SOIL SERIES TOWNSHIP OF TOWNSHIP OF CITY OF
DEPOSITION NAMES (FROM 1:25,000 MAPPING) NORFOLK DELHI NANTICOKE
Mainly lacustrine silty clay Brantford, Beverly, Toledo 121 160 3054

Mainly lacustrine heavy clay Smithville, Haldimand, Lincoln - - 2504

Mainly silty clay loam till Muriel, Gobies, Kelvin 1194 536 -

Mainly lacustrine silty loam Brant, Tuscola, Colwood 870 368 1119
40-100 cm of sandy or loamy sediments Brookton, Berrien, Wauseon, Walsher, Vittoria,
1238 874 1721
over lacustrine clays or loams Silver Hill, Tavistock, Maplewood
Mainly lacustrine sands with Fox, Brady, Granby, Wattford,
2175 3367 1233
wind-modified surfaces Normandale, St. Williams, Lowbanks
Mainly eolian sands at least 100 cm Plainfield, Walsingham, Waterin
4121 1944 162
thick, often duned
Gravelly sands of fluvial or till Burford, Wilsonville, Scotland, Oakland,
derivation, or 40 to 100 cm of sandy Vanessa 41 756 541
sediments over gravelly sands
LANDSAT TM CLASSIFIED IMAGE AREA
9760 8005 10334
TOTAL FOR CORN:
CORN AREA AS A % OF MUNICIPALITY
23.0% 22.2% 20.8%
AREA THAT IS CLEARED AND
AVAILABLE FOR AGRICULTURE:

1991 CENSUS AREA TOTAL FOR CORN: 9634 6736 10195

* after Presant and Acton (1984).

10-18
preconsolidation stress" from dry bulk densities (void ratios) measured in situ and from other soil
properties needed to estimate the NCL. Preliminary testing of this simple function was carried out
by assembling the necessary data on soil physical properties for all soil horizons characterized by the
Ontario Centre for Soil Resource Evaluation (O.C.S.R.E.; formerly the Ontario Institute of Pedology)
for soil series mapped during the course of the last five county-level soil inventory upgrades in
southwestern Ontario and that met the minimum data requirements (i.e., 210 soil horizons from over
100 soil series profiles). The size of this data set was reduced considerably when soils that had a
liquidity index of less than 0.25 were eliminated as being severely overconsolidated and where an
estimate of the preconsolidation stress would likely have little meaning. These severely compacted
soil horizons were largely clay-rich materials, and were eliminated so that the assumption could be
made that the pedotransfer function would at worst tend to estimate conservatively.

In assembling the data set for statistical analysis using the General Linear Models procedure in
PC-SAS (SAS Institute Inc., 1985), we deduced that there were at least four factors that might have
some influence on the preconsolidation stress. Climatic variation across this region was not believed
to be sufficient to be included amongst the following factors: i) mode of deposition (geomorphology),
ii) soil family particle-size class, iii) soil horizon (depth), and iv) cropping history. Statistical
analysis was performed in accordance with a completely random design with unequal replication.
Variance (F) ratios showed that the soil horizon and the soil particle-size class variables were the
only significant factors (P< 0.05). Multiple range testing showed that the first approximation
preconsolidation stress increased significantly with depth (i.e., from the topsoil to the subsoil
horizons) and with increasing clay content. With the exception of some no-till Ap horizons, plow
layer horizons generally showed little evidence of overconsolidation due to the loosening effect of
regular tillage operations.

In summary, the solum (A and B) horizons of agricultural soils in southwestern Ontario are in a less
compacted condition than the C horizon subsoils. If a tillage- or traffic-induced compaction problem
existed, then it might have been expected to show up in higher estimated preconsolidation stresses
in the B horizon but this was not the case. Furthermore, many of the finer particle-sized soils are quite
overconsolidated, whereas the coarser soil groups are much less so. The soil horizons with the
greatest estimated preconsolidation stress values (mean > 150 kPa) tended to be glaciolacustrine
subsoil horizons that belonged to the fine clayey or very fine clayey particle-size classes and which
had low organic carbon contents. Therefore, we might expect the coarser soils to exhibit compressive
behaviour in the field very much like normally consolidated soils (i.e., very little prestress history in
evidence) with a high degree of vulnerability to soil compaction from wheel loading. Fortunately, the
amelioration of subsoil compaction by subsoiling is more easily achieved in these soils. The clay-rich
soils, however, are not likely to be as vulnerable to the loss of total soil porosity with normal wheel
loadings of 0.1-0.2 MPa at the soil surface, but may undergo other forms of soil structural
degradation (shear, plastic deformation) that can be equally or more limiting to root growth.

10-19
NLWR Parameter Estimates from Other Soil Data

The application of the NLWR concept as an indicator of soil physical quality depends on one being
able to estimate the NLWR from readily available soil data. This limited study indicates the
possibility for using the NLWR to indicate changing soil quality within a particular soil. Yet there
remains a need for continued testing of the NLWR concept to determine the conditions governing
its use for various soil and cropping situations. Here we address the possibility of estimating the
limits of the NLWR from existing data or measurements.

If the aeration limit is adopted based on a fixed air-filled pore space, then the upper limit of NLWR
can be estimated from soil bulk density and soil-water desorption data as was done for this study.
In many of the more detailed soil studies and surveys being undertaken, both the bulk density and
the -10 kPa water content are determined from core desorption measurements or directly in situ. Thus
the upper limit of the NLWR is a feasible estimate. There is some question now whether air-filled
pore space of 10%, as used here, is adequate (One dimensional oxygen diffusion: Comparison of
numerical and analytical solutions with an exponential respiration function. F.J. Cook, personal
communication).

Estimates of soil strength pertaining to root growth should emulate the soil deformation caused by
the root. Soil deformation around cone penetrometers is a reasonable approximation to that around
roots. The usual parameters (cohesion and angle of internal friction) used to characterize soil strength
are not easily nor directly related to cone penetrometer resistance. The strong dependence of soil
strength on water content or water potential further complicates the possibility of estimating the soil
strength at which root growth is restricted. In addition, there are few examples where basic soil
properties have been related to soil strength as a function of water content or potential. Koppi and
Douglas (1991) show a relationship between shear strength (the combined effect of cohesion and
friction) and clay content for water contents "close to field capacity". The increasing concern about
soil strength and stability of soil structure will lead to improved measurement and estimation
procedures over those which are available now. "Permanent wilting point", the usual lower limit of
availability of water (-1500 kPa), is a more readily obtained parameter, both because it has been often
a part of many laboratory routine procedures but also because a number of models are available to
estimate such values from the soil constituents (clay, sand and organic matter) and bulk density. In
regard to estimation of the NLWR the lower limit remains for now only the lower limit of available
water until better models relating soil strength, water potential and root growth are available.

SUMMARY AND ASSESSMENT OF THE PREDICTIVE CAPABILITY

OF THE PEDOTRANSFER FUNCTIONS

The compressive behaviour of many of the Haldimand-Norfolk soils under saturated and unsaturated
conditions showed that the compression index tended to increase with increasing initial total soil
porosity, but contrary to the findings of many studies involving structurally disrupted soils, the degree
of saturation had only a limited influence on the relative positions of the compression lines. It is
possible to segregate the soils of this regional municipality into three categories based on the relative

10-20
positions of measured stress-strain functions under structurally intact and remoulded conditions
(saturated). Furthermore, the study results suggest that this relationship between the VCL and NCL
for a given soil is a more sensitive measure of the degree of soil overconsolidation than the more
traditional preconsolidation stress variable. The results from the stochastic analysis suggested that
the soils which are the most susceptible to a significant loss of total soil porosity with wheel traffic
are the medium-textured soils that are imperfectly to poorly drained. Fine-textured soils were
frequently found to be already sufficiently overconsolidated that they did not show any significant
susceptibility to further compaction at any point during the corn growing or harvesting season. This
finding was corroborated by high preconsolidation stresses estimated with a simple pedotransfer
function for clay-rich soils in southwestern Ontario.

The NLWR concept, which incorporates and integrates a number of plant response factors, has been
shown to be useful when used relative to the traditional available water capacity. We have used the
NLWR to show that intense long-term cultivation has resulted in a decreased NLWR:AWC ratio. The
major factor indicating this decrease was an apparent increasing soil strength restricting the growth
of roots. The estimate and application of the aeration limit from soil data is feasible but the level of
air-filled pore space which is adequate requires validation. The estimate of limiting soil strength for
incorporation into the NLWR concept remains an impediment for its application based on the use of
existing soil data bases.

ACKNOWLEDGEMENTS

The assistance of Les Bober, Kathy Howe, Pamela Joosse and Duncan Wood was invaluable in
maintaining progress in the varied components of the soil compaction risk study. We also
acknowledge the financial assistance from the Environmental Youth Corps Program which funded
the summer positions of the latter two students in 1991 and 1992, respectively.

The assistance of Mark McGovern, Bruce Compton, Keith Wires and Jean-Marc LeClerc has been
absolutely vital to the NLWR method development and data collection. Co-op-work term students,
Peter Chesney, Cora Henderson and Tim Teefy, were often responsible for keeping things going and
maintaining a sense of appropriate urgency with good humour.

REFERENCES

Bailey, A.C., Johnson, C.E. and Shafer, R.L. 1986. A model of agricultural soil compaction. J.
Agric. Eng. Res. 33: 257-262.

Blake, G.R. and Hartge, K.H. 1986. Bulk density. Pages 363-375 in A. Klute, Ed. Methods of soil
analysis. Part 1. Physical and mineralogical methods. Agronomy Monograph No. 9, Amer.
Soc. of Agron., Madison, WI.

10-21
Bradford, J.M. 1986. Penetrability. Pages 463-478 in A. Klute, Ed. Methods of soil analysis. Part
1. Physical and mineralogical methods. Agronomy Monograph No. 9, Amer. Soc. of Agron.,
Madison, WI.

Bradford, J.M. and Gupta, S.C. 1986. Compressibility. Pages 479-492 in A. Klute, Ed. Methods
of soil analysis. Part 1. Physical and mineralogical methods. Agronomy Monograph No. 9,
Amer. Soc. of Agron., Madison, WI.

Can-Ag Enterprises. 1988. Assessment of soil compaction and structural degradation in the
lowland clay soils. A report prepared for Agriculture Canada under the Soil and Water
Environmental Enhancement Program (Technology Evaluation and Development sub-
program).

CoHort, 1990. CoStat. CoHort Software, Berkeley, CA.

Cutforth, H.W., Jefferson, P.G. and Campbell, C.A. 1991. Lower limit of available water for three
plant species grown on a medium-textured soil in southwestern Saskatchewan. Can. J. Soil
Sci. 71: 247-252.

Deloitte and Touche Management Consultants. 1991. 1990 survey of Ontario tillage and soil
conservation: Practices, perceptions and attitudes (executive summary). A report prepared
for the Agriculture Development Branch of Agriculture Canada. 32 p.

Gee, G.W. and Bauder, J.W. 1986. Particle-size analysis. Pages 383-411 in A. Klute, Ed. Methods
of soil analysis. Part 1. Physical and mineralogical methods. Agronomy Monograph No. 9,
Amer. Soc. of Agron., Madison, WI.

Häkansson, I. 1990. A method for characterizing the state of compactness of the plough layer. Soil
Tillage Res. 16: 105-120.

Hohner, B.K. and Presant, E.W. 1985. Seasonal fluctuations of apparent water tables in selected
soils in the Regional Municipalities of Niagara and Haldimand-Norfolk between 1978 and
1984. Ontario Institute of Pedology Publ. No. 85-6. 139 pp.

Koppi, A.J. and Douglas, J.T. 1991. A rapid, inexpensive and quantitative procedure for assessing
soil structure with respect to cropping. Soil Use Manage. 7:52-56.

Larson, W.E., Gupta, S.C. and Useche, R.A. 1980. Compression of agricultural soils from eight
soil orders. Soil Sci. Soc. Am. J. 44: 450-457.

Letey, J. 1985. Relationship between soil physical properties and crop production. Pages 277294
in B.A. Stewart, Ed. Advances in soil science. Springer Verlag, NY.

10-22
McBride, R.A. 1988. Estimation of compactibility and overconsolidation in unsaturated, structured
soils. Page 281 in Agronomy Abstracts, Amer. Soc. Agron., Madison, WI.

McBride, R.A., Ed. 199r. Soil compaction susceptibility and compaction risk assessment for corn
production in the regional municipality of Halimand-Norfolk. Final report to the Soil Quality
Evaluation Program Tehnical Report 3, Centre for Land and Biological Resources Research,
Research Branch, Agriculture and Agri-Food Canada, Ottawa, ON. 193 pp. (Appendix report
- 200 pp.)

McBride, R.A. and Baumgartner, N. 1992. A simple slurry consolidometer designed for the
estimation of the consistency limits of soils. J. of Terramechanics 29: 223-238.

Nelson, R.E. 1982. Carbonate and gypsum. Pages 181-197 in A.L. Page, R.H. Miller and D.R.
Keeney, Eds. Methods of soil analysis. Part 2. Chemical and microbiological properties.
Agronomy Monograph No. 9, Amer. Soc. of Agron., Madison, WI.

Nelson, D.W. and Sommers, L.E. 1982. Total carbon, organic carbon and organic matter. Pages
539-579 in A.L. Page, R.H. Miller and D.R. Keeney, Eds. Methods of soilanalysis. Part 2.
Chemical and microbiological properties. Agronomy Monograph No. 9, Amer. Soc. of
Agron., Madison, WI.

Presant, E.W. and Acton, C.J. 1984. The Soils of the regional municipality of HaldimandNorolk.
Report No. 57 of the Ontario Institute of Pedology, LRRI Contribution No. 8413. 100 pp.

SAS Institute Inc. 1985. SAS/STAT guide for personal computers. Version 6 ed. SAS Institute, Inc.,
Cary, NC.

Science Council of Canada. 1986. A growing concern: soil degradation in Canada. Ottawa, ON.
24 pp.

Topp, G.C., Galganov, Y.T., Ball, B.C. and Carter, M.R. 1993. Soil water desorpotion curves.
Pages 569-579 in M.R. Carter (Ed.). Soil sampling and methods of analysis. Lewis
Publishers of CRC Press Inc., Boca Raton, FL.

Topp, G.C., Galganov, Y.T., Wires, K.C. and Culley, J.L.B. 1994. Non-limiting water range
(NLWR): An approach for assessing soil structure. Soil Quality Evaluation Program
Technical Report 2, Centre for Land and Biological Resources Research, Research Branch,
Agriculture and Agri-Food Canada, Ottawa, ON. Contribution No. 94-41. 36 pp.

Voorhees, W.B. 1987. Assessment of soil susceptibility to compaction using soil and climatic data
bases. Soil and Tillage Research 10:29-38.

10-23
CHAPTER 11
PREDICTION OF AGROCHEMICAL MIGRATION

W.D. Reynolds, R. de Jong, S.R. Vieira and R.S. Clemente

Agriculture Canada, Research Branch,

Centre for Land and Biological Resources Research, Ottawa, Ontario

INTRODUCTION

This study is concerned with developing the capability to quantify soil quality in terms of the soil's
potential for preventing the pollution of ground water resources by the downward percolation of
agrochemicals (e.g. nutrients, pesticides). Such a capability would be of great assistance to scientists,
planners and policy makers in the development of soil quality inventories of "pollution potential",
and in the development of agricultural land use practices and guidelines that maintain agrochemical
inputs to the ground water at acceptable and sustainable levels.

As a first step, methodologies are being developed for characterizing and predicting the downward
migration of the widely used herbicide, atrazine. Spatial and temporal variability in the atrazine
migration is being accounted for via the combined use of a solute transport model, pedotransfer
functions, geostatistical analyses, and a geographical information system.

BACKGROUND

Although most pesticide contamination of ground water is below current Canadian drinking water
guidelines, there are growing public concerns over potential health hazards related to long - term
exposure to low levels of pesticides and their metabolites (Agriculture Canada, 1990). Pesticide
residues, especially atrazine, have been detected in surface, ground and tile drainage waters of many
agricultural watersheds, particularly where there is some combination of high pesticide usage,
intensive agriculture, high rainfall, irrigation, coarse and other highly permeable soils, high water
tables, and sloping topography (Millette and Torreiter, 1992).

Pesticide contamination of ground water has traditionally been considered to be due primarily to
spills, and to improper storage, disposal and application practices. There is increasing evidence,
however, that normal agricultural practices can also result in low-level, non-point source
contamination of ground water via the downward migration of pesticides through the soil profile
(Agriculture Canada, 1990). There is consequently a need to determine how important and
widespread this type of contamination might be, what the controlling soil quality, land use and
environmental factors are, and what agricultural practices are required to maintain this type of
pollution at acceptable levels.

11-1
Essential steps in obtaining the above information include, identification of the primary mechanisms
controlling pesticide movement through the soil profile, and development of the capability to
characterize and predict the pesticide movement in space and time with acceptable accuracy. The
approach being taken to achieve these steps is to employ a sophisticated solute transport simulation
model in combination with pedotransfer functions, geostatistical analyses and a geographical
information system (GIS). The solute transport model, which is an in-house modification of a well
established and tested modelling package called LEACHM (Hutson and Wagenet, 1989), integrates
the major processes that occur in the soil-plant-atmosphere system, including: soil horizonation;
saturated, unsaturated, steady and transient water flow; crop growth and transpiration; solute (e.g.
atrazine) sorption, degradation, advection and dispersion; precipitation and evaporation; soil heat
flow; and water table elevation. The pedotransfer functions are used to estimate, from available soil
data, the soil attributes that are required as input to the solute transport model (e.g. hydraulic
conductivity function, soil water characteristic). The geostatistical analyses are used to extend the
model predictions of solute percolation behaviour from a point basis (the model is one-dimensional)
to an areal basis (e.g. farmer's field, watershed), using procedures that take into account the inherent
spatial variability of the area. The GIS is used to create maps of the geostatistically extended solute
percolation behaviour, and to overlay these maps with those of soil attributes, land management
practices, cropping practices, weather, etc. Such maps and overlays are the "end product" which can
be used to show the importance and distribution of ground water contamination by downward
percolating pesticides; to determine the major soil, land use and environmental factors controlling
the contamination; and to estimate the potential environmental impact of changes in land
management practices.

The remainder of this report includes a brief description of the LEACHM modelling package and the
modifications made to it; a discussion of the testing and evaluation of two of the submodel
components; an outline of how the modelling package is being applied within the pedotransfer
function/ geostatistics/ GIS framework; and a preliminary application of the methodology for the
prediction of atrazine leaching through the soils of the Grand River watershed (Fig. 11-1. Greater
detail can be found in the corresponding technical report (Reynolds et al., 1994).

DESCRIPTION OF THE LEACHM MODELLING PACKAGE

LEACHM is a general acronym (Leaching Estimation And CHemistry Model) that refers to four
submodels of a large and comprehensive computer simulation package that describes the one -
dimensional storage, transmission and dissipation of water and solutes within the soil profile. These
submodels include LEACHW, which describes soil water flow only, LEACHP, which describes
sorption, migration and degradation of pesticides, LEACHN, which describes nitrogen transport and
transformations, and LEACHC, which describes the movement of inorganic salts. The submodels
consider a the main processes that occur in the plant root zone, including, transient fluxes of water,
heat and solutes; alternating periods of rainfall and evapotranspiration; and variable soil conditions
with depth. All submodels utilize similar numerical solution schemes, based on procedures developed
from several earlier models (Bresler, 1973; Nimah and Hanks, 1973; Tillotson et al. 1980).

Only the LEACHW and LEACHP submodels are being used in this study (they are discussed further
below). The reader is referred to Hutson and Wagenet (1989) for a more detailed discussion of the
entire LEACHM package.

11-2
Fig. 11-1. Location map for the Grand River Watershed.

11-3
LEACHW (Water Storage and Transmission Submodel)

The equation for transient vertical soil-water flow, derived from Darcy's law and the equation of
continuity, is described by:

Cw(2) Mh /Mt = (M/Mz) [K(2)(M H/M z)] - U(z,t) (1)

where Cw(2) = d2/dh is the differential water capacity relationship (L3 L-3 L-1), 2 is volumetric water
content [L3 L-3], H = h-z is hydraulic head [L], h is pore water pressure head [L], z is depth below the
soil surface [L], K(2) is hydraulic conductivity relationship (LT-1), U is a sink term representing
water uptake by plant roots [T-1] (calculated according to the Nimah and Hanks, 1973), and t is time
[T].

Empirical functions characterizing the soil water characteristic, 2(h) (used in the Cw(2) relationship),
and the hydraulic conductivity, K(2), are required in LEACHW. A combined parabolic-power
function is currently used to describe the 2(h) relationship (Clapp and Hornberger, 1978; Hutson and
Cass, 1987): The parabolic component is given by:

[1 - (2/2s)]½ (2i/2s)-b
h = -------------------------- ; hi # h # 0 (2)
[1 - (2i/2s)]½

and the power function component by:

h = a(2/2s)-b ; -4 # h # hi (3)

where hi = a[2b/(1+2b)]-b , 2i = 2b 2s/(1+2b), 2s is volumetric water content at saturation [L3 L-3],

and a and b are empirical constants. The point, (hi, 2i) locates the intersection of the parabolic and
power function segments. The corresponding hydraulic conductivity relationship is given by
(Campbell, 1974):

K(2) = Ks (2/2s)2+(2+p)/b ; h $ hi (4)

K(h) = Ks (a/h)2+(2+p)/b ; h < hi (5)

where Ks [LT-1] is the value of K(2) at saturation (i.e. 2 = 2s) and p is an empirical pore interaction
parameter.

Equation (1) is solved using finite difference techniques (Hutson and Wagenet, 1989) to obtain
estimates of h at each depth interval (node) into which the soil profile has been subdivided. Water
contents are calculated at each node using Eqs. (2)-(5). Water flux densities between each node,
(q, LT-1) are calculated using:

q = K(2) ()H/)z) (6)

These values of 2 and q are then used in the simulation of solute (pesticide) transport in the LEACHP
submodel.

11-4
LEACHP (Pesticide Migration, Sorption and Degradation Submodel)

Solute (pesticide) migration through the soil profile is described in LEACHP using the convective
- dispersion equation, written as:

(MC/Mt) (DKd + 2 + ,KH) = (M/Mz) [2D(2,q) (MC/Mz) - qC] ± N (5)

where C is pesticide solution concentration [ML-3], D is soil bulk density [ML-3], Kd is the distribution
coefficient [L3 M-1], , is air filled porosity [L3 L-3], KH is Henry's Law constant for solute
volatilization [dimensionless], D(2,q) is the apparent diffusion coefficient [L2 T-1] and N represents
sources and sinks of solute [ML-3 T-1] such as dissolution and degradation. The apparent diffusion
coefficient is defined by:

D(2,q) = Dm (q) + Dp(2)/2 + Dog KH/2 (6)

where Dm(q) is the hydrodynamic dispersion coefficient (L2T-1) that describes mechanical mixing as
a result of local variations in mean pore water velocity, Dp(2) is the effective solute diffusion
coefficient in the liquid phase (L2T-1) and Dog is the solute diffusion coefficient in the gaseous phase
(L2 T-1).

Pesticide sorption reactions, which are assumed to be sufficiently rapid that local equilibrium exists
between dissolved and sorbed forms of the chemical, are described by:

S = Kd C (9)

where S is pesticide sorbed concentration (MM-1). Kd values are estimated from commonly reported
Koc values (see e.g. Jury et al., 1984) and the organic carbon content of the soil.

Degradation of pesticides is assumed to obey first-order kinetics, so that:

N = -k(2 + DKd + ,KH) C (10)

where k is a dissipation rate coefficient (T-1), dependent upon soil water content and soil temperature.
LEACHP can simultaneously simulate the transformation and fate of up to five chemical species.
The transformation products of a chemical species can form sources for other chemical species.

LEACHP must operate in concert with LEACHW, as it requires the water content and water flux
distributions at each time step in order to solve the convection - dispersion equation. In solving for
the pesticide distributions, LEACHP proceeds through a series of discrete time steps to estimate
C(z,t) at each node. Bypass of plant related subroutines is possible if no plants are present. Output
tables of cumulative totals and mass balances of water and chemicals include: the amount of
material initially in the soil profile; the amount currently in the soil profile; the simulated change,
additions and losses of material; and composite material balance errors. A summary by depth (node)
of predicted 2, h, q, C, and plant extraction of water and solute is also provided.

11-5
Input Requirements for LEACHW and LEACHP

The principal input data requirements for LEACHW and LEACHP include:

i) LEACHW
- soil profile characteristics (e.g. depth, layers)
- top and bottom boundary conditions (e. g. water table, free draining profile, etc.)
- soil hydraulic properties (e.g. 2(h) and K(2) functions)
- weather data (e.g. daily air temperature, precipitation and potential evapotranspiration)
- crop data (e.g. planting, emergence, maturity and harvest dates; canopy growth; rooting and
transpiration characteristics)

ii) LEACHP
- soil profile characteristics (e.g. dispersivity; organic carbon content; soil, air and water
thermal properties)
- solute (pesticide) chemical properties (e.g. solubility; vapor pressure; dissipation and
sorption kinetics; transformations rates)
- chemical (pesticide) applications (e.g. dates and amounts of chemical applied)

The LEACHM user's manual (Hutson and Wagenet 1989) should be consulted for a complete and
detailed listing of all inputs.

MODIFICATIONS MADE TO LEACHW AND LEACHP

The LEACHW and LEACHP submodels required several modifications before they could be applied
effectively within the pedotransfer/geostatistics/GIS framework. The most important of these
modifications are outlined briefly below.

Soil Hydraulic Properties

The soil hydraulic property functions used in LEACHW (Eqs. 2-5) poorly reflect the often observed
rapid decrease in water content and hydraulic conductivity at pore water pressure heads between 0
and -0.2 kPa (see e.g. Topp et al., 1980). The functional relationships between pressure head (h),
soil water content (2) and hydraulic conductivity (K) proposed by Van Genuchten (1980), on the
other hand, usually give a much more accurate representation of the observed behaviour.
Consequently the Van Genuchten functions were incorporated into LEACHW as an optional
alternative. The 0(h) relationship is given by:

2 s - 2r
2 = 2r + __________________ 2r # 2 # 2S (11)
1 + |"h| ]
n [1-(1/n)]

where a [L-1] and n [dimensionless] are empirical parameters, and 2S and 2r, refer to the saturated
and residual volumetric water contents, respectively [L3L-3]. The K(h) relationship has the form:

11-6
 [1 + |"h|n ][1-(1/n)] - |"h|(n-1)]2
K(h) = Ks _______________________ (12)
[1 + |"h|n ][1-(1/n)] (L+2)

where L = 0.5 is usually assumed.

Water Extraction by Plants

Preliminary runs with LEACHW indicated that in wet soil profiles, (i.e. pore water pressure heads
greater than -10 kPa), the Nimah and Hanks (1973) function for water extraction by plant roots,
U(z,t), would predict unrealistic (oscillating) water uptake patterns. An alternative function based
on the work by Feddes et al. (1978) was therefore added as an option. This function has the form:

U(z,t) = Rdf (z,t) $(h) Smax (13)

where Rdf is the root distribution function [dimensionless] (already calculated in the GROWTH
subroutine), $(h) is a dimensionless sink term variable, ranging from zero to 1 as a function of soil
water pressure head, and Smax represents the maximum possible rate at which plant roots can extract
water from the soil, and is given by:

Smax = PT / Zr (14)

where PT is the potential transpiration rate [LT-1] and Zr [L] is the rooting depth. Root water uptake
is adjusted to reflect non-optimal conditions using the $ function and specified pressure head limits,
h1, h2, h3 and h4 (see Feddes et al., 1978). In the Grand River watershed application, the h1 !h4
values of Dierckx et al. (1988) and Veenhof (1993) were used: h1 = -1 kPa, h2 = -2 kPa, h3min = -60
kPa, h3max = -40 kPa, h4 = -1500 kPa. The Rdf function was that of Tillotson et al. (1980), with the
provision that Zr increased linearly from 5 cm depth at planting to 90 cm depth at root maturity.

Dissipation Rates

The dissipation rate constant, k, of most pesticides (including atrazine) is known to exhibit a
substantial dependence on soil water content and temperature. Empirical water content and
temperature corrections for k were therefore added to LEACHP. The water content correction has
the form (Walker, 1978):

kW = 1n2 / A(100 2)-B (15)

where kw is the water content corrected dissipation rate constant [T-1], and A and B are empirical
constants. The temperature correction, Tcf [dimensionless], is given by:

Tcf = Q10 0.1(t-tbase) ; 0ºC < t < 35 ºC (16)

where Q10 [dimensionless] is the factor by which k changes over a 10 ºC temperature interval, t [ºC]
is soil temperature, and tbase [ºC] is the base temperature from which the temperature correction
was

11-7
determined. The water content - temperature corrected dissipation rate constant, kwt, is then
calculated as:

kwt = kw Tcf (17)

For the Grand River watershed application, practical constraints dictated that Eqs. (15) and (16)
could be calibrated to atrazine only for the 3 predominant soil types in the watershed (Table 111).
Consequently, every soil in the watershed was assigned one of the 3 calibrated kwt functions on the
basis of which predominant soil type they were the most similar to.

Table 11-1. Dissipation rate constants, k*, Q10 values and constants A and B (Eqs. 15 and 16) for
atrazine. (Data provided by E. Topp and W.N. Smith, CLBRR, Agriculture and
Agri-Food Canada, Ottawa, ON).

Soil Type
Variable Woodstock Dalhousie Alliston
Clay Clay loam Sandy Loam
k* (d-1) 0.0468 0.0217 0.0215
Q10 4.896 3.715 3.687
A (d) 686.9 122.9 198.4
B 1.061 0.369 0.514
k* is a "reference level" dissipation rate constant (determined at tbase = 25 ºC and 2 = 0.70 2 s)
which is required for the calculation of Q10, A and B.

Bypass Flow

It is well established that preferential, or bypass, flow in macropores (cracks, worm holes, root
channels, etc.) and "finger zones" can cause rapid movement of water and solutes through the soil
profile and into the ground water (e.g. Thomas and Phillips, 1979; Hendrickx and Dekker, 1991).
As a result, many attempts have been made to incorporate bypass flow into the traditional
mechanistic water and solute transport models, which neglect this phenomenon. However, all of
these representations have serious limitations (Beven, 1991), and consequently a simplistic, but
measurement based, approach to bypass flow was adopted for this work.

Chloride tracer breakthrough curves measured in situ in the soil profile often show the 0.5Co
concentration (C0 = chloride concentration in input spike of tracer) occurring before one pore volume
(PV) of soil water is eluted. This is indicative that some of the water in the soil profile was bypassed
by the chloride solute (White, 1985). The proportion of soil water bypassed by the solute was
estimated here using:

F = 1-(2T /2) = 1-PV0.5 ; 2T # 2; 0 # F # 1 (18)

where F is a "bypass flow" factor [dimensionless], 2T is the amount of soil water that participated
in solute transport [L3L-3], and PV0.5 is the number of pore volumes at which the 0.5C0 tracer
concentration occurred. This factor was incorporated into the advection and mechanical mixing
terms

11-8
of the convection-dispersion equation, i.e.:

qC ! qC/(1-F) (19)
Dm(q) = 8q / 2 ! 8q/ 2 (1-F) (20)

where q [LT-1] is water flux density, C [ML-1] is solute solution concentration, Dm [L2T-1] is the
mechanical mixing component of the hydrodynamic dispersion coefficient (Eq. 8) and 8 [L] is the
soil dispersivity. The above implies that all of the soil water participates in solute sorption,
dissipation and diffusion, but only a fraction of the water (specified by F) participates in advection
and mechanical mixing. The effect of F on the convection-dispersion equation is to increase the
average velocity with which solute migrates through the soil profile relative to the average linear
pore water velocity (q/2). This in turn causes the predicted 0.5Co concentration of a nonreactive
solute (e.g. chloride) to occur at less than one PV (i.e. PV0.05 # 1).

Bypass flow factors (F) were determined from field measured chloride breakthrough curves collected
in Southern Ontario for several soil types, water contents and depths in the soil profile. The F values
were highly variable (CV = 168 %) and did not show any consistent patterns with soil texture, water
content, pore water flux, or depth in the soil profile. Consequently, the mean value of F = 0.20 (n =
56) was used to represent bypass flow in the Grand River watershed application.

Multiple Year Simulations

LEACHW and LEACHP were modified to run over a number of consecutive years to determine if
the annual pesticide mass loading at a specified depth becomes constant with time, continuously
increases or decreases with time, or changes erratically from year to year. The modifications
included development of multiple year weather and crop management input files and various
program changes to account for over-winter redistribution of water in the soil profile, snow
accumulation during winter, and spring runoff.

It was assumed that during the "winter" (defined as when the weekly mean air temperature is less
than 0 ºC), the soil surface is frozen and there is no water flux across the atmosphere/soil interface.
The result of this is that the existing water and solutes in the profile slowly redistribute during the
winter via gravity drainage. Precipitation during the winter period is assumed to occur as snow which
accumulates with the assumption that 30% is lost due to blowoff, evaporation and sublimation. The
remaining 70% is distributed between infiltration and snowmelt runoff during the first seven days
of "spring" (defined as when the weekly mean soil surface temperature is greater than 0 ºC). Runoff
is calculated with the USDA Soil Conservation Service (1972) curve number method. Soil surface
temperatures were assumed to equal air temperatures.

Crop Management

The crop management component of LEACHP was modified to take into account crop type, soil
permeability and air temperature when determining the dates for pesticide application, planting,
emergence, maturity and harvest. For the Grand River watershed application, corn (Zea mays L.)
was planted in soils with intermediate surface permeability (100 mm/d #Ks # 250 mm/d) when the
mean air temperature reached 12.8ºC (Brown, 1976). Relative to this date, planting was advanced

11-9
seven days on soils with high surface permeability (Ks > 250 mm/d), and delayed seven days on soils
with low surface permeability (Ks < 100 mm/d). Emergence was assumed to take place seven days
after planting, and atrazine was applied two weeks after emergence. Crop maturity (full crop cover
and maximum root depth) occurred when 1250 corn heat units had been accumulated since planting.
The crop was harvested in the fall when either the mean air temperature dropped below 12ºC, or the
minimum air temperature dropped below -2ºC.

TESTING AND EVALUATION OF LEACHW AND LEACHP

The modified LEACHW and LEACHP models were tested and evaluated for their ability to
simulate water and solute transport with acceptable accuracy. Some example results are given below.
A complete description of model testing and evaluation is given in Reynolds et al., 1994.

Laboratory Column Study

The modified LEACHW and LEACHP models were tested for their ability to simulate water content
profiles (LEACHW) and chloride and atrazine breakthrough curves (LEACHP) obtained from
steady, saturated and unsaturated flow through two 20 cm diameter by 65 cm long intact columns
of structured clay loam soil. The overall regression between predicted (Y-axis) and measured
(X-axis) water content (expressed as volume percent) gave a slope of 1.11, a Y-axis intercept of
-5.19 %, and an R2 value of 0.93. The overall regression between predicted (Y-axis) and measured
chloride effluent concentration (in mg/L) gave a slope of 1.19, a Y-axis intercept of 1.23 mg/L, and
an R2 value of 0.98. The regression for atrazine effluent concentration (in µg/L) gave a slope of 0.90,
an intercept of 2.40 µg/L, and an R2 of 0.92. These results are within acceptable limits of accuracy,
considering the observed heterogeneities within the soil column and the many sources of
measurement and prediction error.

Field Studies

LEACHW and two other soil water flow models (SWATRE and SWASIM) were tested and
evaluated against field data from a structureless (single grain) Caledon sandy loam, cropped to
soybeans (near Simcoe, ON.) and a structured (medium subangular blocky) Rideau clay, cropped
to grass (near Ottawa, ON.). All three models were found to perform well within the limits of
acceptance, as evidenced by an average prediction error within ± 0.04 cm3 cm-3 and a relative
prediction error within ± 29 % (Clemente et al., 1994). Discrepancies between the measured and
modelled soil water contents were attributed to the fact that none of the models accounted for
hysteresis and flow in macropores. Differences between the models were found to be due primarily
to the different approaches used for describing evapotranspiration.

The LEACHW and LEACHP models were further tested for their ability to simulate field-measured
water content profiles and chloride breakthrough curves obtained from constant flux rainfall
simulation experiments conducted at six field sites distributed across Southern Ontario, which ranged
in texture from sandy loam to clay. For a clay loam soil near Ottawa the regression between
predicted (Y-axis) and measured (X-axis) water content (expressed as volume percent) between 25
and 100 cm depth gave a slope of 0.96, a Y-axis intercept of 0.015 %, and an R2 value of 0.94. The

11-10
corresponding regression between predicted (Y-axis) and measured (X-axis) chloride concentration
(mg/L) at the 25 cm depth gave a slope of 0.91, a Y-axis intercept of 44.5 mg/l, and an R2 of 0.97.
These results are typical of what was obtained for all six of the field sites, and are considered
acceptable.

It was concluded from the column and field tests that the modified LEACHW and LEACHP models
can adequately simulate water, chloride and atrazine transport both in the laboratory and in the field.
It was also noted that both models usually satisfied the criterion for model acceptance set by the
Prediction Exposure Assessment Workshop (Hedden, 1986), which states that a model should be
able to replicate field data within a factor of 2 for site-specific applications.

APPLICATION OF THE MODIFIED LEACHW AND LEACHP MODELS WITHIN

THE PEDOTRANSFER FUNCTION/GEOSTATISTICS/GIS FRAMEWORK

The LEACHW and LEACHP models simulate water and solute movement in the vertical direction
only, i.e. they are one-dimensional. Extension of the models to an areal basis (e.g. Grand River
watershed) requires running them at a number of georeferenced locations distributed throughout the
area, and then applying interpolation procedures that account for the inherent spatial variability
within the area. This is accomplished using archived soil survey and weather databases, pedotransfer
functions, geostatistical analyses, and a GIS.

The area used for the Grand River watershed application actually consisted of a large (. 3 million
ha), arbitrarily defined, "map window" encompassing the watershed, rather than just the watershed
alone (Fig. 11-2). This was done to increase the amount of data available, and thereby precision, of
the geostatistical calculations; and to eliminate inaccuracies in the geostatistical and GIS calculations
along the watershed boundary due to border effects.

Soil Survey and Weather Data

The input requirements of the LEACHW model include (among other things) soil physical
properties, soil hydraulic properties, and weather data. When the model is being applied to large
areas, such as watersheds, the main sources of these data (in Canada) are the National Soil Data Base
(NSDB) and the Archived Weather Data Base (AWDB).

The soil input data required for the Grand River watershed application was identified and extracted
from the NSDB on the basis of the dominant soil type in the 1:1 million scale soil landscape
polygons (Soil Landscapes of Canada, Shields et al., 1991) that fell within the map window. The data
for each polygon were extracted for 3 soil layers (approximately A, B and C horizons), and
included the upper and lower depths of the layer; sand, silt and clay contents; bulk density; organic
carbon content; saturated hydraulic conductivity; and 2-4 points on the soil water characteristic. The
data were assigned to the georeferenced centroids of the landscape polygons. A total of 119
landscape polygon centroids fell within the map window, 18 of these falling within the Grand River
watershed (Fig. 11-2).

11-11
Fig. 11-2. Map window encompassing the Grand River watershed with
locations of soil landscape polygon centroids.

11-12
The climate data for the map window were extracted from the AWDB, and included monthly 30
year normals (1950-1980) of maximum and minimum air temperatures, precipitation, and days
with precipitation. These data were assigned to each of the landscape polygon centroids within the
map window, using the values from the nearest weather station. The monthly normals were
converted to daily data, required by LEACHW, using the procedures described by Brooks (1943)
and Van Diepen et al. (1988). Potential evapotranspiration (also required by LEACHW) was
calculated according to Baier and Robertson (1965).

Pedotransfer Functions

For the Grand River watershed application missing data from the NSDB included: bulk density
(61 % missing), the soil water characteristic (61 % missing), saturated hydraulic conductivity (91
% missing), and the unsaturated hydraulic conductivity function (100 % missing). These were
estimated using pedotransfer functions based on soil texture and organic carbon (OC) content, for
which there were no missing values in the NSDB. Bulk density (BD) was estimated using the Gupta
and Larson (1979) model. The soil water characteristic, 2(h), was estimated using the model of
McBride and Macintosh (1984), and then least squares fitted to the Van Genuchten (1980) function
(Eq. 11). Saturated hydraulic conductivity, Ks, was estimated using the model of Jabro (1992). The
unsaturated hydraulic conductivity relationship, K(h), was estimated from Eq. 12 using the 2(h) and
Ks results. The estimated soil physical and hydraulic property data, as with the available data, were
assigned to the georeferenced landscape polygon centroids.

Simulations and Output

The LEACHW - LEACHP simulations were conducted for all 119 soil landscape polygon centroids
in the map window (Fig. 11-2), using the appropriate soil and weather input data at each centroid.
The simulations were run for 10 consecutive "simulation" years, assuming an initially atrazine - free
soil profile and repeating the 30 year normal weather each year. Corn was grown every year over
the entire map window using the crop management scheme described above. Atrazine was applied
each year at the recommended rate of 150 mg/m2 (Ontario Ministry of Agriculture and Food, 1993),
3 weeks after planting. A representative mean soil dispersivity (8 = 15.5 cm, obtained from the field
tests of LEACHP) and atrazine partition coefficient (Koc = 160 ml/g, Jury et al., 1984) was assumed
for all soil types and depths in the soil profile. Atrazine dissipation rate constants were determined
using Table 11-1 and the procedures described above. A constant water table depth of 120 cm was
assumed because of inadequate water table data in the NSDB. The predicted annual mass loading
of atrazine at the 90 cm (average tile drain) depth at the end of the 10-year simulation was used as
an estimate of ground water contamination.

Geostatistical Analyses

A geostatistical technique known as kriging was used to account for spatial variability when
extending the soil and weather input data and model predictions from a point basis to an areal basis
(e.g. watershed). Kriging is essentially a weighted moving-average technique for interpolating
between known data values at georeferenced locations, with the weighting factors determined from

11-13
a semivariogram. The latter characterizes the spatial variability of the known values by defining
both the maximum distance over which the values are related to each other (i.e. the range of the
semivariogram), and the functional nature of this relationship (i.e. the shape of the semivariogram).

Kriging is used within the modelling-geostatistics-GIS framework to convert relatively small

numbers of irregularly spaced and highly variable point values of soil properties, weather attributes
and predicted agrochemical loadings, into large numbers of interpolated values that extend
throughout the area of interest on a regular, fine-mesh grid. The interpolations provide the required
extension from a point basis to an areal basis, and also retain the spatial variability characteristics
of the original data sets. The interpolations also provide the spatial detail necessary for effective use
of a GIS.

For the Grand River watershed application, semivariograms of soil properties (texture, BD, OC, 2s,
2r, Ks, ", n), precipitation (spring, summer, fall, winter) and predicted atrazine loadings (mg
atrazine/m2/yr) were calculated using the 119 locations in the map window. The data were then
kriged (interpolated) on a 2 km x 2 km grid to produce a total of 7381 georeferenced grid points
(1657 within the watershed) containing estimates of soil hydraulic properties, precipitation and
predicted atrazine loadings. These kriged data formed the input to the GIS.

Geographic Information System (GIS)

The main function of the GIS was to produce, quantify and overlay maps of the kriged soil, weather
and pesticide loading data. This allowed estimation of the importance and distribution of atrazine
contamination of ground water, as well as determination of the soil, weather and land management
factors that control the contamination.

The GIS chosen for this study was the Integrated Land and Water Information System (ILWIS,
version 1.3), developed at the International Institute for Aerospace Survey and Earth Sciences
(ITC), Enschede, The Netherlands. ILWIS was developed specifically, as suggested by its name,
to handle land and water information.

RESULTS AND DISCUSSION OF THE

GRAND RIVER WATERSHED APPLICATION

Soil texture (sand, silt, clay content), bulk density, and 2s, exhibited moderate to high variability
across the watershed (CV = 8.1 to 85.1 %), but only modest changes with depth (Table 11-2). The
high lateral variability in texture is also reflected in the surface texture map for the watershed (Fig.
11-3), which shows a wide range in soil types (sand to silty clay), as well as very complex spatial
distributions. The soil texture semivariograms for the 3 soil horizons are similar, with a small
nugget (20 - 40 % of the variance) and a correlation distance (range) of about 60 km. The similarity
between the semivariograms, coupled with the extreme and intricate lateral variability in soil
texture, probably reflects the complex glacial origin of most soils in this watershed.

The OC and Ks values are moderately to extremely variable across the watershed (36.0 % # CV
# 64.0 % for OC; 79.7 % # CV # 156.7 % for Ks) at any particular depth, and also decrease

11-14
Fig. 11-3. Soil surface texture map of the Grand River watershed

11-15
substantially in mean value with increasing depth (Table 11-2). The Ks distributions for soil layers
1 and 2 also have very large positive skewness and kurtosis values (2.537 # skewness # 3.217; 9.22
# kurtosis # 12.76), indicating that many low Ks values exist close to the mean value, many large
Ks values exist far above the mean value, and relatively few Ks values fall in between. The decrease
in mean OC and Ks with increasing depth is not surprising because of the usual decrease in
biological activity and soil structure with depth.

The atrazine loadings across the watershed are highly variable (CV = 136.9 %) and form a statistical
distribution that is positively skewed (skewness = +0.956) and flat (kurtosis = 2.30), which
indicates that many high loading values exist far above the mean value. The loading distribution
also appears to be multimodal, as several peaks occur in the loading histogram and many of the
histogram classes contain no values. In contrast to this extreme and complex spatial variability, is
the temporal (year to year) variability, which declines to zero as the predicted annual atrazine
loadings become constant at any particular location after about 5-8 simulation years (example given
in Fig. 11-4). Evidently, the spatially and/or temporally distributed weather and soil attributes
interact in such a way as to enhance the spatial variability, but eliminate the annual variability, of
atrazine loading at the 90 cm depth.

It should also be noted from the example soils in Fig. 11-4 that the rate and path by which atrazine
loading stabilizes, as well as the final loading value, appear to be determined by complex
interactions among weather, soil properties and solute transport mechanisms. The predicted atrazine
loadings all start at zero, reflecting the fact that initially atrazine-free soil profiles were assumed.
For a few soils (e.g. Haldimand clay), the loadings stay at zero for the entire 10-year simulation,
which implies that the atrazine applied to these soils is either degraded entirely or sufficiently
retarded in its movement that it does not reach the 90 cm depth after 10 simulation years. It is
assumed in this work that these soils will never contribute significant quantities of atrazine to the
ground water. The majority of soils, however (e.g. Fox sand, Guelph loam, Huron clay loam; Fig.
11-4), contribute increasing quantities of atrazine with time until a plateau is reached after about
5-8 simulation years. These constant final loadings are somewhat soil dependent, increasing with
coarser textures, but exceptions are frequent. For example, Fig. 11-4 shows that the annual atrazine
loading is initially greater in the Fox sand (years 2 - 4) than in the Guelph loam, which is consistent
with the much higher sand content of the Fox soil (. 80 % sand in Fox sand; . 35 % sand in Guelph
loam). After 5 years, however, the trend reverses, and the Guelph loam contributes a slightly
greater annual atrazine loading than the Fox sand (by about 14 %).

An ILWIS - generated map of kriged atrazine loadings throughout the Grand River watershed is
given in Fig. 11-5. It confirms, both the high variability and the complex spatial distribution of the
loadings indicated in Table 11-2. Correlation analysis shows that atrazine loading is significantly
correlated with many soil and weather attributes. In particular, low loadings tend to coincide with
clayey soils and low summer precipitation, while intermediate to high loadings tend to coincide
with sandy to loamy soils and moderate to high summer precipitation. The magnitudes of the
correlations are generally low, however. This supports the indication in Fig. 11-4 that atrazine
loading tends to be determined by complex interactions among several soil, weather, crop
management and solute transport factors, rather than by one or two dominant factors.

11-16
Fig. 11-4. Predicted annual atrazine loading (mg/m2) versus time (yr) at 90
cm depth for Fox sand, Guelph loam, Huron clay loam and
Haldimand clay.

11-17
Fig. 11-5. Predicted annual atrazine loading (mg/m2/yr) at the 90 cm depth
for the Grand River watershed.

11-18
Table 11-2. Basic statistics for the Grand River watershed application, based on the 18 landscape
polygon centroids within the watershed. Thick = layer thickness; rest of parameters
defined in text.

Parameter Unit Mean CV Min. Val. Max. Val. Skew. Kurt.

LAYER 1
Thick cm 14.4 33.3 5.0 25.0 0.204 2.60
Sand % 32.9 68.9 11.0 75.0 0.901 2.13
Silt % 43.8 32.9 17.0 64.0 -0.682 2.13
Clay % 23.2 54.1 8.0 45.0 0.345 1.75
BD g/cc 1.34 11.8 1.0 1.57 -0.433 2.11
OC % 1.83 36.0 0.5 3.10 -0.218 2.18
2s % 49.5 13.9 40.0 63.0 0.227 1.80
Ks cm/s 5.8E-4 156.7 2.7E-5 3.8E-3 2.537 9.22
LAYER 2
Thick cm 16.7 48.3 5.0 30.0 0.525 1.79
Sand % 34.5 72.5 4.0 80.0 0.693 2.07
Silt % 38.8 34.7 15.0 62.0 -0.361 2.16
Clay % 26.7 65.0 5.0 61.0 0.422 1.98
BD g/cc 1.45 9.0 1.25 1.70 0.311 1.98
OC % 1.21 47.4 0.17 1.91 -0.306 1.63
2s % 45.7 11.2 35.9 56.2 -0.116 2.50
Ks cm/s 1.2E-4 153.4 1.2E-7 8.2E-4 3.217 12.76

LAYER 3
Thick cm 88.3 11.2 70.0 100.0 -0.213 1.67
Sand % 33.9 85.1 3.0 87.0 0.766 2.00
Silt % 32.1 43.9 9.0 64.0 0.206 2.61
Clay % 34.1 65.1 4.0 64.0 0.015 1.33
BD g/cc 1.50 8.1 1.30 1.71 -0.143 2.31
OC % 0.62 64.0 0.10 1.72 0.977 3.90
2s % 42.1 11.1 35.5 53.0 0.779 2.92
Ks cm/s 6.5E-5 79.7 1.2E-7 1.6E-4 0.418 1.95
Atrazine
Load mg/m2/yr 0.50 136.9 0.0 1.88 0.956 2.30

The predicted atrazine loading to the ground water in the Grand River watershed, based on the
ILWIS compilation of the 1657 kriged loading values, ranges from 0 to 2.5 mg/m2/yr (Fig. 11-5)
with a mean value of 0.67 mg/m2/yr. The maximum and mean loadings given here are somewhat
higher than those in Table 11-2, because the kriging interpolations take into account the LEACHP
-simulated loadings at all 119 polygon centroids in the map window (Fig. 11-2), several of which
are considerably higher than the loadings for the 18 centroids within the watershed. (The maximum
LEACHP-simulated loading in the map window was 8.29 mg/m2/yr.). Both the mean and maximum
predicted atrazine loadings for the watershed (i.e. 0.67 mg/m2/yr and 2.50 mg/m2/yr, respectively)
are quite low relative to the application rate of 150 mg/m2/yr, suggesting that atrazine sorption and
dissipation are extensive within the soil profile.

11-19
The concentration of atrazine in the soil water at the 90 cm depth was also predicted to be generally
low throughout the watershed. The former 60 ppb Canadian drinking water guideline for atrazine
(Canadian Water Quality Guidelines, 1989) was never exceeded at the 90 cm depth during the
10-year simulation. The 3 ppb USEPA standard (USEPA, 1987) was exceeded, however, on or
before the 10th simulation year in about 27% of the watershed area (Fig. 11-6). The areas where this
occurs also have predicted annual atrazine loadings that fall within the top half of the loading range
(0.5 - 2.5 mg/m2/yr, Fig. 11-5), which suggests that atrazine concentration and loading rates are
related, but this relationship is not direct or simple. It also suggests that the areas where the 3 ppb
concentration is exceeded (Fig. 11-6) represent regions of potentially significant low-level
non-point source contamination of ground water by the downward percolation of atrazine through
the soil profile. Figure 11-6 thus demarks regions in the watershed where more detailed
investigation may be warranted.

The Grand River watershed predictions are consistent with the results of a recent survey of ground
water quality in Southern Ontario (Agriculture Canada, 1992), which included over 900 farm water
supply wells (approximately 100 fell within the Grand River watershed) and 144 monitoring wells
(approximately 11 fell within the watershed). The survey concluded (as did the predictions) that
non-point atrazine contamination of ground water was infrequent, highly variable spatially, and
low-level; that the former Canadian drinking water guideline for atrazine (60 ppb) was never
exceeded; that the USEPA drinking water standard (3 ppb) was exceeded occasionally; and that
the contamination was not strongly related to soil type or land use, implying control by many
interacting soil, weather and land use factors rather than one or two dominant factors. This lends
credibility to the LEACHM-Kriging-ILWIS methodology, notwithstanding that more extensive
comparisons with field data are required before definite conclusions can be drawn.

CONCLUDING REMARKS

Although the LEACHM-Kriging-ILWIS methodology still requires further development and

testing, the preliminary results are very encouraging. The modified LEACHW and LEACHP
models appear capable of simulating both laboratory- and field-measured transport of water,
chloride and atrazine with acceptable accuracy. Predictions of potential ground water contamination
by atrazine for the Grand River watershed seem plausible and compare favourably with recent
ground water survey results. It is consequently felt that this methodology will ultimately prove very
useful in the development of agricultural practices and guidelines that maintain agrochemical
inputs to the groundwater at acceptable and sustainable levels.

11-20
Fig. 11-6. Predicted time (yrs) for atrazine to reach 3 ppb at the 90 cm
depth in the Grand River watershed.

11-21
 REFERENCES

Agriculture Canada. 1990. Report to Ministers of Agriculture. Federal-Provincial Agriculture

Committee on Environmental Sustainability, Communications Branch, Agriculture Canada,
Ottawa, ON.

Agriculture Canada. 1992. Ontario Farm Groundwater Quality Survey, Winter 1991/92.
Federal-Provincial Environmental Sustainability Initiative, Research Branch, Agriculture
Canada, ON.

Baier, W. and Robertson, G.W. 1965. Estimation of latent evaporation from simple weather
observations. Can. J. Plant Sci. 45:276-284.

Beven, K.J. 1991. Modeling preferential flow: an uncertain future? Pages 1-11 in T.J. Gish and
A. Shirmohammadi, Eds. Preferential Flow: Proceedings of the National Symposium. American
Society of Agricultural Engineers, St. Joseph, MI.

Bresler, E. 1973. Simultaneous transport of solutes and water under transient unsaturated flow
conditions. Water Resour. Res. 9:975-986.

Brooks, C.E.P. 1943. Interpolation tables for daily values of meteorological elements. Q. J. R.
Meteorol. Soc. 69:160-162.

Brown, D.M. 1976. Heat units for corn in southern Ontario. Ontario Ministry of Agriculture and
Food, Guelph, ON.

Campbell, G. 1974. A simple method for determining unsaturated conductivity from moisture
retention data. Soil Sci. 117:311-314.

Canadian Water Quality Guidelines. 1989. Water Quality Branch, Environment Canada, Ottawa,
ON.

Clapp, R.B. and Hornberger, G.M. 1978. Empirical equations for some hydraulic properties.
Water Resour. Res. 14:601-604.

Clemente, R.S., De Jong, R., Hayhoe, H.N., Reynolds, W.D. and Hares, M. 1994. Testing and
comparison of three unsaturated soil water flow models. Agric. Water Manage. (in press).

Dierckx, J., Gilley, J.R., Feyen, J. and Belmans, C. 1988. Simulation of soil water dynamics and
corn yields under deficit irrigation. Irrig. Sci. 9:105-125.

Feddes, R.A., Kowalik, P.J. and Zaradny, H. 1978. Simulation of field water use and crop yield.
Centre for Agric. Publ. and Docum., Wageningen, The Netherlands.

Gupta, S.C. and Larson, W.E. 1979. A model for predicting packing density of soils using particle
size distribution. Soil Sci. Soc. Am. J. 43:758-764.

11-22
Hedden, K.F. 1986. Example field testing of the soil fate and transport model, PRZM, Dougherty
Plain, Georgia. Pages 81-101 in S.C. Hern and S.M. Melancon, Eds. Vadose zone modeling of
organic pollutants. Lewis Publishers, Chelsea, MI.

Hendrickx, J.M.H. and Dekker, L.W. 1991. Experimental evidence of unstable wetting fronts in
homogeneous non-layered soils. Pages 22-31 in T.J. Gish and A. Shirmohammadi, Eds.
Preferential Flow: Proceedings of the National Symposium. American Society of Agricultural
Engineers, St. Joseph, MI.

Hutson, J.L. and Cass, A. 1987. A retentivity function for use in soil-water simulation models. J.
Soil Sci. 38:105-113.

Hutson, J.L. and Wagenet, R.J. 1989. LEACHM, Leaching Estimation And Chemistry Model.
Version 2. Center for Environmental Research, Cornell Univ., Ithaca, NY.

Jabro, J.D. 1992. Estimation of saturated hydraulic conductivity of soils from particle size
distribution and bulk density data. American Society of Agricultural Engineers (ASAE).
35:557-560.

Jury, W.A., Spencer, W.A. and Farmer, J. 1984. Behaviour assessment model for trace organics
in soil: III. Application of screening model. J. Environ. Qual. 13:573-579.

McBride, R.A. and Macintosh, E.E. 1984. Soil survey interpretations from water retention data:
I. Development and validation of a water retention model. Soil Sci. Soc. Am. J. 48:1338-1343.

Millette, J.A. and Torreiter, M. 1992. Nonpoint source contamination of groundwater in the Great
Lakes Basin: a review. Centre for Land and Biological Resources Research, Research Branch,
Agriculture Canada, Ottawa, ON.

Nimah, M.N. and Hanks, R.J. 1973. Model for estimating soil water, plant and atmospheric
interrelations. I. Description and sensitivity. Soil Sci. Soc. Am. Proc. 37:522-527.

Ontario Ministry of Agriculture and Food. 1993. Guide to weed control. Publication No. 75.
Queen's Printer for Ontario, Toronto, ON.

Reynolds, W.D., De Jong, R., Vieira, S.R. and Clemente, R.S. 1994. Methodology for predicting
agrochemical contamination of ground water resources. Soil Quality Evaluation Program,
Technical Report, Centre for Land and Biological Resources Research, Research Branch,
Agriculture and Agri-Food Canada, Ottawa, ON.

Shields, J.A., Tarnocai, C., Valentine, K.W.G. and MacDonald, K.B. 1991. Soil landscapes of
Canada: Procedures manual and user's handbook. Land Resource Research Centre, Research
Branch, Agriculture and Agri-Food Canada, Ottawa, ON.

Thomas, G.W. and Phillips, R.E. 1979. Consequences of water movement in macropores. J.
Environ. Qual. 8:149-152.

11-23
Tillotson, W.R., Robinson, R.J., Wagenet, R.J. and Hanks, R.J. 1980. Soil water, solute and
plant growth simulation. Utah Agric. Exp. Stn. Bull. No. 502. Utah State Univ., Logan, UT.

Topp, G.C., Zebchuk, W.D. and Dumanski, J. 1980. The variation of in situ measured soil water
properties within soil map units. Can. J. Soil Sci. 60:497-509.

USDA Soil Conservation Service, 1972. National Engineering Handbook, Section 4, Hydrology,
Washington, D.C., USA.

USEPA. 1987. Atrazine health advisory. U.S. Environmental Protection Agency, Office of
Drinking Water. Washington, D.C., USA.

Van Diepen, C.A., Rappoldt, C., Wolf, J. and Van Keulen, H. 1988. CWFS Crop growth
simulation model WOFOST. Documentation version 4.1. Centre for World Food Studies,
Wageningen, The Netherlands.

Van Genuchten, M.Th. 1980. A closed form equation for predicting the hydraulic conductivity
of unsaturated soils. Soil Sci. Soc. Am. J. 44:892-898.

Veenhof, D.W. 1993. The compression characteristics and susceptibility to compaction of soils
from the Regional Municipality of Haldimand-Norfolk. Unpublished M.Sc. thesis, Univ. of
Guelph, Guelph, ON.

Walker, A. 1978. Simulation of the persistence of eight soil-applied herbicides. Weed Res.
18:305-315.

White, R.E. 1985. The influence of macropores on the transport of dissolved and suspended matter
through soil. Pages 95-.120 in B.A. Stewart, Ed. Advances in Soil Science, Vol.3,
Springer-Verlag, New York, NY.

11-24
CHAPTER 12
INDUSTRIAL ORGANIC COMPOUNDS IN SELECTED CANADIAN
MUNICIPAL SLUDGES AND AGRICULTURAL SOILS

M.D. Webber

Wastewater Technology Centre,

Operated by Rockcliffe Research Management
P.O. Box 5068, Burlington, Ontario L7R 4L7

INTRODUCTION

There is growing concern about degradation of the environment by mankind. Degradation has been
documented internationally and there are growing "Green" movements in many countries.
Maintaining the quality of land resources for future generations is a major concern of these
movements.

Soil degradation in Canada is well documented and conservation of our soil and water resources
is a recognized need. Indeed, the focus of Agriculture and Agri-Food Canada's research effort
shifted recently from increasing crop production to development of sustainable agricultural systems.
Consistent with this shift "A Soil Quality Evaluation Program (SQEP) for Assessing Agricultural
Sustainability in Canada" was undertaken.

The SQEP objectives included developing procedures for predicting the fate of inorganic and
organic additions to soils. The Wastewater Technology Centre (WTC) has recognized heavy metal
and industrial organic expertise and was invited to participate.

Literature review indicated a considerable body of heavy metal information, particularly in relation
to land application of municipal sludge (Black et al., 1984). There was limited information
concerning industrial organics in municipal sludge (Webber and Lesage, 1989) but little information
concerning their fate in soil. It was decided that the WTC studies would focus on industrial organics
in Canadian agricultural soils. A report of these studies follows.

OBJECTIVES

The WTC undertook to determine; (1) the concentrations of selected industrial organic contaminants
in Canadian agricultural soils and (2) the potential for organic contamination of soils from
municipal sludge application to agricultural lands.

12-1
 EXPERIMENTAL

Thirty soil samples supplied by Dr. C. Wang, Centre for Land and Biological Resources Research,
Agriculture and Agri-Food Canada, Ottawa were analyzed for industrial organic compounds.
Twenty-four of the samples were obtained from eight SQEP national benchmark sites each located
in a different province. The majority of the samples were taken from the plough layer but seven were
subsoils. These benchmark soils were thought to represent typical Canadian agricultural soils. The
other six samples were plough layer soils obtained from intensively cropped southern Ontario farms
where there was repeated heavy use of pesticides for many years.

Hamilton and Sarnia municipal sludges were monitored for industrial organic contaminants during
a 5-month period from September 1992 to January 1993. These sludges were thought to represent
the worst cases of organic chemical contamination in Ontario. Hamilton is a city with considerable
heavy industry, including the Stelco and Dofasco foundries, while Sarnia has a concentration of
petrochemical industries.

Hamilton and Sarnia sludges were applied to sandy loam and silt loam soils at rates of 0, 8, 32 and
200 t ha' dw. The sludge treated soils were incubated aerobically at 20ºC and approximately field
water capacity for eleven weeks prior to analysis.

Soil samples (0-15 cm) were taken during November, 1992 from eight sludge treated southern
Ontario fields. Each sample was a composite of twenty-five, 2-cm diameter cores taken while
walking a "W" pattern over the field. One field had received a single application and six had
received multiple applications of sludge according to recommended practise. The eighth field was
a future landfill area owned by a regional municipality and it had received repeated surface
applications of sludge every year from 1982 to 1989 inclusively.

The Canadian agricultural soils were analyzed for base-neutral and acid extractable organic
compounds, total polychlorinated biphenyls, organochlorine pesticides, organophosphorous
pesticides and neutral, phenoxy acid and carbamate herbicides. The sludges and all other soils
included in these studies were analyzed for polynuclear aromatic hydrocarbons, total polychlorinated
biphenyls and organochlorine pesticides.

RESULTS AND CONCLUSIONS

Canadian Agricultural Soils

Results for the twenty-four soils from the Agriculture and Agri-Food Canada, Soil Quality
Evaluation Program - benchmark sites and the six intensively cropped soils from southern Ontario
indicated that there was no significant contamination with base-neutral and acid extractable
organic compounds (BN&As). Despite excellent analytical sensitivity these compounds were
seldom detected in the soils. The method detection limits (MDLs) were usually <0.1 mg kg-1 dw and
in many cases were <0.05 mg kg-1 dw. There was no detection, or only trace amounts of polynuclear
aromatic hydrocarbons (PAHs) - except naphthalene, chlorinated benzenes, heterocyclic nitrogenous
compounds (HNCs), nitrosamines, haloethers, a variety of "other compounds" and acids - except
phenol and pentachlorophenol. Naphthalene exceeded the 0.05 mg kg-1 dw MDL in thirteen soils,

12-2
one value was 1.2 mg kg-1 dw but all others were <0.21 mg kg-1 dw. Phthalate ester concentrations
frequently exceeded the MDLs but only three values for di-n-butylphthalate and two values for
bis-2-ethylhexylphthalate exceeded 1 mg kg-1 dw. The largest value was 3.1 mg kg-1 dw of
di-n-butylphthalate. Phenol concentrations in twenty-one soils exceeded the 0.02 mg kg-1 dw MDL
and in five soils exceeded 0.15 mg kg-1 dw but none exceeded 1 mg kg-1 dw. Pentachlorophenol was
reported in eight soils but did not exceed 0.36 mg kg-1 dw.

It is unlikely that the BN&As observed in these soils represent significant agricultural or
environmental hazards. The compounds (naphthalene, phthalate esters, phenol and
pentachlorophenol) which exceeded MDLs, occurred at small concentrations and degrade in soils.
The author has no satisfactory explanation for the apparent widespread occurrence of naphthalene
in soils but phthalate esters, which lend flexibility to plastics, are ubiquitous in laboratories and the
environment. It is possible that small amounts of these compounds resulted from soil contamination
during sample collection, preparation and analysis. Phenol may have resulted from degradation of
a variety of materials including pesticides and soil organic matter and pentachlorophenol is a
widely used wood preservative.

Production and use of polychlorinated biphenyls (PCBs) has been banned in Canada since the
mid-1970's but they, like phthalate esters, are ubiquitous in the environment and small amounts
(<250 :g kg-1 dw) occurred in all soils. They are not taken up by plants and it was concluded that
they do not represent a significant hazard to agricultural crops.

Production and use of organochlorine pesticides (OCs), except endosulfan, has been banned in
Canada since the mid-1970s. Thus, a higher incidence of alpha chlordane, dieldrin and aldrin in the
intensively cropped soils than in the other soils was assumed to reflect greater past use of these
compounds in intensive than in general crop production. However, in most cases only trace amounts
(<10 :g kg-1 dw) of OCs were observed in the soils from either group. Endosulfan (Thiodan) is a
stomach insecticide used on field crops and the small concentrations (<100 :g kg-1 dw) observed
in several soils were thought to reflect its use during the growing season in which the soil samples
were taken. Total DDT (i.e., op'-DDT + pp'-DDT + p,p'-DDE) concentrations in several soils
approached 1 mg kg-1 dw and in one apple orchard soil exceeded 70 mg kg-1 dw. DDT is very
persistent in soil and even the >70 mg kg-1 dw residue probably resulted from past use of this
compound in apple production. DDT represents no hazard to crop production and residues in soil
approximating 1 mg kg-1 dw probably represent little environmental hazard, however, larger residues
may represent a significant environmental hazard.

The results of organophosphorous pesticide (OPs) analysis were similar for both groups of soils.
Fonofos, which is moderately persistent in soil, was the only compound detected and it was observed
at concentrations of <100 :g kg-1 dw in all soils. These residues were consistent with its use in crop
production and probably do not represent a significant hazard to the environment.

Neutral, phenoxy acid and carbamate herbicides were measured in surface soils from seven of
the Agriculture and Agri-Food Canada, Soil Quality Evaluation Program - benchmark sites and the
six intensively cropped farms. Results were similar for the two groups of soils and indicated no
detection of most of these compounds in most of the soils. It was concluded that residues of these
compounds probably do not represent a significant hazard to the environment.

12-3
PAHs, OCs and PCBs in Hamilton and Sarnia Sludges

The PAH, OC and PCB results for Hamilton and Sarnia sludges indicated good reproducibility
between duplicate analyses and among replicate samples. There were variations in concentrations
over the five-month sampling period but the individual PAHs in Hamilton sludge generally were
<50 mg kg-1 dw and in Sarnia sludge <5 mg kg-1 dw; total PCBs in the Hamilton sludge were <2.5
mg kg-1 dw and in Sarnia sludge <1.5 mg kg-1 dw; and with few exceptions OC concentrations in
both sludges were <100 :g kg-1 dw. It was concluded that the results accurately reflected the
concentrations of PAHs, OCs and PCBs in the Hamilton and Sarnia sludges.

PAHs, OCs and PCBs in Soils Incubated with Hamilton and Sarnia Sludges

In general PAHs, OCs and PCBs were not detected or concentrations approximated the respective
MDLs in the two southern Ontario surface (0-15 cm) soils that were not treated with sludge.
Treating these soils with 32 t ha-1 dw of Hamilton and Sarnia sludges, which approximated the
maximum acceptable loading for Ontario soils, did not measurably change their PAH, OC and PCB
concentrations. Treating them with 200 t ha-1 dw of sludge, a rate greatly exceeding acceptable
Ontario practise, caused only minor increases in the concentrations. Moreover, there was evidence
for degradation of some compounds, particularly the low molecular weight PAHs, during the
eleven-week experimental period. The Hamilton and Sarnia sludges were thought to represent
worst-case industrial organic contamination, however, it was concluded that even application of
these sludges according to recommended Ontario practise does not represent a significant PAH, OC
and PCB hazard to agriculture and the environment.

PAHs, OCs and PCBs in Sludge Treated Southern Ontario Field Soils

Polynuclear aromatic hydrocarbons, OCs and PCBs were not detected or their concentrations
approximated analytical MDLs in the seven southern Ontario soils which had received from one
to three sludge applications according to recommended practise. The concentrations in the eighth
soil which had received much more than the recommended maximum sludge application rate were
slightly larger than the MDLs. It was concluded that land application of southern Ontario sludges
according to recommended practise probably does not represent a significant PAH, OC and PCB
hazard to agriculture and the environment.

REFERENCES

Black, S.A., Graveland, D.N., Nicholaichuk, W., Smith, D.W., Tobin, R.S., Webber, M.D. and
Bridle, T.R. 1984. Manual for land application of treated municipal wastewater and sludge.
Environment Canada Manual EPS 6-EP-84-1, Ottawa, ON.

Webber, M.D. and Lesage, S. 1989. Organic contaminants in Canadian municipal sludges. Waste
Management & Research 7: 63-82.

12-4
CHAPTER 13
DEVELOPMENT OF INDICATORS OF SOIL QUALITY AND SUSTAINABLE
LAND MANAGEMENT, USING KNOWLEDGE AND INFORMATION OBTAINED
FROM INNOVATIVE CONSERVATION FARMERS

M. Cann1, J. Dumanski1, S. Gameda1 and M. Brklacich2

1
Agriculture and Agri-Food Canada, Research Branch, Centre for Land and
Biological Resources Research, Ottawa, Ontario
2
Carleton University, Geography Department, Ottawa, Ontario

INTRODUCTION

Recent research has examined the effect of changes in soil quality on soil productivity through the
use of predictive soil degradation models and long-term field experiments. Although these studies
demonstrate the negative effects of degradation, soil productivity may often appear to be unaffected
at the farm level due, in part, to technological advances such as improved seed varieties, more
effective fertilizers, improved pesticides, and better land management. In the past, these technical
advances have been effective in improving or maintaining crop productivity at a reasonable cost
even though soil quality may be declining. How can these declines in soil quality be identified, and
corrective actions taken before irreparable damage to the soil occurs? The development of indicators
of soil quality deterioration that can be used by individual farmers in their own fields can contribute
to the resolution of this issue.

In rural areas, there are farmers who are very innovative in adoption of technology, and who possess
an intuitive knowledge of soil conditions and sustainable production methods. These innovative,
conservation farmers make use of indicators of soil quality and sustainable land management in
their on-going agricultural practices. This study was designed to capture this local knowledge, and
to utilize it to estimate the impacts of soil degradation on crop yields, as well as providing some
guidance on sustainable land management systems. If properly validated, and provided through
suitable extension programs, these indicators could also be used by many other farmers to monitor
the quality of their soil resources.

OBJECTIVES

The objectives of this study were to use the observations of innovative conservation farmers to: i)
identify indicators that distinguish between the kinds and severity of soil degradation; ii) determine
losses of crop productivity due to soil degradation; and iii) ascertain threshold levels at which
ameliorative measures should be taken. These farm-level observations were also to be applied to a
framework for evaluating sustainable land management, and to be used to develop. an expert system
which would function as a conservation planning and research tool.

13-1
IDENTIFYING INDICATORS OF SOIL DEGRADATION

Two sources of information were integrated to create an information base of indicators of soil
degradation, namely a review of the available literature and questionnaire data from innovative
conservation farmers.

Review of Literature

A literature review revealed several indicators that are factors of soil quality such as topsoil depth,
substrate properties, soil texture, bulk density, electro-conductivity, and soil pH. Much of the
literature focused on the effect of topsoil loss on crop productivity. Unfortunately, most of the
published literature provided data on wheat and corn, with very little information on the impacts
of degradation on other crops.

The relationship between topsoil depth and crop productivity is best described by a
Mitscherlich-Spillman function which incorporates the following four elements of agronomic
response: i) it should conform to the law of diminishing returns; ii) when topsoil depth is zero, some
yield should still be possible; iii) there should be a finite maximum yield; and iv) topsoil depth in
excess of the rooting zone should not decrease yields (Christensen and McElyea, 1988).

A review of the last 50 years of erosion research by Pierce (1991) concluded that: yields from
experimental field studies were lower than those from actual farm yields; restoration of yields by
increased use of fertilizer is dependent on subsoil properties; yields were related linearly to topsoil
depth in many studies; it is uncertain if productivity losses are permanent; uneroded sites are
becoming harder to find; some effects of erosion have yet to be studied; technology can mask the
gradual productivity losses over the short-term, making it difficult to detect yield decreases
immediately; and spatial relationships and soil variability within a landscape have generally been
ignored.

These observations suggest that other sources of information may be needed to supplement
experimental results if timely answers are to be provided.

Innovative Conservation Farmers

The term "innovative conservation farmers" is used in this report to refer to farmers who are early
adapters and refiners of new conservation technology. Many of these farmers modify new
technologies to take advantage of their particular physical production environments and the
prevailing economic conditions. In most cases, these farmers integrate research results about soil
quality changes with their own indicators of soil degradation, with implicit criteria for the adoption
of remedial cropping practices, and with economic indicators, to develop an adaptive or sustainable
land management program for their farm. In these ways, they intuitively employ the five pillars of
sustainable land management (SLM): productivity, security, protection, viability, and acceptability,
in their decision making process.

13-2
Experience has shown that innovative conservation farmers can be valuable sources of information.
Despite the lack of precise measurements and experimental layouts, these farmers often are keen
observers and are very sensitive to changes in crop production levels on their farms. Therefore, they
can accurately predict comparative changes in yield caused by various factors of degradation on
different parts of their fields.

METHODS

A detailed questionnaire was designed to capture the observations of innovative conservation

farmers as well as to identify the indicators used by the farmers to define the symptoms of soil
degradation. Appendix 1 contains the erosion questions1, including examples meant to orient the
response. From these, the farmers reported on the specific indicators, management practices, etc. that
they employed. An example of one farmer's response is included.

A total of 84 structured questionnaires were distributed to innovative conservation farmers selected

by local conservation officers in Alberta, Saskatchewan and Manitoba. In Prince Edward Island
30 questionnaires were administered directly to selected farmers by conservation officers. Follow-up
interviews were conducted subsequent to questionnaire administration in order to clarify responses
and fill information gaps observed from responses. Questionnaire responses were obtained from 35
farmers in the Prairies and from all 30 farmers in Prince Edward Island. The farmers were asked
to:
• identify degradation indicators;
• quantify yield changes due to degradation;
• estimate soil quality changes; and
• describe at what point in the degradation process ameliorative practices became essential
(cost effective) and what practices they found to be effective.

The innovative conservation farmer observations were compared to results of conventional

degradation studies and they were reviewed by regional agronomists and conservation officers.

RESULTS AND DISCUSSION

Soil erosion and salinization are the most prevalent types of soil degradation in the provinces
surveyed (Agriculture Canada, 1985). The innovative conservation farmers were asked to describe
the severity of soil degradation in three levels; slight, moderate and severe. Generally, a distinction
was made between indicators of soil degradation due to wind and water erosion, although no such
distinction was made for estimated yield reductions.

_______________
1
Impact of soil degradation on crop yields in the Canadian prairies: Collection registration number:
AGR/RBR-195-05080.

13-3
Soil Degradation Indicators Identified by Innovative Conservation Farmers

Examples of indicators that were reported by the farmers as ones that they use to identify the degree
of soil degradation in a field are given in Table 13-1.

Table 13-1. Degradation indicators as reported by innovative conservation farmers

Degradation
Wind Erosion Indicators
Level
Slight increased stoniness on knolls; soil in road-side ditches
cobble-sized stones; heavy dust clouds; mixing of topsoil with subsoil;
Moderate
soil accumulation along fences; uneven plant growth
very heavy dust clouds; stunted to no plant growth; burial of fence lines; increased
Severe
frequency of stones throughout the entire field; very small soil aggregates
Water Erosion Indicators
Slight poor germination; cultivable rills
larger rills and small gullies; soil deposition at base of slope;
Moderate
localized increased stoniness; yellow clay in topsoil
visible water runs or gullies; no topsoil left; very uneven crop growth;
Severe
buried seedlings at base of slope
Salinization Indicators
white powder on topsoil; salt crystals in shallow soil samples;
Slight
weed growth increasingly noticeable
salt crystals visible on surface; widespread kochia and other weed growth; absence
Moderate
of saline sensitive plants; surface crusting is more prominent
soil is always wet; hard-crusted topsoil; topsoil is very white due to salt crystals;
Severe only salt-resistant weeds grow; waterlogging in discharge areas; seedbeds are very
difficult to prepare

Yield Reductions Due to Soil Erosion as Estimated by Innovative Conservation Farmers

The innovative conservation farmers estimated yield reductions due to soil erosion for several crops.
Fig. 13-1 shows the medians of the estimated yield reductions for each level of soil erosion.

In general, yield losses in the small grains and oilseeds were 10 to 15% due to slight erosion, 15 to
30% due to moderate erosion and 40 to 65% due to severe erosion. Flax and canola were reported
as the most sensitive to soil erosion at the slight erosion level, followed by the cereals. Potatoes and
alfalfa were the least sensitive to soil erosion in that yield reductions were not noticeable until the
moderate erosion level.

13-4
 Fig. 13-1. Yield reductions due to soil erosion as estimated by
innovative conservation farmers (based on 65 responses).

Yield Reductions Due to Salinity as Estimated by Innovative Conservation Farmers

Yield reductions were as high as 15% for slightly saline soil, 50% for moderately saline soil, and
from 65 to 100% for severely saline soil (Fig. 13-2). Wheat, flax and canola were reported as the
most sensitive crops, whereas alfalfa was slightly more salt tolerant.

Yield reductions for slight salinity were similar to those estimated for slight erosion. However, the
yield reduction estimates for moderate and severe salinity were greater than the estimates for
comparable levels of soil erosion. For example, wheat yields from moderately eroded fields were
about 30% less than yields from non-eroded fields whereas moderately saline fields had yield
reductions of approximately 50%.

Ameliorative Cropping Practices Suggested by the Innovative Conservation Farmers

It is noteworthy that soil degradation control practices were generally initiated by the farmers only
when soil degradation had advanced to the moderate level, i.e. the effects of slight degradation were
normally internalized by the fanner. At this point, a variety of control practices were used depending
on the condition and the farmers' preferences. In many cases, the conservation farmers continued
to use these control practices even after the level of soil degradation had decreased. Table 13-2 lists
some of the most common ameliorative practices.

13-5
 Fig. 13-2. Yield reductions due to soil salinity as estimated by
innovative conservation farmers (based on 35 responses).

Table 13-2. Cropping practices initiated by innovative conservation farmers in response to soil
degradation

Degradation
Erosion Salinity
Level
Slight grassed waterways; less summerfallow continuous cropping; planting salt-tolerant
crops
Moderate seed to forages; longer crop rotations; longer crop rotations; recharge area
reduce tillage operations; winter cover management; draining excess water; leave
crops; chemical weed control; increased native prairie intact; avoid working low areas
use of shelterbelts
Severe continuous cropping; plant alfalfa; plant alfalfa; seed to perennial forages
increased fertilizer on eroded knolls; seed
to perennial forages

Integrating Scientific and Innovative Conservation Farmer Results

Several studies reported on the impacts of soil erosion and salinity. For slight and moderate
degradation, yield reductions estimated by the innovative conservation farmers were similar to the
results in: Dormaar et al. (1986); Ives and Shaykewich (1987); Massee and Waggoner (1985);
Fowler and Hamm (1980) and Bole and Wells (1979). For severe degradation, yield reduction
estimates by innovative conservation farmers tended to be larger than those reported in the scientific
literature.

13-6
 RELATIONSHIP BETWEEN SOIL QUALITY AND
SUSTAINABLE LAND MANAGEMENT (SLM)

Sustainable Land Management combines technologies, policies and activities aimed at integrating
socio-economic principles with environmental concerns so as to simultaneously:

• maintain or enhance production/services (Productivity);

• reduce the level of production risk (Security);
• protect the potential of natural resources and prevent degradation of soil and water quality
(Protection);
• be economically viable (Viability); and
• be socially acceptable (Acceptability).

These objectives are identified as the five pillars of SLM (Dumanski et al., 1991) and soil quality
elements are key factors in the first four of these pillars.

Indicators Identified for the Framework for Evaluating Sustainable Land Management

A framework for evaluation of Sustainable Land Management (FESLM) is being developed by an

international working group. The objective of this work is to develop a scientifically-sound
procedure for evaluation of whether certain types of land management will lead towards
sustainability or away from it (Smyth and Dumanski, 1993).

There are five steps in the framework for evaluating sustainable land management ( Smyth and
Dumanski, 1993):

• definition of the purpose - objective and means;

• analysis - evaluation factors, diagnostic criteria, indicators and thresholds;
• assessment endpoint;
• validation; and
• implementation of recommendation.

This study focused on the analysis step of the FESLM.

The analysis step is divided into identification of evaluation factors, the development of diagnositic
criteria, and the determination of key indicators and thresholds. The prairie farmers were asked to
comment on the concept of sustainable agriculture. For this study, and for the questionnaire given
to the farmers, the definition of sustainable land management (SLM) was modified as follows:

Sustainable land management is a group of technologies applied at the farm level, which
individually or in combination contribute to sustainable agriculture. Sustainable land
management attempts to maximize efficiency of inputs in relation to maximizing outputs,
but it also incorporates considerations of the long-term environmental and social costs
associated with the outputs. Sustainable land management is a set of recommendations (set
of practices), based on sound scientific principles, which simultaneously promote

13-7
 agricultural productivity, ensure economic and social return and promote or enhance the
quality of the environment and the land (Dumanski et al., 1991).

Many of the farmers related SLM to soil conservation practices. The cost of these practices were
perceived as high when considering the low crop prices of recent years. Education through
demonstration and government incentives were recommended for any package that was designed
to promote sustainable agriculture. Local processing of crops and livestock was viewed as a method
to ensure and, perhaps, even resurrect agricultural communities. The innovative conservation
farmers reported the use of per cent organic matter as an indicator of soil quality and suggested
various sustainable practices that may maintain or increase organic matter content. Indicators of
SLM identified by the innovative conservation farmers were:

• decrease in the area that is summerfallowed;

• increase in acreage allocated to forage or grazing pasture (livestock integration);
• increased use of winter crops to distribute the work load throughout the year and provide
soil cover;
• decrease in field sizes;
• crop diversification by including crops such as legumes or forages in a rotation;
• decrease in the application of commercial fertilizers;
• increase in erosion and moisture control activities such as shelterbelt establishment, grassing
of natural waterways, contour plowing, and standing stubble;
• increased use of herbicides for weed control on fallow;
• improved drainage and recharge/discharge management;
• local processing of produce;
• extension programs with on-farm demonstration and the availability of new technology;
• increased research activities into conservation practices and new crop varieties;
• return of soil organisms such as earthworms; and
• development of new markets for crops such as winter rye, winter wheat, and forages.

These indicators could be included in the analysis step of the evaluation framework. Thresholds for
these indicators were not identified by this group of innovative conservation farmers.

CONCLUSIONS

Early recognition of soil degradation is a key component for sustainable land management. The
innovative conservation farmers reported several indicators which may be used to determine the kind
and severity of soil degradation. All of these are observable indicators that can be assessed in a field
or landscape without the need for special equipment. These indicators describe, to some extent, what
is not sustainable and in this way assist in identifying ameliorative practices which will enhance the
sustainability of a farm operation.

The indicators reported by the innovative conservation farmers can be linked to changes in soil
quality. Also, credible estimates of yield sensitivity to the degree and the type of degradation can
be obtained from innovative conservation farmers, but these estimates should be verified against
published information and by regional agronomists.

13-8
Farmers will internalize the impacts of soil degradation up to the threshold of moderate degradation.
At this point, they begin soil degradation control practices and continue to use these long after soil
productivity has improved. This indicates that the ethic of conservation, once established, will tend
to be maintained.

Finally, the structured questionnaire approach provides credible information in a timely and
cost-effective manner. The successful implementation of this approach requires the collaboration
of local conservation agencies as well as experienced conservation farmers.
This compilation of information has provided indicators that can serve as analytical tools within a
framework for evaluating sustainable land management.

REFERENCES

Agriculture Canada. 1985. Agricultural soil and water resources in Canada: situation and outlook.
Ottawa. 18pp.

Bole, J.B. and Wells, S.A. 1979. Dryland soil salinity: effect on the yield and yield components
of 6-row barley, 2-row barley, wheat, and oats. Can. J. Soil Sci. 59: 11-17.

Christensen, L.A. and McElyea, D.E. 1988. Toward a general method of estimating
productivity-soil depth response relationships. J. Soil Water Conserv. 43: 199-202.

Dormaar, J. F., Lindwall, C.W., and Kozub, G.C. 1986. Restoring productivity to an artificially
eroded Dark Brown Chernozemic soil under dryland conditions. Can. J. Soil Sci. 66:
273-285.

Dumanski, J., Eswaran, H., and Latham, M. 1991. A proposal for an international framework for
evaluating sustainable land management. In Evaluation for sustainable land management in
the developing world. Vol. 2, Technical papers, Proceedings no. 12(2). International Board
for Soil Research and Management, Bangkok, Thailand.

Fowler, D.B. and Hamm, J.W. 1980. Crop response to saline soil conditions in the Parkland area
of Saskatchewan. Can. J. Soil Sci. 60: 439-449.

Ives, R.M. and Shaykewich, C.F. 1987. Effect of simulated soil erosion on wheat yields on the
humid Canadian prairie. J. Soil Water Conserv. 42: 205-208.

Massee, T.W. and Waggoner, H.O. 1985. Productivity losses from soil erosion on dry cropland
in the intermountain region. J. Soil Water Conserv. 40: 447-450.

Pierce, F.J. 1991. Erosion productivity impact prediction. Pages 35-52 in R. Lal and F. J. Pierce
(Eds.). Soil management for sustainability. Soil and Water Conservation Society, Ankeny,
IA.

Smyth, A.J. and Dumanski, J. 1993. FESLM: An international framework for evaluating
sustainable land management. FAO Discussion Paper. Land and Water Development
Division. FAO. Rome. 74pp.

13-9
 APPENDIX 1. QUESTIONNAIRE SAMPLE
2
Section 3. Soil Erosion

Topsoil depth, the darker surface layer of the soil, is often used to determine the severity of soil erosion.
Three severity levels based on topsoil depth could be defined as:

Erosion Level Topsoil Remaining (%)

slight 75 - 90
moderate 50 - 75
severe less than 50

8. Do you use indicators other than topsoil depth to identify areas where soil erosion has
occurred?
Yes x No (go to question 9)

If Yes, please describe the levels of erosion severity using your indicators. (These indicators could
include soil properties and/or landscape features such as increased stoniness on hilltops, more
widespread soil crusting, uneven plant growth, change in soil colour, appearance of rills or gullies,
etc.)
Slight

Moderate
Change in soil colour
Uneven plant growth - vigorous alfalfa, poor wheat etc.
Increased stoniness, especially unpickable pebbles and fist-sized stones Difficulties with soil
incorporated chemicals, i.e. Treflan
Severe
Drastic changes in soil colour, i.e. white topped knolls visible across valley
Gullies cut down to subsoil exposure

9. What is the range of topsoil depths on the NON-ERODED soil in your area?
1-12"

Not known (go to question 11)

10. What is the range of the total soil depths (A and B horizon or depth to gray layer) on the
NON-ERODED cropland in your area?
2-18"

Not known (go to question 11)

11. Have you noticed soil erosion occurring in your area?

Yes x No (go to section 4)

_____________________
2
Examples of one farmer's response are presented throughout in italics. Note that although examples
were provided for several of the questions, the farmers reported only those indicators and practices
that they used on their farms.

13-10
 If Yes, please indicate the level of severity and the percent of your area associated with each
level of erosion.

Erosion Level Percent of Cropland

Slight _______________ 70%

Moderate _______________ 15%

Severe _______________ 5%

12. Do you think that there are some locations in your area where soil erosion is more common
than others?
Yes x No _______ (go to question 13)

If Yes, please describe these locations. (These locations could include characteristics such as soil
texture, position on hillslope, steepness of the terrain, certain soil moisture regimes, water table
levels, type of soil substrate, etc.)

Slopes of Pipestone Valley, particularly the more arid north east side.
A peculiar aspect of wind erosion in this area is that newly broken virgin soils,
particularly those that have been tree covered, blow very easily if not protected by
crop residue. When dry, these soils become flour like and will actually flow around
tires and form puddles of dry dirt. Such fields require very careful management
because of the lack of structure when dry.

13. Did the soil erosion in your area occur gradually, as a result of extreme weather conditions or
by some combination of the two?
Gradual _____________
Extreme weather _____________
Combination _______x______

If extreme weather was at least part of the cause, please describe the nature of these weather
events and indicate when and how often these conditions occurred.

In 17 years at this location I have seen four severe wind storms during daylight hours
in which it was impractical to be outside. Two of these occurred when the soil was
unprotected by active plant growth. The one on Sat. June? 1985 would have been the
greatest natural calamity to strike the prairie region had it occurred in mid May.

14. Once erosion has been identified, are any cropping practises employed to restore soil
productivity and/or to prevent further losses?
Yes x No ____ (go to question 15)

13-11
 If Yes, please describe these cropping practises and indicate at what level of erosion severity
these practises were initiated. (These practises could include increased fertilization, the use of green
manure (under seeding), a change in crop rotation, grassed waterways, contour cropping, planting
trees as windbreaks, etc.)

A majority of farmers in this area have gone to continuous cropping. Some by sudden
decision following a dust storm and others having achieved good yields with extended
cropping. Both situations have resulted in increased fertilization and better weed
control. The acceptance of air seeders has played a part. Trees are generally resisted
- reasons:
1) recent expenditures of money and labour to clear land and
2) the high number of sloughs already cut up the fields

15. Please describe any changes in soil quality that you consider a result of soil erosion. Indicate the
changes that took place at each level of erosion severity. (This could include soil characteristics
such as decreased soil moisture, loss of organic matter, more widespread surface crusting, more
frequent waterlogging, decreased rooting depth, increased stoniness, etc. Also indicate if no changes
were observed.)

Slight
decreased infiltration rates
increase in flooding of sloughs
Moderate
drastically decreased infiltration rates stoniness
decreased soil moisture
water logging in "bath tub rings" increase of flooding in sloughs
Severe
severe loss of organic matter, alfalfa probably the only profitable cropping
option. Even this option is curtailed by crusting, stoniness, cost of phosphate
fertilizer and lack of moisture retention

16. Do you think that, in the absence of increased inputs or technologies, the changes in soil quality
outlined in question 15 would have resulted in lower crop yields?

Yes x No _____ (go to question 17)

If Yes, please estimate these yield reductions, identify the soil characteristics which seem to
have the greatest impact for each crop and indicate the level of erosion severity at which the
yield reductions would be observed. (Please be as quantitative as possible and remember to include
the units of measure for each yield estimation. For example you could identify a 10% wheat yield
reduction, or a 200 kg/ha barley yield reduction, or a 0.5 bu/ac corn yield reduction.)

13-12
Crop Yield on Non-Eroded Yield on Eroded Soils by Erosion Change in Soil
Soils Level Characteristics
LevelYield
loss of organic matter &
CWRS Slight 28 bu
30 bu/ac decrease in infiltration
Wheat Moderate 20 bu rates
Severe 15 bu
LevelYield same as wheat
Barley 45 bu/ac loss of organic matter
Moderate 30 bu
Severe 22 bu
LevelYield
Canola
and 25 bu/ac Slight 24 bu _________________
Mustard Moderate 18 bu __________________
Severe ?? __________________
LevelYield
Oats 60 bu/ac Slight 55 bu __________________
Moderate 48 bu __________________
Severe 35 bu
LevelYield
Flax 15 bu/ac Slight 13 bu __________________
Moderate 10 bu __________________
Severe 5 bu
LevelYield
Br-Alf 1.5 tons/ac Slight 1.5 t ___________________
Moderate 1.2 t ___________________
Severe 1.0 t

13-13
CHAPTER 14
SOILCROP: AN INNOVATIVE CONSERVATION FARMERS' EXPERT
SYSTEM FOR CONSERVATION FARM PLANNING

S. Gameda and J. Dumanski

Agriculture and Agri-Food Canada, Research Branch,

Centre for Land and Biological Resources Research, Ottawa, Ontario

INTRODUCTION

Agricultural land in the Canadian Prairies undergoes extensive and continual degradation primarily
due to wind and water erosion as well as salinization. Maintenance of the productive capacity of
land requires implementation of methods to assess and reverse trends towards greater degradation.
Although empirical and process models are effective in determining soil degradation, the extent of
parameters required, as well as difficulties in validating and interpreting the results, make their use
in management at the farm level difficult. Moreover, such models tend to be specific to landscape
or individual degradation events, whereas the problems encountered in agriculture tend to be
multi-causal and occur on a range of landforms and soil types under diverse cropping sequences.

Often contrasting levels of soil degradation can be observed in otherwise similar landforms within
any given agricultural region. These differences reflect the ability of certain farmers to make early
identification of soil degradation problems and implement the remedies required to amend them.
These innovative conservation farmers have intuitively developed and make use of indicators of the
status of soil quality on their land to assist them in their diagnosis of degradation problems and
choice of management practices. The ability to capture and disseminate this knowledge base to all
farmers would greatly assist in maintaining the productive potential of agricultural land.

In dealing with such a knowledge base, it should be remembered that the diagnostic and analytical
methods of farmers primarily make use of combinations of quantitative, qualitative and symbolic
or analogical knowledge drawn implicitly from experience. In contrast, researchers and extension
personnel primarily use quantitative data and information in the diagnosis and analysis of soil
degradation problems.

Until recently, a means of making use of incomplete or analogical information did not exist as such
knowledge was difficult to categorize, or formalize into algorithms. However, the advent of expert
systems, coupled with shells facilitating their development, has made it possible to address such
problems. In particular, expert systems provide an opportunity to make use of both the
predominantly symbolic and qualitative knowledge of farmers and the more rigorous quantitative
information obtained from scientific research.

14-1
The objective of this study was to develop a soil degradation - crop productivity expert system based
on the knowledge of innovative conservation farmers as well as on information obtained from
scientific research. The expert system is designed for diagnosing the type and level of degradation
on a farm, along with associated yield reductions, as well as the ameliorative practices required and
expected improvements in productivity associated with each practice. The expert system is targeted
for delivery by conservation officers working with farmers.

DEVELOPMENT OF EXPERT SYSTEMS

Expert systems have been defined as models and associated procedures that exhibit, within specific
domains, a degree of expertise in problem solving comparable to that of human experts (Ignizio,
1991). In practice, however, they are computer programs that solve complex problems within a
defined domain (Plant and Stone, 1991). Expert systems can make use of symbolic or qualitative
information, incorporate uncertainty and reach a conclusion using disparate and/or incomplete
information. In effect, they provide a framework for capturing and applying non-algorithmic
knowledge (Plant and Stone, 1991).

Expert systems consist of a knowledge base, an inference mechanism and a natural language user
interface (Doluschitz and Schmisseur, 1988). The knowledge base encompasses domain facts and
heuristics, the inference mechanism checks facts against the knowledge base in order to reach a
conclusion, and the user interface elicits specific facts and data from the user. The interface also
provides the conclusions reached along with explanations of the reasoning used to reach those
conclusions.

A critical activity in the development of expert systems is knowledge acquisition and

representation. The knowledge base for expert systems is usually developed by tapping the expertise
of a domain expert. In agriculture, however, the expertise to be encoded does not reside within one
expert but is shared among researchers, extension specialists and experienced farmers.

The procedures used by knowledge engineers to extract expertise from domain experts, and
structure the knowledge base, are also suitable for eliciting the knowledge of researchers and
extension specialists. However, it is relatively difficult to tap and represent the expertise of
experienced farmers for two main reasons. First, this expertise does not lie within one individual but
exists amongst a larger group, and secondly, the knowledge base may overlap and/or be conflicting
within the expert group. This implies the need for large-scale extraction of knowledge coupled with
resolution of any conflicting information that may arise. In particular, there is greater need for
interpretation of farmer-provided information in order to transform it into a form suitable to an
expert system framework.

Although there is contrast in the form and content of information as perceived by researchers on
the one hand and farmers on the other, the wealth of knowledge that resides among farmers should
not be underestimated. When such knowledge has been utilized in conjunction with scientific
principles, it has led to breakthroughs in such areas as, for example, soil classification and land
evaluation in developing countries (Pawluk et al., 1992).

14-2
Expert systems in agriculture can be classified into five categories: i) heuristic expert systems that
utilize the decision making characteristics of domain experts; ii) real-time expert controls for
regulating control systems; iii) model-based systems that link simulation models and expert
systems; iv) expert databases or linked databases and expert systems; and v) problem specific shells
for developing expert systems pertinent to agricultural problems (Jones, 1989). Of these, heuristic
and model-based expert systems are primarily diagnostic and have been the ones most applicable
to agricultural decision making.

Heuristic expert systems are most successful when applicable to very narrow, well-defined domains.
Expert systems that are model based, also known as hybrid systems, function as parameter
initializing tools that structure the required information into a format suited to the model and as
explanatory schemes that interpret model results in a user-friendly manner.

The main limitation of heuristic expert systems is the extensive input required from experts during
the knowledge acquisition and representation stage. Consequently, such systems are suitable in
areas where considerable institutional commitment exists or where the domain has high economic
value to defray expert consultation costs. Limitations to model-based systems are that they require
close collaboration between expert system and model developers. In fact, due to the dynamic nature
of modelling, both expert system and model development should be undertaken jointly. Both types
of diagnostic expert system require ongoing support for long-term viability.

Recently, there has been a profusion in the number of expert systems developed for agricultural
applications. Examples include: systems for pest management in the production of soybeans
(Batchelor et al., 1989), apples (Huber et al., 1990) and honey (McClure et al., 1993); applications
in cotton crop management (McKinion et al., 1989); planning for conservation (Heatwole, 1987;
Montas and Madramootoo, 1992) and crop rotation (Buick et al., 1992); irrigation scheduling
(Clarke et al., 1989), and whole-farm machinery management (Kline et al., 1987; Lal et al., 1992).
Although the success of some of these systems has been primarily in their proof of concept, others
have shown very high client acceptance. For example, the quality of information in the expert
systems BEE AWARE (McClure et al., 1993) and GOSSYM/COMAX (McKinion et al., 1989) for
apiary and cotton production, respectively, has been of such high value that producers are willing
to purchase them for their management use. In all cases, expert systems have served a useful
function in synthesizing existing research knowledge into a form that assists in decision making.

However, none of the agricultural expert systems developed to date incorporate the knowledge of
innovative conservation farmers. As indicated earlier, this is partly due to the difficulty in capturing
and formalizing such knowledge, but it also stems from a bias that underestimates the "know-how"
of experienced farmers. This study is based on the premise that a wealth of information can be found
in the experience of innovative conservation farmers and that the principles underlying farmers'
qualitative knowledge can be linked to information obtained from scientific research.

14-3
 SOILCROP: SOIL DEGRADATION - CROP PRODUCTIVITY EXPERT SYSTEM

Knowledge Base Development

A two tiered approach was taken for knowledge acquisition in the development of the soil
degradation - crop productivity expert system: i) a structured questionnaire was prepared and
administered to 84 innovative conservation farmers selected by local conservation officers in
Alberta, Saskatchewan and Manitoba. Questionnaire responses were obtained from 35 of the
farmers surveyed. A breakdown of the returns by soil zone consisted of 6, 12, 13 and 4 responses
from the Brown, Dark Brown, Black and Gray soil zones, respectively; ii) an extensive review of
the literature was conducted to identify research-based information on soil degradation - crop
productivity relationships. These two sources were supplemented with information from regional
conservation and farming practice manuals. The information obtained from farmers was
cross-checked against findings from the scientific literature and validated by regional agronomists
and conservation officers prior to incorporation into the expert system knowledge base.

The information sought from innovative conservation farmers consisted of: i) identification of
indicators of the level of degradation; ii) expected yield reductions due to each level of degradation;
and iii) the management practices required for soil amelioration at each level of degradation. Levels
of degradation were categorized as slight, moderate and severe, depicting topsoil (Ap horizon)
losses of less than 25 percent, 25 to 50 percent, and greater than 50 percent, respectively.

The indicators reported by innovative conservation farmers are shown in Table 14-1. In identifying
these indicators, the farmers did not differentiate amongst them in terms of their relative diagnostic
importance, although it was clear that they had implicit criteria for selective use of the indicators.
There was, therefore, a need to classify and rank indicators in order to signify their relative
importance.

Knowledge Base Organization and Indicator Classification

In order to categorize farmer-identified indicators, it was important to associate a relative weight to

each indicator, because rule-based expert systems mimic decision making by associating certainty
or weighting factors to each rule. A two-stage weighting system was utilized to achieve this.
Initially, indicators were classified as either strategic, cumulative or tentative. Strategic indicators
make definite identification of type and level of degradation. Cumulative indicators identify
degradation additively. Tentative indicators suggest soil degradation but need to be in combination
with cumulative indicators to identify degradation type and level. Indicators were then ranked
according to their relative diagnostic merit. Class score was multiplied by rank values to give a
weighting or certainty factor for each indicator. Indicator classification and ranking was conducted
iteratively, and was based on input from several domain experts. Examples of indicator
classification and ranking are given in Table 14-2. Rules for the degradation module of SOILCROP
were structured on the basis of these indicators and their associated certainty factors.

14-4
Table 14-1. Indicators of soil degradation severity identified by innovative conservation farmers

Slight
• lighter coloured soil
• soil accumulation in ditches,
other low areas, fence lines
• slightly shorter crops
• increased stoniness on knolls
and hilltops
• cultivated soil appears
coarse
• crusting
• some dust in the air
Slight
• lighter coloured knolls
• thinner plant growth
• soil crusting on knolls and
deposition areas
• appearance of cultivable rills
and small gullies
• poor germination
• increased stoniness on hilltops
WIND EROSION
Moderate Severe
• large, heavy dust clouds • very stunted to no plant growth
• soil accumulation along fence • crusting on more than half of
lines and shelterbelts field
• mixing of top two soil layers • very heavy dust clouds
• soil crusting covering half the • white-topped knolls visible
field from across fields
• texture becomes more coarse • increased stoniness over most
• uneven plant growth of field
• increased stoniness over parts of • heavy soil accumulation at field
field • edges and fence lines
• hilltops light-brown colour • very small soil aggregates
• thinner, shorter plants • disappearance of fence lines
• decrease in soil aggregate size
WATER EROSION
Moderate Severe
• increased stoniness • no topsoil left
• poor crop growth • water runs visible from across
• topsoil subsoil mixing fields
• soil crusting • fully formed gullies that cut
• rills forming around field edges • down to subsoil
• appearance of larger rills and • increased number of rocks on
small gullies ridges
• soil deposition at slope bottoms • very uneven crop growth
• difficulty establishing forages
and other crops

In addition to classifying and weighting the indicators, they were organized into qualitative
relationships describing the degradation - yield phenomena. Scientific research has generated
extensive data and information in the domain of erosion-based degradation. However, conflicting
evidence exists as to the relative effect of different factors or parameters on soil degradation. There
is, also, a large dependence on locational influences. Furthermore, as the relationships that were
developed tend to be empirical, it is difficult to quantify these effects parametrically. For example,
a number of relationships have been developed between the depth of topsoil eroded and crop yield
losses. These relationships vary depending on such factors as depth of topsoil layer, quality of
organic matter in the topsoil, nature of the subsoil and climatic conditions.

14-5
Table 14-2. Examples of degradation indicators with associated class, rank and certainty factors

Deg'n Class
Wind Erosion Indicators
Level Type Score Rank Value
• crusting on more than half of field sv 2 7 7 49
• very heavy dust clouds sv 1 10 10 100
• white-topped knolls visible from across fields sv 1 10 10 100
• increased number of stones on knolls and hilltops sl 1 10 10 100
• very small soil aggregates sv 3 3 7 21
• large, heavy dust clouds md 1 10 10 100
• mixing of top two soil layers md 2 7 10 70
• difficulties incorporating chemicals md 3 3 1 3
• hilltops light-brown colour md 2 7 4 28
• soil accumulation in ditches, other low areas, fencelines sl 1 10 10 100
• cultivated soil appears coarse sl 2 7 7 49
• some dust in the air sl 1 10 10 100
• slightly shorter crops sl 2 7 4 28
• very stunted to no plant growth sv 2 7 10 70
• uneven plant growth md 2,3 7,3 7 21-49

Deg'n Class
Water Erosion Indicators
Level Type Score Rank Value
• gullies that cut down to subsoil sv 1 10 10 100
• appearance of larger rills and small gullies md 1 10 7 70
• appearance of cultivable rills sl 1 10 7 70
• increased stoniness on parts of field md 1 10 7 70
• soil deposition at slope bottoms md 1 10 10 100
• increased soil hardness md 2 7 4 28
• greater frequency of hardpan areas sl 3 3 4 12
• decreased soil moisture md 3 3 1 3
• increased flooding of sloughs sl 2 7 4 28
• very uneven crop growth sv 2 7 7 49
• buried seedlings in slope bottoms sv 1 10 10 100
Legend

Degradation Class Rank Value

Level Type Score
sl = slight 1 = strategic 10 Relative ranking Certainty level for use by expert
md = moderate 2 = cumulative 7 10 = highest system shell
sv = severe 3 = tentative 3 1 = lowest Score x Rank = Value
* Strategic, cumulative and tentative indicators were assigned scores of 10, 7 and 3, respectively, in accordance
with their relative diagnostic merit.

It is necessary, during expert system development, to organize and synthesize the underlying rela-
tionships behind the complexity of such information. This entails identifying the parameters to be
considered, establishing criteria for categorizing identified parameters, indicating the assumptions
made in developing criteria, and stating the qualitative relationships as a set of rules for use in the
expert system. What is entailed in such an exercise is not simply the qualitative interpretation of
quantitative relationships, but rather an extraction of the implied relationship behind the stated ones.
For example, although the stated relationship in Table 14-3 links crop yield to topsoil depth, effects

14-6
due to characteristics such as soil zone and uneroded topsoil depth can be inferred from locational
differences in the relationships. Such a process reveals the underlying links between the qualitative
knowledge base of innovative conservation farmers and the quantitative soil quality information
from scientific research; it is also characteristic of knowledge acquisition for expert system
development. The principles used in developing qualitative relationships for use in SOILCROP were
based on procedures originally described by Hackett (1988). Examples of the associations developed
are given in Tables 14-3 and 14-4. These relationships were then used to develop the set of rules
which drive and control the expert system. Some examples of the types of rules developed for
SOILCROP are given below.

Table 14-3. Criteria for categorizing selected parameters into classes for use in SOILCROP;
based on information from soil erosion crop productivity research (Larney et al.,
1992; 1994)

Examples of parameters used* Rating Developed

Topsoil loss Organic C Normal yields** Yield loss for SOILCROP
(%) (%) Spring Wheat (%)
<25 <2 <1500 <10 low/slight
25-50 2-3 1500-2500 10-35 moderate
>50 >3 >2500 >35 high
IRating Developed
Yield Recovery (% of uneroded)
for SOILCROP
<85 highly insufficient
85-90 moderately insufficient
90-95 slightly insufficient
95-105 restored
105-110 slightly increased
110-115 moderately increased
>115 highly increased
Erosion-Yield Relationships Yield response to incremental
Location (y yield; x=depth of topsoil topsoil loss (SOILCROP
removed) Equivalent)
Taber y = 1121 - 47.5x + 0.74x2 low
Lethbridge y = 1289 - 93.6x + 1.48x2 low to moderate
2
Hill Spring y = 1554 - 117.6x + 2.65x moderate
2
Cooking Lake y = 2578 - 164.1x + 2.73x severe
2
Josephburg y = 2665 - 45.9x - 2.05x severe

* Each parameter is rated individually

** Example of assumptions used in developing criteria for normal spring wheat yields: The average yield for the
region is 2000 kg/ha. Yields within 25 percent of this value are considered moderate; yields below or above
this threshold are considered low or high, respectively. Similar procedures were used for ranking other
parameters.

14-7
Table 14-4. Examples of developing qualitative classes used in SOILCROP in relation to different agroenvironments (location).

Common Biological Characteristics Erosion Effects Productivity Response

Location Soil Zone Topsoil Organic C Yields Erosion Yield loss Yield recovery
depth (cm) ES** ES depth ES ES Amend't % of SOILCROP
(%) (kg/ha) (%)
equiv. equiv. (cm) equiv. equiv. uneroded equivalent
Taber BC* 10-15 1.40 low 1146 low 5 mod 23 mod Fertilizer 105 restored
Manure 119 moderately increased
10 sev 44 sev Fertilizer 80 moderately
Manure 97 restored
Lethbridge DBC 15 1.58 low 1205 low 5 mod 12 low Fertilizer 125 highly increased
Manure 139 highly increased
10 sev 67 sev Fertilizer 84 moderately
Manure 132 highly increased
Hill BIC(Tn) 10-13 2.97 mod. 1522 mod. 5 mod 28 mod Fertilizer 144 highly increased
Spring Manure 202 highly increased
moderately increased
10 sev 59 sev Fertilizer 118
Manure 197 highly increased
Cooking GL 15-18 3.47 high 2585 high 5 mod 32 mod NA
Lake 10 sev 48
Josephburg BIC(Tk) 30-33 4.00 high 2653 high 5 slight 11 mod NA
10-15 mod 19-50 mod-
sev
20 sev 63
sev

* BC=Brown Chernozemic; DBC=Dark Brown Chernozemic; BIC(Tn)=Black Chernozemic (Thin); GL=Gray Luvisol;
BIC(Tk)=Black Chernozemic (Thick).
** ES=expert system (SOILCROP)

14-8
Examples of rules derived from the classifications described in Tables 14-3 and 14-4:
If: soil zone is Brown Chernozemic
Then: yield potential is low (Table 14-4)
and erosion impact on yields is low (Table 14-3c)
and response to amendments is low to moderate (Table 14-4)

If: soil zone is Dark Brown Chernozemic

Then: yield potential is Iow (Table 14-4)
and erosion impact on yields is low to moderate (Table 14-3c)
and response to amendments is moderate to high (Table 14-4)

If: soil zone is Thin Black Chernozemic

Then: yield potential is moderate (Table 14-4)
and erosion impact on yields is moderate (Table 14-3c)
and response to amendments is high (Table 14-4)

If: soil zone is Brown Chernozemic and erosion level is moderate

Then: fertilizer amendments to eroded areas will restore yields (Table 14-4)
and manure amendments to eroded areas will moderately increase yields (Table 14-4)

If: soil zone is Thin Black Chernozemic and erosion level is moderate
Then: fertilizer amendments to eroded areas will highly increase yields (Table 14-4)
and manure amendments to eroded areas will highly increase yields (Table 14-4)

These examples, although based on limited research, illustrate the qualitative relationships that can
be drawn and utilized in an expert system, Such relationships (rules) are easily changed as more
reliable information becomes available.

Development of the Computer Structure for SOILCROP

The knowledge base for the prototype expert system, SOILCROP, was developed by integrating
production rules from the indicators used by innovative conservation farmers with information from
scientific research, and by structuring these to develop qualitative relationships. The SOILCROP
expert system consists of degradation, crop yield, management practices and productivity modules
(Fig. 14-1). The degradation module determines the type and level of degradation on the basis of
a series of questions posed by the expert system to the farmer or conservation officer during
interactive consultation sessions on a computer. The crop yield module identifies expected crop yield
reductions based on the level of degradation diagnosed by the expert system. The management
practices module indicates the range of farmer identified, cost-effective management practices for
ameliorating soil degradation and crop yield losses. The productivity module determines the effect
that chosen management practices may have in reducing soil degradation and crop yield losses. Thus,
the observed indicators obtained during the interactive consultation are combined with external data
sources to provide the physical description of cropland, quantified soil quality changes, the type and
severity of soil degradation and finally recommendations for amelioration. Fig. 14-1 illustrates the
links between the modules within the overall expert system structure.

14-9
 Fig. 14-1. SOILCROP modules and inter-linkage.

The examples below illustrate how rules are formed from indicators to estimate type and level of soil
degradation. First of all, degradation level is identified on the status of the following qualifiers:
soil colour
soil coarseness and aggregate size
soil crusting
stoniness
soil accumulation
presence and intensity of dust clouds
status of rills and gullies
crop establishment and growth

Rules are then developed from this in the following manner:

If: Soil colour has significantly changed, and white-topped knolls are visible across fields
Then: There is a severe level of erosion on the farm - Confidence=100

If: Soil colour has become lighter over the mid slopes of fields and in localized areas
Then: There is at least a moderate level of erosion on the farm - Confidence=70

14-10
These rules are based on the qualifier and corresponding erosion classes shown in Table 14-5. Similar
rules were developed for the other qualifiers.

Table 14-5. Example of qualifier status and corresponding type and level of erosion used in rule
base development
Erosion
Qualifier Status
Type Level
Soil colour not changed wind/water none
become lighter on knolls and hilltops wind slight
significantly changed with white-topped knolls visible from across fields wind severe
become lighter over the mid slopes of fields and in localized areas water moderate
become lighter over most of field wind severe

Development of Diagnosis and Recommendations for Amelioration

The expert system compares the information obtained during a consultation session against the rule
base to assign a diagnostic conclusion (type and level of degradation) along with associated weighting
or certainty factor for each of the indicators identified during the consultation. The diagnostic
conclusions are ranked based on the summation of their weighting factor scores. The conclusion with
the highest score gives the most probable diagnosis. An example of a set of diagnostic conclusions and
associated certainty factors is presented in Table 14-6. This portion constitutes the diagnostic part of
the expert system.

Table 14-6. Example of diagnostic conclusions and cumulative certainty factors obtained during
one expert system consultation
Diagnostic Conclusion Certainty Factor
There is at least a moderate level of wind erosion on parts of the farm 319
There is at least a slight level of wind erosion on parts of the farm 226
There is at least a slight level of water erosion on parts of the farm 70

Once degradation levels have been determined, the impact of these processes are expressed in terms
of yield reductions for a range of crops, including wheat, barley, oats, canola and flax. Table 14-7
gives an example of expected yield reductions for the result displayed in Table 14-6.

Table 14-7. Examples of expected yield reductions for moderate levels of erosion
Crop Yield Reduction (%)
Wheat 25-50
Barley 20-40
Oats 10-20
Canola 25-50
Flax 25-40

14-11
Following yield reduction determinations, a choice of ameliorative cropping practices can be presented
based on the required management practices for each degradation level (Table 14-8). These are
assumed to be highly cost-effective since they have been identified by the innovative conservation
farmers. Subsequent to a choice of ameliorative practices, expected productivity improvements can be
determined. The impact of some management practices on crop productivity were presented earlier
(Tables 14-3 and 14-4).

Table 14-8. Ameliorative practices by type and level of degradation as identified by innovative
conservation farmers
WIND EROSION

Slight Moderate Severe

• maintain soil cover • maintain soil cover • maintain soil cover
• continuous cropping • continuous cropping • grassing entire field
• reduce summerfallow area • direct seeding • use soil structure improvement
• use chemical fallow • use chemical fallow strategies
• reduce field width • re-crop eroded fields • permanent forage
• increase amendment
applications • cropping to alfalfa
• increase fertilizer • reduce field width
application rates • limit field to <330 ft width
• increase phosphate • establish windbreaks
application in eroded areas • extensive shelterbelt
• reduce field width development
• limit field to 400-600 ft
width
• use soil structure improvement
strategies
• use longer crop rotations
• add forages to rotation
• establish windbreaks
• plant shelterbelts

Slight
• maintain soil cover &
improve
• soil structure
• continuous cropping
• direct seeding
• decrease summerfallow
• use chemical fallow
• manage waterways
• leaving natural waterways
• intact
• grassing eroded waterways
WATER EROSION
Moderate
• maintain soil cover & improve
soil
• structure
• continuous cropping
• direct seeding
• plant forages
• decrease/eliminate
summerfallow
• use chemical fallow
• manage waterways
• grass waterways
• stabilize rills and gullies
• use soil structure improvement •
strategies
• incorporate manure or straw
Severe
• maintain soil cover & improve
soil structure
• seed to forages or continuous
crop
• use longer crop rotations
• - increase fertilizer application
• rates in eroded areas
• manage waterways
• grass waterways
• stabilize rills and gullies
use soil structure improvement
• strategies
• incorporate manure or straw

14-12
 FUTURE CONSIDERATIONS

SOILCROP has been developed as a prototype and it still contains a number of limitations. Although
a broad knowledge base was obtained from innovative conservation farmers through the structured
questionnaire, there is still need for further knowledge acquisition, as identified during expert system
development. This would require a Monte Carlo type approach with a small focus group of innovative
conservation farmers, conservation officers and researchers, where iterative knowledge acquisition as
well as testing and modifications of the knowledge base can be conducted. Existing knowledge base
limitations and discrepancies would be identified as well as the strategy for further knowledge
acquisition. A further need is for inclusion of location-specific data by linking with GIS-based
databases and regional models such as LANDBASE, CanHELP, etc.

There is also a lack of sufficient information in the scientific literature that could be translated into
symbolic or analogical relationships for agricultural decision making and management purposes. This
limitation indicates the need for research initiatives that would link qualitative indicators and
quantitative soil attributes. This would tie in well with one of the recognized roles of expert systems,
namely, shifting focus of the research community to knowledge delivery (Jones, 1989). It would also
complement the traditional role of research as a vehicle for the accumulation of knowledge.

SOILCROP must also be linked to process models, particularly to enable predictions of future
performance. This becomes a critical issue for predicting sustainability. Process-based models such as
CENTURY and EPIC have shown great promise in terms of predicting trends in soil degradation -
crop yield relationships. SimPLE is another such model that has been specifically developed for
Western Canadian conditions (Greer et al., 1993), and has the added advantage of requiring a much
smaller level of data input as compared to the other models.

Future developments of SOILCROP will incorporate links to computerized data bases and geographic
information systems for input of location specific physical attributes as well as linkages to process
models for degradation and crop productivity determinations. In its links to process models, the role
of the expert system would be to translate on-farm information into parameters required for model
initialization as well as to translate model results from a management perspective.

REFERENCES

Batchelor, W.D., McClendon, R.W. and Adams, D.B. 1989. Evaluation of SMARTSOY: and expert
simulation system for insect pest management. Agric. Sys. 31:67-81.

Buick, R.D., Stone, N.D., Scheckler, R.K. and Roach, J.W. 1992. CROPS: a whole-farm crop
rotation planning system to implement sustainable agriculture. AI Applications. 6(3):2949.

Clarke, N.D., Tan, C.S. and Stone, J.A. 1989. Expert system for scheduling supplemental irrigation
in Ontario. ASAE Paper No. 89-7572. ASAE, St. Joseph, MI.

14-13
de Jong, E. 1988. Soil erosion. in Land Degradation and Conservation Tillage: Partial Proceedings,
34th Annual CSSS/AIC Meeting, August 21-24, 1988, Calgary, AB. pp 5769.

Doluschitz, R. and Schmisseur, W.E. 1988. Expert systems: applications to agriculture and farm
management. Comput. Electron. Agric. 2:173-182.

Greer, K.J., Schoenau, J.J., Anderson and Hilliard, C.R. 1993. Simulated productivity lost by
erosion (SIMPLE): model development, validation and use. Proc. Soils and Crops Workshop.
Saskatoon, SK. pp 138-148.

Hackett, C. 1988. Matching plants and land: Development of a general broadscale system from a crop
project for Papua New Guinea. Natural Resources Series No. 11. Division of Water and Land
Resources. CSIRO, Australia. 82pp.

Heatwole, C.D. 1987. Conservation planning using expert systems and geographic information
systems. ASAE Paper No. 87-5011. ASAE, St. Joseph, MI.

Huber, B., Nyrop, J.P., Wolf, W., Reissig, H., Agnello, A. and Kovach, J. 1990. Development of
a knowledge-based system supporting IPM decision making in apples. Comput. Electron. Agric.
4:315-331.

Ignizio, J.P. 1991. Introduction to Expert Systems: The Development and Implementation of
Rule-Based Expert Systems. McGraw Hill, New York, NY. 402pp.

Jones, P. 1989. Agricultural applications of expert system concepts. Agric. Sys. 31:3-18.

Kline, D.E., Bender, D.A. and McCarl, B.A. 1987. Farm-level machinery management using
intelligent decision support systems. ASAE Paper No. 87-1046. ASAE, St. Joseph, MI.

Lal, H., Jones, J.W., Peart, R.M. and Shoup, W.D. 1992. FARMSYS - a whole farm machinery
management decision support system. Agric. Sys. 38:257-273.

Larney, F.J. 1992. Productivity of artificially eroded soil: The Lethbridge "scalping" studies. Pages
51-56 in Soil conservation: many ways to make it work. Proc. Soil Conservation Workshop &
Alberta Conservation Tillage Soc. 14th Annual Meeting, January 13-15, 1992, Edmonton, AB.

Larney, F.J., Izaurralde, R.C., Janzen, H.H., Olson, B.M., Solberg, E.D., Lindwall, C.W. and
Nyborg, M. 1994. Soil erosion-crop productivity relationships for six Alberta soils. J. Soil Water
Cons. (in press).

McClure, J.E., Tomasko, M. and Collison, C. 1993. BEE AWARE, an expert system for honey
bee diseases, parasites, pest and predators. Comput. Electron. Agric. 9:111-122.

McKinion, J.M., Baker, D.N. Whisler, F.D. and Lambert, J.R. 1989. Application of
GOSSYM/COMAX system to cotton crop management. Agric. Sys. 31:55-65.

14-14
Montas, H. and Madramootoo, C.A. 1992. A decision support system for soil conservation planning.
Comput. Electron. Agric. 7:187-202.

Pawluk, R.R., Sandor, J.A. and Tabor, J.A. 1992. The role of indigenous soil knowledge in
agricultural development. J. Soil Water Cons. 47(4):298-302.

Plant, R.E. 1989. An artificial intelligence based method for scheduling crop management actions.
Agric. Sys. 31:127-155.

Plant, R.E. and Stone, N.D. 1991. Knowledge-Based Systems in Agriculture. McGraw Hill, New
York, NY. 342 pp.

14-15
CHAPTER 15
RELATIONSHIPS OF SIMULATED EROSION AND SOIL AMENDMENTS TO SOIL
PRODUCTIVITY

R.C. Izaurralde1, M. Nyborg1 , E.D. Solberg2, M.C. Quiroga Jakas1,

S.S. Malhi3, R.F. Grant1 and D.S. Chanasyk1
1
Department of Soil Science, 4-42 Earth Sciences Building, University of Alberta,
Edmonton, Alberta
2
Soil and Crop Management Branch, Alberta Agriculture, Food and Rural
Development, Edmonton, Alberta
3
Agriculture and Agri-Food Canada, Research Station, Lacombe, Alberta

INTRODUCTION

Soil erosion is the most important degradation process affecting land productivity in western Canada 1,2 .
Erosion effects on soil productivity are many, often interrelated, and difficult to quantify. Crop growth on
eroded land is strongly influenced by soil and topographic properties3,4,5,6,7 . Eroded soils affect plant yield
through their increased bulk density, poorer tilth, reduced organic matter content, low nutrient availability,
and reduced water-holding capacity 8,9,10,11,12,13 . Topsoil thickness, for example, has been an integrative soil
property used as an indicator of soil quality and productivity8.

Yield differences between eroded and non-eroded soils, however, are often masked by genotypes that
respond well to the application of fertilizers and herbicides14. Nitrogen and P fertilizers have been used to
restore the productivity of artificially exposed subsoil12,15,16 but their effectiveness has varied with soil type,
climate, crop, and level of management 3,16,17,18,19. However, even if crop yields under eroded-soil conditions
can be increased with the addition of N and P fertilizers, these may not realize the levels obtained under non-
or slightly-eroded conditions10,20,21.

Several studies have assessed erosion in Alberta 11,22,23,24,25 but these studies did not provide a direct method
for quantifying the effects of erosion on crop productivity. Quantification of erosion-productivity
relationships would allow for improved cost/benefit analysis of conservation tillage and erosion control
programs26,27. Three approaches have been used to quantify erosion effects on crop productivity28,29: (I)
assessment of past erosion effects on crop productivity7,30,31, (ii) effects of simulated-erosion on crop yield
by use of artificially-eroded plots15,32, and (iii) computer simulation of long-term effects of erosion on crop
productivity33.

This paper summarizes results from a two-year field study on artificially-eroded plots and a
computer-simulation study. Specific objectives for these studies were to: (i) quantify the relationship between
depth of erosion and productivity of two soils of central Alberta, (ii) assess the effectiveness of soil
amendments, and (iii) test how EPIC, a productivity-erosion computer model, predicted results from the field
experiments.

15-1
 MATERIALS AND METHODS

Artificial-Erosion Experiment

To address the first two objectives we conducted two field trials in 1991 and 1992 at Josephburg and Cooking
Lake, approximately 20 km east of Edmonton, Alberta. The soil at Josephburg is a Black Chernozem (Angus
Ridge series). with a 30-cm thick surface horizon. The soil at Cooking Lake is a Gray Luvisol (Cooking Lake
series) with a 15-cm thick surface horizon. The experimental treatments consisted of five erosion levels
created by removing topsoil in 5-cm depth increments from 0 to 20 cm. Within each erosion level we applied
four amendment treatments: (i) control (C), (ii) addition of 5 cm of topsoil (T), (iii) addition of fertilizer (F)
at rates of 100 kg ha-1 of N (banded) and 20 kg ha-1 of P (applied with seed), and (iv) cattle manure (M) at
a rate of 75 Mg ha-1 on a dry basis. To test residual effects of amendments in 1992, we applied a maintenance
dose of fertilizer to one-half of each plot (100 kg ha-1 as broadcast-incorporated urea and P at a rate of 20 kg
ha-1 as 0-45-0 applied with the seed).

All plots were sown to hard-red spring wheat. Plant variables measured included: plant density, growth stages,
height at maturity, ear density, above-ground biomass at flowering, grain yield, and kernel weight. A plot
combine was used to harvest the plots. Plant samples were analyzed for nitrogen and phosphorus.

In addition, we took soil samples from all plots from the following depths: 0-10, 10-20, 20-30, 30-60, and
60-90 cm. These soil samples were processed and analyzed for texture, field capacity, wilting point, pH,
exchangeable bases, extractable and total phosphorus, available and total nitrogen, and organic carbon. Bulk
density was determined in the field.

Computer Simulation Experiment

The Erosion Productivity Impact Calculator (EPIC)32 is a fully integrated model designed to simulate daily
agro-ecosystem processes and integrate the results over tens of years. Processes modelled include: wind
erosion, water erosion, runoff, plant growth, nutrient uptake, nitrate leaching, and changes in soil organic
matter. Data obtained from the field trials in 1991 at Josephburg and Cooking Lake were compared with those
simulated by EPIC. The treatments selected for the simulation runs were the five levels of simulated erosion
with the fertilizer and manure amendments and a control. Soil data used as input for the simulation runs were
obtained from soil samples analyzed for texture, bulk density, and nutrients. Values of field capacity and
wilting point were estimated by EPIC using a regression method based on clay, sand, silt, organic matter, and
bulk density. Input datasets for fertilizer and composted-cattle manure treatments were made by inserting
lines which described the addition of either commercial or organic fertilizers. EPIC can simulate various
organic fertilizers such as cattle, pig, and chicken manure. In addition, manure applications can be simulated
to be applied fresh, from scrapings, from pits, or composted. We first calculated the amount of organic and
inorganic N and P added in the manure to the experimental plots and proceeded to account for those
quantities in the input datasets of all manure treatments. The simulation runs used crop parameters as defined
in the EPIC crop files. Two of these, maximum leaf area index and fraction of growing season at which leaf
area starts to decline, were modified to better describe spring wheat growth in boreal climates.

15-2
 RESULTS

Artificial-Erosion Experiment

Soil Properties of Control Plots After Topsoil Removal

At Josephburg, the influence of artificial erosion on available soil water (ASW) to 90 cm depth was small.
Available soil water increased slightly from 147 mm on the 0-cm cut to 150 mm on the 10-cm cut. It then
decreased to 144 mm on the 20-cm cut. A similar trend was observed at Cooking Lake where ASW increased
from 140 to 150 mm in the first two cuts but then varied around 145 mm in deeper cuts. At both sites, soil bulk
densities measured in the upper 10 cm increased with depth of simulated erosion. All surface soil layers at
Cooking Lake had greater bulk densities than those at Josephburg. Chemical soil properties changed
substantially with depth of layer sampled and depth of simulated erosion. For example, concentration of total
soil carbon in the surface layer of the 20-cm cut at Josephburg was 60 % of the concentration present in the
normal profile. In contrast, soil carbon at Cooking Lake in the surface layer of the 20-cm cut was only 25%
of the concentration present in the normal profile. Available nitrogen in the first three depths of cut totalled
approximately 100 kg ha-1 at Josephburg but only 80-85 kg ha-1 in the deeper cuts. The non-eroded soil profile
at Josephburg had about 15 kg ha-1 less extractable P than the non-eroded soil at Cooking Lake. The decrease
in extractable P with depth of simulated erosion was more pronounced at Cooking Lake than at Josephburg.

Yield and Yield Components in 1991

Growing season precipitation in 1991 was 274 mm, slightly above the normal of 260 mm. The growing
season of 1991 was excellent. Both erosion levels and amendments affected grain yields (Table 15-1).
Average wheat yields on the non-eroded soil at both sites were similar but they decreased more sharply with
increasing erosion on the Gray Luvisol than on the Black Chernozem. Fertilizer was the best amendment to
restore the productivity of eroded soil, followed by manure and topsoil. On the Black Chernozem, yields
obtained on fertilized plots subject to 20 cm of simulated erosion (4.2 Mg ha-1) were greater than those
obtained on non-eroded control plots (3.0 Mg ha-1). On the Gray Luvisol, however, wheat yields obtained on
fertilized plots subject to the same level of artificial erosion (3,1 Mg ha-1) were only slightly greater than
those obtained on non-eroded control plots (2.9 Mg ha-1).

At Cooking Lake, good environmental conditions during plant emergence resulted in uniform plant densities
(Table 15-2). At Josephburg, plant density on the manure and fertilizer treatments was lower than on the
control treatment. The topsoil amendment had intermediate plant densities. Simulated erosion did not affect
final stands at either location. On average, wheat plants at Josephburg were 10 cm taller than at Cooking Lake.
Both depth of cut and amendments caused significant differences in plant height at both locations. Simulated
erosion, averaged across amendments, reduced tillering. It reduced spike density by 7.1, 24.7, 42.0, and
41.5 % at Cooking Lake by 3.2, 4.1, 16.5, and 23.2 % at Josephburg for the 5, 10, 15 and 20 cm cuts,
respectively. In spite of a plant density 7.4 % lower than the control treatment, the fertilizer treatment
increased tillering and hence spike density. At both locations, seeds obtained from the 0 and 5 cm cut plots
were heavier than those from the 15 and 20 cm cut plots. At Cooking Lake, the manure effects on seed
weight were similar to those of fertilizer and topsoil amendments. Any of these treatments produced seeds

15-3
that were heavier than those obtained from the control plots. A greater weight variability was observed at
Josephburg where manure produced the heaviest seeds. At both locations, inflorescence emergence and
anthesis were not reached evenly. However, growth-stage differences among erosion levels were minimal after
anthesis.

Table 15-1. Effects of artificial erosion and amendments on wheat grain yield (12% moisture) in 1991

Site (kg ha-1)

Treatment
Josephburg Cooking Lake
Cut (cm)
0 4314a † 4228a
5 4184a 3624b
10 3693b 3133c
15 3090c 2256d
20 2769c 1730e
Amendment
Control 2137d 1514d
Fertilizer 4854a 4169a
Manure 4477b 3655b
Topsoil 2972c 2640c
† Means within same treatment and site followed by the same letter are not significantly different at p < 0.05
using a protected Fisher's LSD test.

Yields in 1992

Fair growing conditions prevailed during 1992. At Josephburg, July precipitation was 62 mm, but only 15 mm
fell during the third and fourth weeks. Further, precipitation during the first two weeks of August was only
4 mm. A hail storm at Josephburg caused visible damage on wheat stands and reduced yields. The maximum
yield of the experiment (4.4 Mg ha-1) was obtained at Cooking Lake on non-eroded soil receiving manure in
1990 and a moderate application of fertilizer in 1992 (Table 15-3). Yields at Cooking Lake in 0-cm cut plots,
averaged across amendments, were about 20% greater than those at Josephburg. We surmise that greater
precipitation at Cooking Lake may have contributed to greater yields. This advantage at Cooking Lake was
maintained at the 5-cm cut in the manure-control and manure-fertilizer and topsoil-control but disappeared
in the other treatments and cuts. At Josephburg, the yield decline was drastic in cuts receiving no fertilizer
in 1992. For example, the 20-cm cut without fertilizer yielded 65 % less than the 0-cm cut (Table 15-3). The
same 20-cm cut treatment receiving fertilizer in 1992 yielded only 10 % less than the 0-cm cut. Overall,
manure provided the best residual effect at both sites. At Josephburg, fertilizers N and P as restorative
amendments had a similar effect to manure and topsoil (Table 15-3). The application of a maintenance level
of fertilizer in 1992 helped recover yields even under the most drastic case of simulated erosion.

15-4
Table 15-2. Effects of artificial erosion and amendments on wheat-growth parameters in 1991

Plant density Tiller number Height Spike density Grain weight

m-2 plant-1 cm m-2 g 1000 seed-1
Simulated erosion (cm) Josephburg
0 150a† 3.4a 104a 478a 35.1a
5 156a 3.0b 102a 472a 34.8a
10 165a 3.0b 99a 467a 34.3ab
15 167a 2.5c 90b 407b 33.3bc
20 158a 2.4c 81c 374b 32.6c
Amendment
Control 165a 2.2c 80c 347c 33.5bc
Fertilizer 155bc 3.7a 105a 546a 33.9b
Manure 153c 3.3b 106a 499b 35.4a
Topsoil 163ab 2.3c 90b 367c 33.2c
Simulated erosion (cm) Cooking Lake
0 165a 3.2a 97a 511a 35.6a
5 160a 2.9a 94a 475a 34.4ab
10 165a 2.3b 87b 385b 33.4bc
15 163a 1.9b 74c 296c 32.0cd
20 163a 1.8b 73c 299c 31.2d
Amendment
Control 170a 1.7c 68c 288d 32.4b
Fertilizer 157b 3.1a 95a 489a 33.5a
Manure 162ab 2.7b 97a 431b 34.1a
Topsoil 164ab 2.2c 80b 364c 33.4a
† Means within same treatment and variable name followed by the same letter are not significantly different
at p < 0.05 using a protected Fisher's LSD test.

Nutrient Concentration and Uptake in 1991

At both sites, amendments affected grain-N concentration more so than erosion. The range in grain N
concentration at Josephburg was 2.8 g kg-1 for level of erosion and 7.3 g kg-1 for amendments. The
corresponding ranges at Cooking Lake were 1.7 and 7.1 g kg-1, respectively. The decrease in grain-N
concentration, measured as the level of simulated erosion increased, was accompanied by decrease in total
N uptake. Fertilizer application at Josephburg produced the highest grain concentration and total uptake of
N. Grain N in manure- and topsoil-treated wheat was diluted with respect to wheat in control plots. At
Cooking Lake, wheat on fertilized plots took up the greatest amounts of N but had less grain-N concentration
than either control plots at the same site or fertilized plots at Josephburg.

15-5
In contrast, grain P changed much less than grain N across main treatments. The significant differences in total
P uptake measured among treatments at Cooking Lake are attributed to yield differences and not to limiting
levels of P in plant. There was some dilution effect in grain P at Josephburg possibly as a result of the high
yields obtained.

Computer Simulation of Wheat Yields

A fundamental hypothesis of this study was to test if EPIC could reproduce the observed yield and related
plant parameters given soil, climate, and landscape data as input. At Josephburg, computer-simulated yields
followed the trend of observed yields (Fig. 15-1). EPIC consistently over-estimated wheat yields in the control
treatment but under-predicted those obtained in plots receiving either commercial fertilizers or manure. EPIC,
however, correctly estimated the yield trends. At Cooking Lake, computer-simulated yields declined with level
of simulated erosion but could not follow the sharp yield decrease measured in the last two erosion increments
(Fig. 15-2). There was, however, a close agreement between observed and predicted yields during the first
three levels of erosion on amended plots. Yields were over-estimated by EPIC in the last two erosion levels
on the amended plots.

At Josephburg, EPIC attributed most of the yield reduction in control plots to water and N stress. The number
of days with water stress was greater than those with N stress during the first three levels of simulated erosion.
Nitrogen stress, however, controlled most of the yield reduction during the last two erosion levels. A different
pattern of plant stress was simulated at Cooking Lake where the number of N-stress days predicted rose from
14 to 37. This extreme N limitation, in turn, caused an actual reduction in water stress. At both sites, predicted
days with stress due to P, air temperature, lack of aeration, high bulk density, soil acidity, and low soil
temperatures were much less important than water and N stress. At Josephburg, the number of days with bulk
density causing stress on root growth rose slowly from 1.7 to 4.1, as the simulated-erosion level increased
from 0 to 20 cm. At Cooking Lake, these predictions stood steady at 2.4 days across the five erosion levels.

DISCUSSION

Research results from the Prairie Provinces 34,35,36 consistently showed an increase in crop-yield response to
applied N when extractable NO3-N decreased. Yield reductions due to soil loss can be also explained as a
function of nutrient losses. In this study, there was a close relationship between nutrient loss and yield
reduction. At both sites, the rate of yield reductions per unit of available N lost was a function of the mass
of available N lost. Erosion affects not only nutrient-pool sizes but also nutrient-release dynamics37. In this
study, simulated erosion curtailed not only the amounts of available N but also the soil's ability to release
it.

In a previous paper38 we described the general relationships between simulated erosion and yield for six
Alberta soils. Two of the soils were the ones at Josephburg and Cooking Lake. In five out of six cases,
including Cooking Lake, the simplest relationship that described yield (y) as a function of simulated erosion
(x) without any amendment added was: y = a - bx + cx2. This equation tells us that marginal yields diminish
with the incremental loss of topsoil; i.e., topsoil near the surface influences yield more than topsoil farther
from the surface. The exception was the equation for the soil at Josephburg that had also a negative c
coefficient.

15-6
Fig. 15-1. Observed and simulated spring wheat yields at Josephburg in 1991.

15-7
Fig. 15-2. Observed and simulated spring wheat yields at Cooking Lake in 1991.

15-8
At Josephburg, therefore, the topsoil near the surface had less value than deeper topsoil. These relationships
are presented in the top two graphs of Fig. 15-3.

Table 15-3. Effects of artificial erosion and amendments on wheat yield (12 % moisture) in 1992

Site
Treatment Josephburg Cooking Lake
-1
kg ha
Non-fertilized
Cut (cm)
0 2677 3191
5 2494 2443
10 1557 1633
15 1249 962
20 920 1026
Amendment
Control 1526 1268
Fertilizer 2075 1605
Manure 1981 2523
Topsoil 1748 1813
Fertilized
Cut (cm)
0 3506bc† 3721a
5 3895a 3407b
10 3633ab 2874c
15 3199c 2569d
20 3146c 2393d
Amendment
Control 3314b 2539c
Fertilizer 3260b 2628c
Manure 3870a 3841a
Topsoil 3460b 2965b

† Means within same treatment and variable name followed by the same letter are not significantly different
at p < 0.05 using a protected Fisher's LSD test.

What then, if anything, can be done to restore the productivity lost due to simulated erosion? The application
of fertilizer N and P compensated the loss of yield even at the greatest level of simulated erosion. The
application of cattle manure helped improve yield at Josephburg and, at the same time, reduce the impact of
yield loss due to erosion. At Cooking Lake, yields improved substantially up to 10 cm of simulated erosion
with addition of cattle manure. From there onwards, yields decreased very rapidly. In fact, the rate of yield
loss was 5.6 times greater at Cooking Lake than at Josephburg. The return of 5 cm of topsoil to the plots made
the rate of yield loss, at both sites, an independent function of erosion (Fig. 15-3). It appears that added topsoil
was largely responsible for the plant productivity realized at that site.

15-9
Fig. 15-3. Effects of depth of simulated erosion and restorative amendments on
wheat yields in 1991.

15-10
Results obtained in 1992 generally supported the findings of 1991. They confirmed that soil productivity was
a function of topsoil depth, which in turn was strongly a function of soil nutrients. The application of a
maintenance level of fertilizer N at Josephburg not only restored yields but also, in the case of 20-cm cut,
offset control yields of non-eroded soil by 18 % (Table 15-3). These results were not repeated at Cooking
Lake where the soil has a much thinner topsoil. Wheat on 5-, 10-, 15-, and 20-cm cuts, receiving a
maintenance level of fertilizer, yielded 107, 90, 81, and 75 %, respectively, of wheat grown on 0-cm cut
without maintenance fertilizer. These results at Cooking Lake compared closely with others reported for
southern Alberta.11

As in other studies11,12,39 the yield-soil depth relationships of this study attributed substantial weight to
fertility-chemical soil properties as means of explaining yield responses to simulated erosion (Fig. 15-2).
Other factors, however, such as increased bulk density and compaction can reduce root penetration and soil
aeration40 and therefore reduce crop yields.12 Results from a root study conducted on these plots in 199141
indicated that root growth was not affected by levels of simulated erosion but improved with addition of
amendments such as fertilizer and manure. Apparently, wheat plants responded to the highest level of
simulated-erosion imposed (20-cm cut) by maintaining the amounts of root mass but decreasing the
accumulation of above-ground biomass (e.g., root/shoot ratios under simulated erosion increased). Bulk
densities measured in the upper 10-cm were not sufficiently high enough so as to reduce root growth. We
did not observe soil compaction effects on germination and plant establishment. Reduced yield on eroded
soil arose from low-tillering wheat plants with reduced leaf areas and lighter seeds rather than from low
plant densities. Fertilizer increased yield by increasing tiller density, spike density, and leaf area. Whereas,
cattle manure increased yield by increasing tiller density, spike density, leaf area, and seed weight.

Results of the simulation runs showed that EPIC predicted grain yield, above-ground biomass, and grain N
at both sites within levels of simulated erosion ranging from 0 to 10 cm. At Cooking Lake, the model did not
predict well the effects of erosion on crop productivity for levels greater than 10-cm of topsoil removal. EPIC
accounted for the effects of commercial fertilizers and manure in restoring lost productivity and detected
the slight advantage of commercial fertilizer over cattle manure in restoring this productivity. EPIC appeared
to attribute too much weight to the active nitrogen fraction of severely-eroded soil at Cooking Lake resulting,
in turn, in over-predictions of plant biomass and grain N. We suggest that revisions in the plant growth and
nitrogen algorithms could improve the model's performance for climatic-soil conditions such as those in the
Canadian prairies.

This study showed that soil productivity of artificially-eroded soil was mostly a function of nutrient removal.
Lost productivity was partially restored by using either fertilizer or cattle manure. Several questions however
remain: (i) what would be the optimum fertilizer recommendation rates that could meet yield expectations
and ensure, at the same time, environmental protection (e.g., nitrate leaching)?; (ii) at what rates should
organic amendments be applied?; (iii) what would be the best recommendable sources of organic
amendments?; and (iv) how would soil quality evolve over time on severely-eroded soils subject to an
specific management?

15-11
 CONCLUSIONS

We conclude that: (i) the incremental loss in productivity was a function of the depth of soil removed, (ii)
simulated erosion reduced yields more at Cooking Lake, with a 15-cm thick surface horizon, than at
Josephburg, with a 30-cm thick surface horizon, (iii) yield reductions were associated with the amounts of
available and total nitrogen and phosphorus removed, (iv) in 1991, the order of amendment effectiveness
was: fertilizer>manure>topsoil, and (v) in 1992, the residual effectiveness of amendments was:
manure>topsoil fertilizer.

With regards to the testing of the simulation model EPIC we conclude that: (i) it satisfactorily predicted, at
both sites, grain yield, above-ground biomass, and grain N within levels of simulated erosion that ranged
from 0 to 10 cm, (iii) it did not predict well at Cooking Lake erosion effects on crop productivity for levels
greater than 10-cm, (ii) it accounted for the effects of commercial fertilizers and manure in restoring lost
productivity, and (iii) it detected the slight advantage of commercial fertilizer over cattle manure in restoring
lost productivity. Revisions in plant growth and nitrogen algorithms could improve the model's performance
for climatic-soil conditions such as those prevailing in the Canadian prairies.

ACKNOWLEDGMENTS

We thank Z. Zhang, M. Molina, B. Hoar, J. DeMulder, J. Brown, and C. Nguyen for technical assistance
throughout this study. We gratefully acknowledge Mr. Doug Gabert and Mrs. Judy Gabert of Josephburg
and Mr. Cliff Horricks of Cooking Lake for allowing us to conduct this project on a portion of their farm.
Financial support for this study was provided by the National Soil Conservation Program of Canada.

REFERENCES

1. Coote, D.R. 1984. The extent of soil erosion in western Canada. Pages 34-38 in Soil erosion and land
degradation. Proc. 2nd Ann. Western Provincial Conf. Rationalization of Water and Soil Research and
Management. Saskatchewan Inst. of Pedology, Saskatoon, SK.

2. Sparrow, H.O. 1984. Soil at risk: Canada's eroding future. Report of the Standing Committee on
Agriculture, Fisheries, and Forestry. Supply and Services Canada, Ottawa, ON.

3. Langdale, G.W., and Schrader, W.D. 1982. Soil erosion effects on soil productivity. Pages 41-51 in
B.L. Schmidt et al., Eds. Determinants of soil loss tolerance. Spec. Publ. 45. Am. Soc. Agron.,
Madison, WI.

4. Whitman, C.E., Hatfield, J.L., and Reginato, R.J. 1985. Effect of slope position on the microclimate,
growth, and yield of barley. Agron. J. 77:663-669.

5. Daniels, R.B., Gilliam, J.W., Cassel, D.K., and Nelson, L.A. 1985. Soil erosion class and landscape
position in the North Carolina Piedmont. Soil Sci. Soc. Am. J. 49:991-995.

15-12
6. Schertz, D.L., Moldenhauer, W.C., Franzmeier, D.P., and Sinclair, H.R. 1985a. Field evaluation
of the effect of soil erosion on crop productivity. Pages 9-17 in Erosion and soil productivity. ASAE
Publ. 8-85 Am Soc. Agr. Eng., St. Joseph, MI.

7. Schertz, D.L., Moldenhauer, W.C., Livingston, S.J., Weesies, G.A., and Hintz, E.A. 1985b. Effect
of past soil erosion on crop productivity in Indiana. J. Soil Water Conserv. 44:605-608.

8. Power, J.F., Sandoval, F.M., Ries, R.E., and Merrill, S.D. 1981. Effects of topsoil and subsoil
thickness on soil water content and crop production on a disturbed soil. Soil Sci. Soc. Am. J.
45:124-129.

9. Eck, H.V., Hauser, V.L., and Ford, R.H. 1965. Fertilizer needs for restoring productivity on Pullman
silt loam after various degrees of soil removal. Soil Sci. Soc. Am. Proc. 29:209-213.

10. Frye, W.W., Ebelhar, S.A., Murdock, L.W., and Blevins, R.L. 1982. Soil erosion effects on
properties and productivity of two Kentucky soils. Soil Sci. Soc. Am. J. 46:1051-1055.

11. Dormaar, J.F., Lindwall, C.W., and Kozub, G.C. 1986. Restoring productivity to an artificially
eroded Dark Brown Chernozemic soil under dryland conditions. Can. J. Soil Sci, 66:273-285.

12. Tanaka, D.L., and Aase, J.K. 1989. Influence of topsoil removal and fertilizer application on spring
wheat yields. Soil Sci. Soc. Am. J. 53:228-232.

13. Bauer, Armand, and Black, A.L. 1992. Organic carbon effects on available water capacity of three
soil textural groups. Soil Sci. Soc. Am. J. 56:248-254.

14. Krauss, H.A., and Allmaras, R.R. 1982. Technology masks the effects of soil erosion on wheat yields
- a case study in Whiteman County, Washington. Pages 75-86 in B.L. Schmidt et al., Eds. Determinants
of Soil Loss Tolerance. Spec. Publ. 45. Am. Soc. Agron., Madison, WI.

15. Engelstad, O.P. and Shrader, W.D. 1961. The effect of surface soil thickness on corn yields. II. As
determined by an experiment using normal surface soil and artificially exposed subsoil. Soil Sci. Soc.
Am. Proc. 25:497-499.

16. Morrison, R., and Shaykewick, C.F. 1987. Effect of simulated soil erosion on wheat yields on the
humid Canadian prairies. J. Soil Water Cons. 42:205-208.

17. Eck, H.V. 1968. Effect of topsoil removal on nitrogen supplying ability of Pullman silty clay loam. Soil
Sci. Soc. Am. Proc. 32:686-691.

18. Eck, H.V. 1969. Restoring productivity of Pullman silty clay loam subsoil under limited moisture. Soil
Sci. Soc. Am. Proc. 33:578-581.

15-13
19. Arce-Diaz, E., Featherstone, A.M., Williams, J.R., and Tanaka, D.L. 1993. Substitutability of
fertilizer and rainfall for erosion in spring wheat production. J. Prod. Agric. 6:72-76.

20. Malhi, S.S., Nyborg, M., Penney, D.C., Kryzanowski, L., Robertson, J.A., and Walker, D.R. 1992.
Yield response of barley and rapeseed to P fertilization as influenced by soil test P level and method
of placement. Comm. Soil Sci. Plant Anal. 24:1-10.

21. Mielke, L.N., and Schepers, J.S. 1986. Plant response to topsoil thickness on uneroded loess soil. J.
Soil Water Cons. 41:59-63.

22. Toogood, J.A. 1963. Water erosion in Alberta. J. Soil Water Conserv. 6:238-240.

23. Chanasyk, D.S., and Woytowich, C.P. 1987. A study of water erosion in the Peace River Region.
Farming for the Future Project No. 83-0145. Dep. Soil Sci., Univ. of Alberta. Edmonton, AB.

24. Van Vliet, L.J.P., and Hall, J.W. 1992. Effects of two crop rotations on seasonal runoff and soil loss
in the Peace River Region. Can. J. Soil Sci. 71:533-543.

25. Howitt, R.W. 1991. Measuring the characteristics of soil erosion on agricultural landscapes in
East-Central Alberta. Ph.D. Diss. Dep. of Soil Sci., Univ. of Alberta. Edmonton, AB.

26. Van Kooten, G.C., Weisnel, W.P., and de Jong, E. 1989. Estimating the costs of soil erosion in
Saskatchewan. Can. J. Soil Sci. 37:63-75.

27. Smith, E.G., and Shaykewich, C.F. 1990. The economics of soil erosion and conservation of six soil
groupings in Manitoba. Can. J. Agr. Econ. 38:215-231.

28. Meyer, L.D., Bauer, A., and Heil, R.D. 1985. Experimental approaches for quantifying the effect of
soil erosion on productivity. Pages 213-234 in R.F. Follett and B.A Stewart, Eds. Soil erosion and
crop productivity. Am. Soc. Agron., Madison, WI.

29. Lal, R. 1988. Monitoring soil erosion's impact on crop productivity. Pages 187-200 in R. Lal, Ed. Soil
Erosion Research Methods. Soil and Water Conserv. Soc., Ankeny, IA.

30. McDaniel, T.A., and Hajek, B.F. 1985. Soil erosion effects on crop productivity and soil properties
in Alabama. Pages 48-58 in Erosion and soil productivity. ASAE Publ. 8-85 Am Soc. Agr. Eng., St.
Joseph, MI.

31. Daniels, R.B., Gilliam, J.W., Cassel, D.K., and Nelson, L.A. 1987. Quantifying the effects of past
soil erosion on present soil productivity. J. Soil Water Conserv. 42:183187.

32. Mbagwu, J.S.C., Lal, R., and Scott, T.W. 1984. Effects of desurfacing of Alfisols and Ultisols in
southern Nigeria: I. Crop performance. Soil Sci. Soc. Am. J. 48:828-833.

33. Williams, J.R., Renard, K.G., and Dyke, P.T. 1983. EPIC a new method for assessing erosion's effect
on soil productivity. J. Soil Water Conserv. 38:381-383.

15-14
34. Soper, R.J., Racz, G.J., and Fehr, P.I. 1971. Nitrate nitrogen in the soil as a means of predicting the
fertilizer nitrogen requirements of barley. Can. J. Soil Sci. 51:45-49.

35. Carson J.A., Harapiak, J.T., Hennig, A.M.F., Janke, W.E., Nyborg, M., Penney, D.C., and
Walker, D.R. 1974. Use of soil testing to predict nitrogen fertilizer needs of barley and rapeseed in
Alberta. Pages 74-82 in Proc. Soil Fertility Workshop, Saskatoon, SK (available from Extension
Division, University of Saskatchewan, Saskatoon, SK).

36. Nyborg, M., and Malhi, S.S. 1990. Nitrogen requirements for the most economical yield of barley as
influenced by nitrate-N level in soil. Pages 651-652 in Abstracts Vol. IV, 14th International Congress
of Soil Science, 12-18 August 1990. Kyoto, Japan.
37. Bauer, Armand, and Black, A.L. 1992. Quantification of the effect of soil organic matter content on
soil productivity. Soil Sci. Soc. Am. J. 58:185-193.

38. Larney, F.J., Izaurralde, R.C., Janzen, H.H., Olson, B.M., Solberg, E.D., Lindwall, C.W., and
Nyborg, M. 1993b. Soil erosion-crop productivity relationships for six Alberta soils (in press).

39. Verity, G.E., and Anderson, D.W. 1990. Soil erosion effects on soil quality and yield. Can. J. Soil
Sci. 70:471-484.

40. Sands, R., Greacen, E.L., and Gerard, C.J. 1979. Compaction of sandy soils in Radiata pine forests.
I. A penetrometer study. Aust. J. Soil Res. 17:101-113.

41. Izaurralde, R.C., Nyborg, M., Chanasyk, D.S., and Solberg, E.D. 1992. Effects of artificial erosion
and amendments on soil and crop productivity. Environmental Sustainability Initiative
(PFRA-Agriculture Canada). Univ. of Alberta, Edmonton, AB. 15 pp.

15-15