131. .

^'I

I" G I, I - yj ^ ,, I 1:1 ..; I

USER S GU DE

::I IFi

I ^I' ''4;"I I

';'I "11 U ' ' ' I;' 11

,,

11 '^I

t'i
,

: IIJ I I

. ,1~" * *

1:11 pill'I

ITT0 I 10FC/I

I, t?,: I. I

International Tropical Timber Organization

Japan Overseas Forestry Consultants Association

J O ^'C I^*,
ITTO

Contents
Introduction

A Background of Forest Canopy Density Mapping Model

B Development of the Semi-expert System for the FCD Model
C Principal Features of the Semi-expert System

I . FCD Mapping Methodology

I- I

I. , Space Remote Sensing for Forest Management

I- I

I. 2 Concept of the FCD Mapping Model and Semi-Expert System

I-29

I. 3 Algorithms used in the FCD Mapping Procedures

1-40

2. How to Use the FCD Mapper

2.1 Introduction to the FCD Mapper
2.1. , The calculation procedure of FCD Mapper
2.12 The menu list of FCD-Mapper
A. The pull down menu of the main window
B. FCD Menu

2. I
2. I
2- I
2- 7
2- 7
2- 8

2.1.3 File Specification

A. Expert Sheet
B. Project Sheet
C. linage Format File
2. t. 4 Operating environment

2- 9

2.1.51nstallation Guide

2-15

2- 9
2-10
2.11
2-14

f
I
.
".

2.2 FCD Mapper Operations Guide

2.2. , Start-up and Registration Process
A. How to Start the FCD Mapper
B. Registration of the expert file
C. Setting up of the Project Name
2.2.21mage Input Process
A. How to Input Satellite linage Data

B. Selection of linage Area to be Analyzed

2.2.3 Noise Reduction and Niormalization Process

2-20

2-20
2-20
2.22
2-23
2.27
2.27

2-33
2-35

A. Procedures for masking bqdies of water

2-37

B. Procedures for masking clouds

2-39

C. Procedures for removing (masking) the cloud shadow area

2.41

~I

2.2.41ndex Process

2.43

A. Designation of the Vegetation Index (\/I)

B. Calculation of the Thermal Index CTl)
C. Calculation of the Shadow Index (SI)
2.2.5 Vegetation Density GID) Process
A. Designation of the vegetation area
B. Designation of the Bare Soil
C. Designation of the range which calculates the VD
D. Calculation of the VD

2.43
2.45

2-47
2-55
2-55
2-57
2-59
2-61

2-63
2.2.6 Scaled Shadow Index (SSl) Process
A. Confirming the conditions for processing SSl from Vl-Bl-SI false 2-63
color display
2-65
B. Designation of the forest area

C. Calculation of SSl

2-67

2-69
2.2.7 FCD Process and Map Statistics
2-69
A. Producing the FCD Map
B. Accounting the area in each FCD category and saving the FCD Map 2-71
2-75
2.3 FCD-Mapper File Menu
2-75
2.3. , Close of the pro^C to be processing
2.3.2 Changing the expert file, the name of the person conducting 2.77
the analysis, and comments on the project
2.3.3 Adding to or deleting from the project file, the expert file, or the 2-79
name of the person conducting the analysis
2.79
A. Deleting the name of a person in conducting the analysis
2-79
B. Adding and Deleting of the project file
2.79
C. Adding to and Deleting of the expert file
2-81
2.3.4 False color display of the processing image
2-81
2.3.5 Pseudo color display of the processing image

2.41mage display window operation

A. False Setting
B. Pseudo Setting
C. Save Bitmap
D. Printer Setting
E. Print

2-83
2-85
2.85
2.87
2-87
2-87

F. Zoom
.

2-87

G. Histogram

2-87

H . G rid

2-88

I. Scale

2-88

Appendix

2-89

Introduction
A.

Background of Forest Canopy Density Mapping Model

As generally applied in forestry, conventional R'S methodology is based on

statistical analysis of information derived from "training areas" (i. e.
ground-truthing). This has certain disadvantages in terms of the time and
cost required for training area establishment, and the accuracy of resints
obtained. In response to these problems, new methodology was developed
duning ITTO Project FD 32193 Rev. 2 (F), "Rehabilitation of Logged-over
Forests in Asia-Pacific Region, Sub-project 111", In this new methodology,

forest status is assessed OIL the basis of canopy density. The methodology is
.
\

presently identified as the Forest Canopy Density Mapping Model, or for

short the FCD Model. Unlike the convention. al statistical method, the FCD
Model indicates the growth phenomena of forests. The degree of forest
density is expressed in percentages: i. e. 10% FCD; 20%; 30%; 40% and so on.
FCD data indicates the intensity of rehabilitation treatment that may be
required. The method also makes it possible to monitor transformation of
forest conditions over time including degradation. . Additionally, it can
assess the progress of reforestation activities.

Field tests were conducted in both tropical (evergreen) and monsoon

(deciduous) forests to compare results derived using the FCD Model with
data obtained from conventional forest inventory methods. Correlation of
ground-based assessment data and results of RS"based assessment verified
an acctiracsi of more or less Din. ety percent (90%) using the FCD Model.
.

\
*

Among others, the methodology makes it possible to differentiate between

vegetation in the canopy and vegetation on the ground (e. g. grass)* Because
the data is in digital format, tables quantif^ing the rillm. ber of hectares in

each canopy density category from 10% and up, are prepared and prtn. ted o11t
with ease. In the preparation of national land use and management plans,
current forest stattis can be determined without incurring the costs and time
required to establish training areas, Additionally, imagery produced by the
FCD Model conveys a clear picture of forest conditions which is easy for
.

foresters, non-foresters and the .general public to understand.

B. Development of the Semi-expert System for the FCD Model

At present, RS assessment of forest conditions is a task reserved for
personnel who have received special training in remote sensing analysis.
Flirtherm. ore, the images derived from conventional RS analysis have rarely

been simple enough to lie readily understood by practicing forest ex, s,

planning officers and other concerned individuals.
As a follow-on phase of the previoiis project, the Japan Overseas Forestry
Consultants Association (JOFCA) has been implementing ITTO Project, FD
13/97 Rev. I (F), and developed the Semi-expert System to make the new
methodology more available to persons who are not RS experts. The system
is now available on CD-ROM format.

C. Principal Features of the Semi-expert System

The Semi"expert System, or simply the FCD Mapper, is a coinptiter
software package compatible with Windows-type personal computers. The
FCD Mapper contains the algorithms and other formulas utilized to compute
vanies of the several indices contained in the FCD Model for the analysis of
satellite imagery data.
The system will upgrade the planning capacity of decision-makers in forest
management and increase the information collecting capability of foresters,
while reducing costs and saving time.
ITSe of the Semi-expert system will facilitate production of "user-friendly"
satellite imagery that accurately portrays forest conditions. Access to such
imagery will increase the capability of foresters to provide acctirate,
11nbiased data on the status of forests in a format that is easy to understand.

, . FCD Mapping Methodology

I. , Space Remote Sensing for Forest Management

I-I

Space Remote Sensing for Forest Management

P. SRoy
hzdi@, 1113stit"te of Remote Sensing
Ovation@JRemote Sensing, 4ge, Icy)
Deftr"dun
e-mail:

earn

air

I&

I. Introduction.

Adopting sustainable development strategies is no more an option, but an

imperative necessity for achieving economic and food security for humanity. This has
been central theme of agreement of earth Sunnnit in Rio-de"Janeiro in 1992 popularly
called as Agenda 21. The scientific and technological I:nowledge should contribute to
achieve it. While the spectacular developments in space technology during the last
four decades have very forcefully revealed the concept and nature of the single fragile
global village. The survival of our planet as a whole is dependent on promoting a
reasonable quality of life on a sustainable basis in harmony with nature and its
environment. Satellite Remote sensing play a crucial role in determining, enhancing
and monitoring the overall carrying capacity. The repetitive satellite remote sensing
over various spatial and temporal scales offers the most economic means of assessing
the environmental parameters and impact of the developmental processes, The
repetitive nature of spatial data provides excellent capability to monitor ecosystem
composition, impact of management and degradation processes. The integrated
information derived from the satellite data set assists in evolution of appropriate
action plans for initiating sustainable development 0<a0,1996).
The arithropogenic interventions in the natural forest reduces the number of
trees per unit area and canopy closure, which in torn affect the regeneration of the
plant species and leading to uneven age class distribution and invasion of weeds.
Iudicious management through appropriate prescription for improving the forest stock
of any area requires the identification of various vegetation types and status at real
time. Therefore knowledge of forest types along with the density classes is necessary
for scientific forest management, Satellite remote sensing provides a mean to obtain a
synoptic view of forest and their condition on real time basis (Lines end and Kiefer
1987). Satellite remote sensing has played a pivotal role in generating infomiation
about forest cover, vegetation type and landuse changes (^IOUgton and Woodwell
1981; Botkin at at. 1984; Malingf6au 1991; Roy 1993). The standardization of ground
sampling methods, understanding of spectral and temporal responses of vegetation,

coupled with the recent advancements in the digital image processing techniques have
brought about a profound acceptance of the application of satellite remote sensing
data in forest inventory and mapping. Conservation of forest ecosystem require
understanding of the biospherical processes, understanding of their spatial
internnkages and continuous monitoring of the human interventions, in natural

I-2

resource management all the issues are interrelated in any environmental settin .
Hence, to achieve sustainable forest management and environmental rotection an
integrated approach is required. The spatial information hiked with non-s atial data

makes a proper assessment & monitoring of forest environment. The paper resents

the scope of applications of remote sensing in the field of forest and environment

management and attempts to identify future prospects.

2. Forest Managemiemt
Forests are known to be one of the most important renewable natural resource.

However due socio-economic and SOCio-political reasons developing world of SouthEast Asia has faced one of most serious degradation of forests in the tro ical world.

Present, world average of forest per person is 7000~sq in as ag^inst Indian avera e of

700 sq, rut.

The forest resources and forestlands need to be managed in such a wa that

they meet the socio-economic and forest produce needs of not only of this oneration
but for future too. In this context, forest management strategy in india should be based
on very reliable database on forest area, their productive potential and amenoration
sites. The detailed forest cover type maps are often required for Ia in of individual

felling coupes, PIaiming of roads, fire lines, wildlife habitat management etc. Hence a
quick repetitive and accurate information about forest cover is required at the local,
regional state and national levels for various purposes. Satellite remote sensing and
Geographic information System (GIS) have demonstrated that together they have

potential to provide comprehensive information on various facets of forest

management in India (Fig I). The experiences and methodologies to address the user
needs are described, It is also attempted to summarise future a Iications ' and
technological trends.

2.1. Forest density depletion process

,
.
.

Conventional forest management with the objective of maximisin

coriumercial returns is no longer adequate to cope with the challanges of forestry in the
developing world particularly Asia Pacific region. The social interface and forest

resource utilisation pattern has given rise to three zones viz. forest, fringe and nonforest, The first zone (forest) is a result of natural forces while second and third zones

(fringe and non-forest) are primarily man-made systems Fig. 2,

Due to the continuous interaction between the forest and non-forest s stems,

there is an outward flow of resQurces from the forest towards the man-made (nori
forest) systems tlirough the fringe Fig. 2. This outward flow is characterised by

fragmentation of the forest and. its fringe thereby resulting in low densit of the
primary forests. The corrective measures that can reverse this shift vary accordin to
the prevailing forest, non-forest and fringe situations. Tile innnediate challan e in

forest rehabilitation is to bring about a stabilisation of the boundaries b elmaricin

the productivity of the fringe,

I-3

Forest types stratification based on

Satellite Remote Sensing provides. ..
*
$5, ,'~--', -* * ."-.* *,.-, - ~J**,*$$;,;F:. rta**~ **;-*;:..,,$
:$*'-?,*.",,, 14*,,~,*^$*".." , .* * . *,. *

A- *;*r" -';:.' ~,. ~**., * ~ .* . ***. ::-*\~ ' *."*' ,-;'*" ';***
*, tt *,'~~ 3:.,.'*^ .-L ,",~~ , _-:'. ,*.*.,* CS, *r;*'\ :!.,*,!,*^'r. *F
.; .

**-. e^*^** ,+~-,.**s*~^-;~,;,;*t *.,

. ..

*;****'. ,*~ 4.4 I*'%'*' "; *$-*\*$;;:**'***?

*+:^^I?66^^.^ti^6i^;$^.#;^^**:
$4^**a*.*, P*^;:*.;**;.;?
..

..

,...

* *^esS^,
91i$^^','*:*';^*^
,,,
^^, ss^!$^$^43<*-^lit4*,,

;^-$;$^^*$*
eww:*.*;,*^t*;.".*^*\^^

I-. ;~;;,**,..'~ 91^.XI;;^;*'~ - "$;

Fig. I. F, Icets of Remote sensing for multiple levelofi, !formation
needed in forestry activities

Very few attempts have been reported to stratify the forest density using satellite
remote sensing digital data. So far forest density mapping has been carried out on the
basis of tonal variations within a scene by most of the workers (Madhavanunni et al.

1991). However with the increase in spectral and spatial resolutions volume of data
has increased considerably. With this, there is also an increase in the information
content and 'mixed pixels', which poses challange to the existing classification

techniq^^s of forest stratification (MCUoy 1980).

3. Forest Cover Mappimg and Monitoring

Satellite remote sensing .data has been extensively used to map forests of

tropics where up to date data about spatial distribution are absent or lacking. in Southeast Asia region at national level attempts have been made to map forests on I: I
million or 1:250,000 scale ' using interpretation of satellite data, The techniques are
operational in many of the countries and are being used extensively.

1-4

\
A

"

'* wi"
"'*-*"'"""";;BORES";' ' "' "' ' ' ' ' "' ' """' ' ' "" """'*' ' ' ' "' ' ' *""* * ""' ' * "E', 13:1N GE ' ' ' ' ' ' ' ' ' ' ' ' '
, - , - - - - ,- - - - - - -.- - -.- -.-. -.- - , .- , .- - -.- - - - , .-. , ,, . .- .- . .- -.- -.- -.-.-.- , -.- -.-.- -.- . , .- . ~ . , .

it:;*;;;;;:;:NON;;F'OREST

.
.

!;^;, rest 2385@?\$?e. Z>gp!atit}ra Proc^36.83

,

. Demographic Pressure . Demographic Pressure

. Social/Gender unequality . Urbanisation
. Unemployment
, Depletion of arable land
, Pollution
. Low Agricultural
Production

. Lack of Opportunities

( land, water, air )

, Agriculture -Low income

for economic return

Rus', It^"t
. Resource

Depletion
. Boundary
Shifting

Direction

It^-

jilt\a'

464 ' , ,

':tit'<

"-,*-

ant:*

. increased

Impact

Fig. 2

\.

3.1. Forest Density

Forest density expressing the stocking status constitutes the single major stand

physiognomic characteristic of the forest. The canopy closure, number of trees per
unit area and basal area are often taken as parameters to measure density. Satellite
remote sensing provides a measure of the canopy closure. The percent crown closure

is a measure of area occupation rather than stand density. However it is an important

parameter used to stratify the forests. The ecological conditions like, jiglit penetration
through the canopy, surface reflection and rainfall interception are dependent on the
crown density. Forest stock maps are the basic data used by the forest divisions in

preparing a working plan for forest management. It is prepared on a topographical

map of a large scale (1:15,750 approx. ) by various state forest departments showing
the local forest types with conventional insthod of in akirig stock maps using ground
surveys. in india reliable spatial database on the forest cover, its structure and

function are discrete. Forest stock maps, showing local forest types with their crown
density, site quality and age group of the crop, are the base data used by the forest
division for preparing a working plan. However conventional method of making
stock maps using ground survey is time consuming. It is therefore, difficult to
I-5

complete the revision in scheduled time, As a result most of the stock maps do not
reflect current status of forests.

3.2 Forest Stock Mapping

Stock maps depict forest type, density, encroachments cultivation patches

regeneration status and an idea about available resources. Conventional ground

surveys are time consuming and strenuous* They are also limited by the difficulty in

systematic traversing, boundary delineation etc. Visual or digital classification

interpretation methodology to update stock maps from high resolution satellite data
like ERS-IC PAN and now 11^Nos have shown possibility of achieving more
infonnative stock maps than conventional althougli the latter data sets provide more
details.
.

Many of the countries have starred using satellite remote sensing data for

mapping on 1:50 000 scale and overlaying compartment boundaries to obtain stocking
infonnation. This approach needs further consolidation by incorporating digital
approaches in data enhancements or classification. in intensive forest management
areas this mapping should be done on 1:25 000 scale using ERS LISS 111 and PAN
hybrid images,
3.3

Conventional Methods for Forest Density Estimation

33.1. Measurement with instruments

Various instruments are used for the estimation of forest canopy density.
However all these instruments have the disadvantage as the density estimation is
having certairi bias because of the instrmnent, The 'moosehom' is a simple
instrument projecting a dot grid on the canopy. The instrument produces canopy
density estimates which are not significantly biased. Another instrument is the
spherical densiometer, which employs a concave or convex mirror in which canopy is
reflected. The coptometer is also based on the 'same principle for canopy density
estimation. The estimation of canopy density by this method is time consuming and
is expensive, Also vertical stratification in the forest does create problem in
measurement of the canopy closure using graded mirror technique, specially in the
situation when bamboo occupies the understorey.
3.3.2 Qualitative Occular Method
Qualitative occular method is most commonly used method for canopy density

estimation within a specified raftge. in this method the observer takes a vertical

traverse and determines the canopy density depending on the presence of crown width
in a sample plot. With o00ular estimate of the density, a tendency of under estimation
exists because gaps in the canopy are overestimated rather than underestimated as
observed in the tropical rain forests.

I-6

3.3.3 Application of Angle Count Method

The instrument used consists of a sighting rod and a traverse stick which is

exactly two thirds of the length of the sighting rod and defines the limiting angle or.
The number of tree crowns, projected into the horizontal sighting direction, appearing

broader than the angle (x within the full360' sweep, corresponds to the crown density
expressed in decimal parts of I. This method is easy to perform but has not been
tested sufficiently in the detennination of crown density, This procedure is
particularly suited for relatively low crowns and in open stands. in dense stands
values over 1.0 may occur because of the crown overlap.
33.4 Photographic Recording of the Canopy Density

Photographic recording can be used to estimate the forest canopy density. To

obtain density vanability in the gr'on rid "fish eye" (divergent lens) photographs are
used. An ordinary camera is also used to photograph the canopy. While taking the
photograph the camera is aimed towards the canopy in a direction that excludes lower
storey vegetation. However fish eye camera is not used commonly because its wide

angle often includes adjacent vegetation outside the target area of the photograph.
The photographs thus taken can be digitised enabling computerised analysis and
thereby helping accurate unbiased analysis of forest canopy density. Also the

photographs can be scanned and converted to digital image having various grey levels.
The image thus obtained can be density sliced which will give an unbiased estimation
of forest canopy density. This method is accurate but is time consuming and most
.

expensive.

33.4. IAerialPhoto-interpretation

: .

Canopy density from aerial photographs can be measured either jocularly or

with the help of some measuring device. Ocular estimates requires practice of
studying aerial photographs under stereoscope but the method is highly subjective.
Objective estimation of crown density can, however, be made by use of some device
such as dot sampling desigy or crown density scale. The dot sampling is a transparent
plate with a number of evenly distributed dots. The transparent plate is placed on one
photograph of the stereo pair so that the photograph of the stand whose density is to
be detennined is fully covered. The proportion of the number of dots which fall on
the tree canopy to the total gives the canopy density.

A still better method is to dot sampling techntque which involves comparing

the aerial photogr'aphs with certain standard scale showing various densities, These

standard scales can be made either from actual photographs of area whose canopy
density is to be estimated or by making diagrammatic representation of various
densities. in order to use the scaleb the interpreter slides the scale placed besides the
aerial photographs under the stereoscope up and down till standard corresponds with
the photographs, The accuracy of the canopy density depends on the scale of aerial
photographs. The larger the scale of photographs more is the accuracy. Forest
density class interval depends on the nature of the photo quality and scale of
photogi'aphs, Following class intervals have been suggested by Tomar and Maslekar
for indian conditions :

1-7

S.

No.

Large Scale
(1:20,000 and larger)

Medium Scale

Small Scale

(1:20,000 to 1:40,000)

(Smaller than 1:40,000 )

I.

< 20%

< 20%

< 20%

2.

20-40%
40-60%

2040%

20-60%

3.
4.
5.

40-60%

60-80%
80% +

60% +

60% +

However the canopy density estimation using aerial photogr. aphs could not

become popular due to difficulty in their procurement and high cost.

.

3.3.5 Satellite Based Methods

3.3.5.1 Visual Interpretation

in the visual interpretation method different types and density are separated
depending on the tonal variation, The visual interpretation technique has been widely
used for forest cover mapping (mon. 1983, 1985; Ponyal at at. 1994). However the
visual interpretation technique is subjective and depends on the field knowledge and

aptitude of the interpreter* The National Forest Cover Mapping programme

undertaken by Forest Survey of india is primarily based on visual interpretation
wherein forests are classified into three major density classes viz. dense forest (>
40%), 0p^n forest (40-10%) and scrub forest (< 10%) on 1250,000 scale* The
methodology is subjective and varies with the aptitude of interpreter.
3.3.5.21mage Ratioi, ,g
in general two satellite based methods have been used in conventional remote
sensing analysis of the forest vegetation viz. statistical method and physical method.
The statistical method (supervised and unsupervised classification) classifies similar
types of vegetation into specific categories through multivariate analysis of multi
spectral data. Classification of vegetation status using the statistical methods can
often be misleading because the spectral characteristics of different types of
vegetation are frequently quite similar at certain times of the year, Hence it is often
difficult to conduct detailed classification of vegetation status using the statistical
methods because the spectral characteristics are cuminulative effect of variation due
to type of forest, canopy closure, background canopy roughness, illumination

geometry and spatial resolution o^ the sensor, It is possible to minimize this problem
by overlaying multi-temporal data covering the same area, observed at different
seasons of the year. However this will require a repetition of the analysis and will not
.

"

be cost effective.

in the physical method of deterrnintng vegetation status, the spectral

characteristics of chlorophyll~a are extracted from the satellite data. Vegetation
density is estimated through synthesis of vegetation index and bare soil index. The
NDVl is often used to stratify vegetation and non-vegetation areas. However,
I-8

physical method have the disadvantage as the subtle differences due to canopy density
in the infrared and red are not highlighted in the ratio based indices, The value of the
vegetation index derived from physical method is not so reliable in situation where
the crown closure is < 50%.

3.3.5.3 Forest Density Classification Approach

The vegetation crown closure density can be stratified using combined ms

data and NDVl. The necessary information with respect to forest type and canopy
density can be obtained from ms data with a good percentage accuracy. It has been
observed that NDVl with normal data increases the accuracy in physiogiioinic
stratification. NDVl as such has shown positive relationship with the canopy closure
(R. oy at @I. 1990). However the results are not reliable if crown closure is <50%.

Further the accuracy can be increased by merging IRS Panchromatic data (spatial
resolution 5.8 in) with IRS LISS 111 data. It enables to discriiulnate the maximum
texture with respect to forest canopy density (FCD) and hence provides an unbiased
estimation of FCD.

33.5.4 BiophysicalSpectralResponse Modelling

Biophysical vegetation indices are capable of handling variations introduced
in a scene due to aimosphere or sensor or vegetation background influences. They
can reduce the effect of bias and assist in the extraction of significant features of a

sperm^d g, 'on, d chi^. t (C*"rran 1980; Main^ at at. 1981), These indic^s also help in
categorising mixed pixel effectiveIy (MCCloy and Hall 1991).
In0-10FCA (1993), suggested that the anbtle differences can be improved
by using power degree of the infrared response and it has been found to be more
sensitive to forest density and physiognomic vegetation classes. The index thus
calculated is ternied as advanced vegetation index (AV^. For more detailed

estimation of vegetation status, a bare soil index (BIO is calculated. The underlying
logic for this approach is based on the high reciprocity of bare soil status and

vegetation status. Therefore, by combining both Bl and AVI in the analysis, it

becomes possible to estimate the gi. o1md status on a continuum ranging from
vegetativeIy rich conditions to exposed soil conditions. ' The Bl is formulated with

medium infrared infonnation, in addition to vegetation type and status, the mmque
characteristic of forest is its three dimensional structure, To extract information on

this shadow index (SI) is used through extraction of the low radiance of visible light.
This approach isolates vegetation feature space using advanced vegetation
index (AVD and bare soil index (BD. The vegetation feature space is further stratified
using shadow index (SI) on the basis of texture variation introduced by the canopy
shadow of the forest stand.

The Landsat TM bands (except band 6) Were nomialised using linear

transformation. The temperature calibration using coefficients for Landsat 5 was

done to estimate ground temperature. The temperature data has only been used to
separate soil and non tree shadow* The spectral data set is subjected to physical
transformation using enhancement techniques (Anon. 1993). The vegetation feature
space data was stratified based on the 'texture' of the data as influenced by the canopy
I-9

shadow, Finally a rule based logic is implemented to achieve land cover and forest
density classification, The results can be improved by using water mask from Landsat
TM band 7. The approach adopted in the present study is shown in orig. 2).
3.4 Growing Stock Estimation
Though the strategies have shifted from commercial forestry to conservation
forestry; people oriented productivity in terms of nori wood forest products have to be
estimated for their yield regulation. Hence, estimating gi. owing stock in tenns of wood
and non-wood forest products is a priority. Multistage sampling approach using
remote sensing provides most reliable estimate of forest resource stockings (Chacko,

1964) (Singl, and Roy, 1990) ^stimat^d vol^me of individ^al forest typ^s and further
grouped in to utility groups using Landsat TM (1:50 000 scale images). Ultimately,
stratified populations (at different levels) are sampled to estimate total stock, This
approach with multi-resolution sensor systems can provide higher accuracy in
estimates.

4. Vegetation Type Mapping and Monitoring Changes

Forest vegetation types are classified based on physiognomy, structure,
function and composition and also on heigltt classes and distinction of woody tissue.
Forest classification in the region is based on physiogyomy, climate, sucession and
ecological status, SateUite remote sensing data presents above information on
integrated manner and has been used for vegetation stratification by several authors

(Honghton and Woodw^11 1981; Botkin at a1.1984; Roy, 1993). Th^ different much^s
carried out demonstrates that satoriite remote sensing can stratify forest based on
tonowing criteria:
. Phenological types as a function of leaf duration (e. g. Evergreen, Semievergreen, Moist deciduous and Dry deciduous)
. Major communities and gregarious species (e. g. Sal, Dipterocarpus, Pine,
Teak, Bamboo, Oak, Deodar, etc. )
. Vegetation types of unique environmental setup (e. g. Mangroves, Sholas,
Riverme, Alpine pastures, etc. )
. Canopy closure expressed as forest density (e. g. encroachments, shifting
cultivation and different density levels)
Since vegetation is the indicator of environmental condition, it continuous
monitoring acts as the watchdog for the environment. Remote sensing provides
infonnation on any positive and hegative change in vegetation cover unambigiiously,
being above all the levels of socio-economic or political bias, Hence remote sensing
has the potential to act as a legal tool to overcome all the problems/difficulties &
.

socio-economic conflicts.

The basic premise of the change detection through remote sensing is that the

spectral signatures change coriumensurate with the change in the land cover (R. oy et,
a1. , 1996). Superimposition of two period maps to find the change is an established
procedure. Digital change detection methods involve more time on computer for the
I ~ 10

response and changes in ground facts Iiniit its accuracy. Many procedures exist for

digitally Though it is possible to identify classes on both images, scene dependent

spectral detecting change viz. image differGridng, image rationing, princi at
component analysis, image regression etc.

The medium resolution Wide Field Sensor (WiFS) in ERS -IC/ID and IRS -P3,

WiFS has also demonstrated of its ability in classifying forest type, land use/land
cover, and agricultural production in Indian subcontinent. It is well tuned for

monitoring vegetation status and dynamics. improvement over existing accuracy of

80% using careful analysis and combination of multidate data has been an ested

(Roy at "I. 1995), The infonnation derived from satellite remote sensing is being used

invadedfa
in varied facets of forest management,t ff t
.

5. Deforestation Monitoring

The pressure of forests is greatest in the developing countries. The primary

causes of deforestation are encroachment of forest area for agr. icultural production and
exploitation of forest cover for meeting housing and industrial needs, Deforestation
leads to an increase in the loading of CO2 in the atmosphere. increased albedo and

change in aerodynamic rougliiiess over doforested areas alter the energy balance
bearing implications on atmospheric circulation patterns and rainfall statistics.
Deforestation leads to soil erosion & gradual loss of biodiversity.
The amount of vegetation loss/deforestation due to encroachment can be

estimated by the use of remote sensing technique. The impact of slash and bum

duting and after the 'jinummirig' operations is clearly visible from remote sensing
imageries. The representative relationship between the population density and the
percent of forest cover provides infonnation about the rate of deforestation and

thereby helps in formulating the mitigation plan. Utilization of remote sensing tool for

stock mapping and growing stock estimation for forest management improves

Ii
b'I't
.
'
reliability.

5.11 Encroachment Monitoring

Reserved forests are always prone to encroachment by the locals. To monitor

the extent of forest loss & reserved area boundary, satellite imageries are useful b
providing regular and repetitive inforrnation. Thus it acts as a legal tool for the control
of encroachment and forest loss. Space Application Centre has demonstrated this in

Sanjay Gandhi National Park (Maharastra). However, operational needs demand

mapping ability on I: 10,000 or I^rger with better than loin positional accuracy.
,

6. Biodiversity Conservation Planning

India is identified as one of the important biodiversity pool known for their

genetic, economic and economic prudence. Conservation of biodiversity can only be

achieved through conservation of biological habitats, which require a detailed survey,
and inventory of the existing bioresources, conservation of ecosystems through .
I - 11

appropriate legislation and protection of marine resources, parks, sanctuaries, forests,

wetlands and coral reefs. Remote sensing techniques are useful in locating different
types of bioresources, to identify appropriate corridors surrounding natural habitats
and protect them from human intervention and other harmful influences, which
endanger the existence of these habitats.
ECologists generally classify biodiversity under three categories, which
represent the biodiversity function associated with landscape features, already
identified inventory, differentiation diversity and pattern diversity, Spatial, temporal
and composition patterns and their interrelationships, can be derived from
identification and analysis of landscape units using remote sensing techniques. Using
different landscape ecological parameters along with restricted ground observation
biologically rich areas, hot spots, warm spots, hot specks, can be identified for
conservation prioritisation. A project is presently undertaken by the Department of
Space and Department of Biotechnology at national level to prioritise the biologically
rich areas in the three mega-diversity sites of India. The conceptiial hypothesis and the
approach are provided as Fig, 3 & 4 respectively. Biologically rich areas &
Disturbance index maps will be prepared in .three important biologically sites viz, ,
north eastern region, western Ghats and north west Himalayas. The database will
definitely form the basis for biodivesity conservation and prioritise the sites for more
intensive investigation, The data base will facilitate to formulate legislative measures
for conservation.

6.1 Habitat Management

Preservation and protection of habitats is the gyeatest way of bio-diversity

conservation. The essential parameters for habitat analysis are forest cover, food,
water and topography. important national Parks have been analysed for various
habitat zones based on theiT attributes. Remote sensing provides valuable information
on vegetation landscape structure. Geogi. aphic Information System (GIS) is employed
to analyse for habitat interspersion and juxtaposition which indicate the degree of
suitability of an area for wildlife habitation. interspersion and juxtaposition maps are
used with restrictive factors, availability of water, slope of the terrain, remoteness
from arithropogenic structures to stratify existing landscape structures. Habitat
suitability is modelled using spatial (derived from remote sensing) and non-spatial
(field data) infonnation. Successful habitat analysis leads to the management of
habitat bonstically (Roy at, at. , 1995),
6.2 Protected Areas

For more than two decades, attempts to protect wildlife and wildlife habitats
in india have been pursued by Setting up national parks and sanctuaries. in these
protected areas, however, human. .activities such as animal grazing and forest produce
collection are banned or severely restricted, even though people may have lived in and
conserved the areas for generations. Ensuing conflicts, particularly when combined
with industrial pressures on the areas, have spurred many conservationists, social
activists and forest officials to reconsider on national and local levels the artificial

divide between conservation and human rights. A proper management plan prepared
for the village locals to satisfy their socio-economic needs is possible by using remote
sensing data.
I. - 12

REMOTE SENSINGFORBrODJl, BuslTYCffd. R4C7:EMS4TfONAT

JANDSCAP, :I. Friar.

= HABITAT
= DISTURBANCE

.ENVIRONMENTAL COMPLEXTTY

I;$;;*;. *,. .
TERRAIN
CUMATE
.Rainfall

.Tempemture
.

,' *^'-,~ *,.

,*^^'* a:

""~ '~"'-- *, . .-***, .. 4"

Low

DigruRBANCE

IANDSCAPE
ECOLOGY
.Patch characteristics
.Human Intervennon

REGIMES

, ,:a**~-' . I. .., I*

.' ,$<64;is;
. ,,;;""";f

,;*;,,

HanTTAT

cocosvSIEMS)

VEGETATION I ECOSYSTEM
MAFFDtG

( Remote Sensing )
+

( Ground Tr"th )
+

IExisting Vegetation Maps )

Fig. 3
.

BIODIVERSITY
PRIORITY ZONE

The protected areas become totally isolated from the surroundings resultin in

'islands'. Remote sensing images are utilized to study all the characteristics of
'island' and the balance between linkage from outside the island & confi ration
within it. Hence the identification & study of the protected areas in eneral Tovide us

with 'design principles' that can be used in the planning of 'nature reserves' The

habitat maps can be used in GIS to analyse for suitability of particular wild life (R. oy
at at. 1995).

7. Biomass and Productivity Estimatiom

Vegetation type, its biomass- and productivity estimations are considered

important components affectihg biosphere-atmosphere interactions. The

measurements of biomass per unit area and productivity have also been set as oals

for international Geosphere-Biosphere Frog, ume (IGBP) (Fig. 5 & 6). Biomass

I - 13

<11.1i^in:!!^$:,=*!\:;I.
,,!9-,!$9;t^!^3<;,:,
*:e;*-,'&*@^
' $1\***:^:PI,
,tilt^*;**
*,
;.- -,::.*,,,** .:, -*,~- .,.?;.,*,
*.-,,$;{***';* ~*,**;:$;-*t**
$__ * r^!>,*.
tt**"-*Lt. t*;,*:;,*:*,**

*:;;:::***~^;^*!;!;&^!;^;;S^^^,;^^#;^$^\4:34;^$;;^a
Vegetation

Ancillary

Ground sample

type map

database

data

I simplification

Terrain

Erosysteni uniqueness
representative ness, Species

Forest/1.10n

Broad cover

Road,

forest map

types

Settlement

Fragmentation

porosity
Interspersion
Juxtaposition

richness Biodiversity value

Proximity

Terrain

Analysis

complexity
Landscape I

Analysis I
Package I
______,, I

Disturbance Index = f (fragmentation, porosity,

interspersion. bioticlhuman

disturbances &juxtaposition
of vegetation types)

+.

+,

BiologicalI Richness = f(Ecosystem uniqueness, Species Richness,

Biodiversity Value, Terrain complexity &
Disturbancelndex)

Fig, 4
disttibution in the second method the empirical relationships are used for

converting multispectral response in biomass values. The ground measurements are

done in the hornogenous vegetation sample stratums. The correlation coefficient
between observed and predicted. .biomass is 0.77 (with error of estimate as 10.5%),
Aniong the two methods, the - first one has been found to more accurate and
extendable on larger landscapes.

Primary productivity and their seasonal variations are the key components in
the carbon cycling. Estimation of this provides msiglit for our better understanding of
the global climatic changes and the rates of mass and energy fluxes between plant
canopies and atmosphere, which are intricateIy linked with the landuse changes.
I - 14

Recent research has shown that the integrated vegetation index can be related directly
to the regional primary productivity. However such studies have remained confined
to NOAA"AVIl^ spatial. resolution. Several models are now available for the

assessment of the forest productivity like model based on biome types only
(Whittalcer and Likens, 1975), model utilising the meteorological parameters (Lieth,
1975) and Forest BGC modal (Running and Conghltari, 1988),
With the recent advances in remote sensing for ecological research the
Production Efficiency Model approach has come up which takes into account the
most versatile parameter of vegetation i. e. AFAR,
Thus, models utilising the direct relationship of APAR with satellite
parameters seem to attract the most. The physical basis of these models is that the
measurements of light interception by canopy can be related empiricalIy to dry matter
production over time to produce predictive models of productivity as a function of

intercepted radiation. The relationship between dry matter production and intercepted
radiation under relatively constant environmental conditions at a single site usually is
linear (Monteith, 1977; Under et* @!., 1985; Camel et. at. , 1987).

W ,*, ' e , ,,,,,,, ,

Where, W (,) is the dry matter accumulation during time interval t, B is the
conversion efficiency of plant to convert Itgilt into dry matter and Cp, b, is the absorbed
photosynthetically active radiation during timeinterval t.
(Kuruar and Monteith, 1982) and others noted that vegetation indices

calculated ad the nonnalised difference vegetation index 0'1'[IN~IR]/11N+IR]) or Simple

vegetation index (INAR) are approximately linearly related to the fraction of the

incident photosynthetically active radiation (p'A1^) that is absorbed by the canopy

(FAPAR). These two relationships can be coupled to create a model in which primary
productions is related to the temporary sums of NDVl by means of a light conversion
offtoioncy focior (Pmc^, 1991). (Goward and Dy^, 1987) has bullt unon the
conceptual framework of Monteith (1972) to develop a model of primary production
applicable at the regional scale. The model utilised for the study by Roy & Abhineet,
1998 is illustrated in Fig 7.
NFP = s Z (N*S*)
Where, NFP = Seasonal production in Kg ha-I; N, = integrated NDVl; S, = Total

incident PAR (during time Interval t); e =Light conversion efficiency (gC/unit PAR)
, I

8. Integrated Natural Resources Management

The answer to provide food security to the world without affecting ecological
balance lies in the adoption of new scientific tools available particularly, the use of
vital inputs from space remote sensing and biotec}in o10gical advances. Remote
sensing provides useful infonnation related to agriculture, forestry, land resources,
I - 15

and soil mapping, under ground and surface water, wasteland, coastal mapping,
mineral prospecting and ocean resource. integration of these vital inputs from space
with socio-economic data at micro-level has resulted in the initiation of sustainable

integrated development strategies at each watershed level, for achieving increased

agricultural production without impairing ecological & environmental integrity &
affecting the future habitability of the planet.

. . . ..... ..... * . . ,..............,. * " . * ...... *

* . . ,..... ... ~ * ..... * ~ " . . ... .,. " " " " " ...~..... . . . . .,.~......*

mini;;^;136^;FA;^^ON;;;$3;
$31';*;;:I;.^a^';^':';;;;^;'^:GE:';;=-!':

a^;^;^;!;0^:^at^;;;
...
.

.......

....,......

*~,....*..,........

.'. . .* .'. . . .,,. . . . . .'. '. . .* .'. ,.'.,. .'. * . '.'. .* .'. . .'.'. . .'.' . . .'. . .'.-. .'. . . . .

,... . .. . ... ". .". . ,.,. *.. . . ..... * . . . . " . . . . . . . . . . . , . . . . . . . . . . .

*;*;.;;^I^^$1!^^;;:^i^;$;;;;

......................
..,.........
W. . .'.'.'.'.'.'. * .'.'. W

~ "..... . .,.,.... ... .*..,. *.,. * ... . ... ............

I, *, <\' I
lint;;:;lit^inns^1116/1^;^^.:;^I'Aji^jin;;;;;;:;*

LEE

""" ' ' ' '.. . . * .. * ,..,. V. ' <.'.'

** ->*-- -- *-*, -',','*^;^'I^A'rA;!*INS;:$ '*- -,--' --'-- -'-- -*---, ^*

,.,,".............

by.

.........

."".................

4<'-".."\.".:6:3. ,:**.:*:*:,:tabx. x. *:*'**:.<*.*:**:*:-*:*:<-,<.

by.

$^;^I@^,$;^:;^@^;;;$

~ ','..'.'.'.*,.'. .....* . ... ..... .~.............. ... ..,. " ... . ..*... *... '.' ' ' ' ' ' ' '. '

by.

........ ...... ..... .*. . . ....... . .. .......... .. ...

*.,:-,:.=**=.:.,..**:.*:+: .:\:*:*:.:.**:-,:.=.:-:.=.*:.*:-*a. ., t ...:,:*:

, . ....*...'.,.'.'. '..' '.'.'.'. '.'., '." " "' ' "'r ', ' ' * v "E

as^;'^^$;jg;^:it^$;;;;;;;;
-----------;B - ---~--;;;;;;^;$^:^^inn;;;::;;$^

.... ................................,..... ..<......,,..,,......,... ,...,

... .........,.. ..,.".....,,.. ",."."."..,.,,.. .,,,...........

;B^^tt^it
^-^.^i6^^@^

;^;a^i^t^:$it^ij6^;;;;;;;

'-'.',.'-'"'",.',.',.,,.', '*.'.',. .',.,,.'

am, ""","~',.~,",.~

::>-:***"0^**'S';^j' 'Ii^$;::**;"

",,,"

$3^;^^^;;^;;:^$$;;;;:
..**,....,. ,.*. '.*"' ' *".**.."*. , *..*. *

coq",,*.".~~~"".,.,.~.
.

.....

................

am".',. * .',,".',,.',.'.'.'.'~.',.', ~ "

'%,.:<,.:<.:.:C<C. :.*:<.:.:<.'.:<*.:.:.:*:,.'<.*:<J-',.*r

.......*.....

,.............. ,. " ......................

3:3^;;S^*i:^16;:;;^$t$5;13^
'.'.'.'*t, .'r. ',.'.",.,,.,. v r. . ."..,.",,,, -'" "." 'S

*:<.',.:<<.*:.:*.:<t. :., C. :.:<.:.:<.:.:.'.:.\:<.:.:<.:<.:*,

$3^^^^^-:ti^^;;;;

.*,~- -;I;:^termme-*---,....*....

'<<' '., * ,. . * . . . . . . . . " . . . . . . .-. . ,. . . . * . . ' ' ' '

.'$:'--::',:->:-*:;^!I^t;:* *: tilt. ,:.*: :*: :.*: *:

;;;;;$^;:^;^;^In^^;;;;;

, , -~,,~".-,.-,,.-,.- -.-,,.-.-.-.-~~.-,,.-, *

----------- - ;!my - -, -------

$5^,::,$;t'i^;;^;^jib;e^""""'

,' *",-' 4.4'.'.'.*L. .<.,...<u. ................\.. t-.*....<- .

"' + V V ' "' ' f, " V ' ' ' V " ' ' ' ' V ' ' ', ' " ' '

.-::*,.-*:-. v*::*..*.:,:.**.:;3:<,.+<.'-':'.:.'.'*:'*,.'

... t. .<".<, I. -. '. .. , ., :*:-* *.. ..~,..-:.-*.-..,.... .r-

. ...*,. .-.- -.- --- - ;11n^ ,....,.,,.,,.,

...

..........

..

..

..

3:3:1^*^t^;^:^I^ii@I^;;jin;$^iai$;;;;;;;;
. ..... *.....,,..... .. *. am* ,,- -*'-~- - - e a;,' " ' ' ' ' ' ' ' ' "' ' ' ' ' ' "' ' -'-'- ' - '-'--- *
. .,.'.'.* . . . . . . . . .,..,. . ..,. . , . .,. .,. . . ..,.~..... . .,. . , . * . * ,. * " . . ... ..,. am . . ,

*,>,*;"*,'*, .' J' > ** ' ' v * '.'. '.. < , :\ *..*.... CL, r

Fig. 5. lipFROACH'FOR BIOM4SS ESTML-ITJONUSD^IG REMOTESENS. nVG

( Statisticc, ! Sampling Technique )

I - 16

I~

o00

I1.10 Vl
*_

FOREST

'

TYPE
MAP

' .I'..

SLOPE MAP

: -','1,110MOGENl^. ons. '-';.'

. - yEGETAnOl! STRIA. TA

.,

...,.,~

..~..

..

Extraction of

Spectral
Signatures

' Ground Measured ~

Sample

Seasonal

Spectral Data '

^+..

Parameters

* Biom@ss: * ZIP, 4. R
* ,..- J, . E. g. .E, ,

EMFURICAL

SPECl^L MODELS

is':^. q^^*!, in^^s^-^'^.;^^>;,

Fig. 6. lipPR0,4CHFORBfOM4SS/PRODUCTmn'
BSTLAL, I T/ON

WreckATED
NORMAUSEn""mmNCE
VECRrAnON UNDER
innWl

3.17ELUTE' .!
}. SUB ,. :+;
MODEL ' ' 4;

fig. 7 MODEL FORES77M4770"OFrvETPRl, *, RY

PRODUCz70" us\G 547E"'7E DERIVED WORM4usEo

0117'ERE\CF IE'GETA770\"VD^'(IV0,7)

', ~
"~
"~
' ,t
. '/', It,
- ~ , *.. ~. *
A, Ak

IA, ,, = I, ( z row )

in

NETP

SATELUIEDERIVED

Y . ,;

FRODU

AFAR

CLIE", AnONS

MODEL

I^
*.

. A . . . t',,'.*,,**~,,'

NFP = e I^ AFAR )
vEGErAnoN

FIELD BASER AFAR

MEASURED, ERrs

TYPES &

PRO"^vnY

RATE OF co.
INTAKE

ENERGY CONVERSION

EmuERCv t .I

I, = Monthly total i". ichnt PAR ( MJ in. 2 )

t;, PAR = Ratio of fraction of PAR intercepted absorbed
by the photosy, Ithetic elements

ripnr, Nn grip Morin, .

... ....~... ,

,..

.
.,

NDVl= bitegyaied Noonalised Difference Vegetation

Index ( NDVl ) for growing season

I - 17

8.1. Management of Land Resources

Continuous monitoring of land use I land cover with remote sensing imageries
agricultural planimig have been of immense use in providing information on temporal
changes in forest, mangrove and shifting cultivation areas, enabling to take timely
measures for arresting and conserving its valuable resources. The ability to identify
saline/alkaline or reduce soil sannity/acidity as the case may be, and adoption of
alternate suitable crops or cropping patterns to restore the fertility of soil to the
original level (Ragliavaswamy et. at, 1992).

National level land use mapping on 1:250 000 scale has been done using
satellite remote sensing data. The data has been put to use for agro-climatic zoriation.
The physiographic analysis and landuse infonnation has made it possible to map soil
association up to the scale of 1:25000 scale. National Bureau of Soil Survey (ICAR)
has mapped the soil of the country On 1:250000 scale* Besides, methods are available
to map soil degradation processes like satinisation, erosion and water logging.
Grasslands or grazing lands are another important biome, which have been
degyaded due to ever increasing cattle population. The pilot studies carried out have
demonstrated that besides physiogiiomic classification of grasslands, satellite data
also provide valuable information on the degradation al status of the grasslands. The
spatial infonnation of gt'assland types, stratumwise biomass estimation and sodoeconomic information has been used to estimate carrying capacity of these grasslands,
8.2 Management of Water Resources
Remote sensing satisfies the first task in the management of water resources
by regular monitoring and inventorying of both surface and gi. ound water potential.
Regular monitoring of the contour level changes of reservoirs over a period of few
years and use of appropriate regression models based on the well known cone
formula, has made it possible to quantitativeIy compute water storage capacity of the
reservoir as a function of time.

Integr. ation of satellite derived & ground-based information using GIS has
helped in evaluating irrigation system pertonnance, including the utilisation
efficiency of canals and distributaries, and to identify problem areas requiting correct
action. Staggering of crop showing dependence on viater availability, particularly at
tail end of the canal system, identification of water logged areas to relieve water
congestion pockets, & detailed characterisatton of deficiency in selected distributaries
to promote equity across and along distributaries. These information's have generally
resulted in enhancing the paddy yield by almost 25% i. e. from 38 Qtsnia. to 48 Qts/}Ia,
in the Bhadra command area in ^Icha. -

Application of space remote sensing for identifying underground aquifers in

India has enabled identification of drinking water resource in rural areas. Regular
receipt of feed backs from over 200,000 case studies, have confirmed that use of
remote sensing imageries followed by conventional geo-chemical methods has
increased the success rate of locating aquifers to 93% as against less than 45%
achieved using conventional methods alone. Remote sensing provides advance
I - 1.8

drouglit warning, detection of crop pests and accurate assessment of crop damage due
to natural disasters.

8.3 Wetland Conservation

Wetlands are ecologicalIy important habitats, which are tremendous pressure

during recent years. Besides being economicaly important the wetland are termed as
kidney of terrestrial ecosystem. They harbour varied and unique biodiversity. Remote
sensing data has been used to map and monitor major wetlands on 1:250000. Selected

wetlands are also mapped on 1:50000 scale to characterise their surrounding. These
maps have been first reliable database on the wetlands. The seasonal characteristics
are analysed using pre and POStinonsoon satellite images. It is expected that future

satellites with higher resolutions will prove an asset in inventory and management of
wetlands in the country.
8.4 Resources and Infrastructure Development

Remote sensing can play a vital role in the infrastructure development in the
following manner:
assessing the need for infrastructure
assessing the feasibility of themhastructure
providing baselineirifonnation needed for planning
assessing the environmental implications of the infrastructure &
suggesting corrective, 'management measures.
monitoring of updating of the infomiation about infrastructure.

While infrasimctore is an essential component of growth process, there are

several stages involved in its development, starting from assessment of the need,

location, technical feasibility & developmental strategies; remote sensing inputs are
useful in all these stages,
.

With increasing industrialisation, migration to cities & urban centres,

increasing industrial demand for water and increasing environmental concerns, it is

expected that use of remote sensing infrastructure development would dramatically

.

Increase.

ERS-IC hybrid data (PAN + LISS-Iru on 1:15,000 scale is used in urban

planning and infra structure development & development of infonnation bigliway.

The enhanced output was utilised in developing City Infonnation System (CIS), in
site selection for setting up of industrial park.
8.5 Urban Forestry

Appropriate location of sites for plantation based on environmental

components like soil, water, rainfall, temperature, maintenance cost, plant growth rate
etc. are needed for proper planning. This is not possible by the conventional method
of surveying. Remote sensing data provide infonnation with respect to extent and
location of available lands and its spatial distribution of execution of urban social
forestry schemes.
I - 19

9. Forest Fires

Forest fires are recurrent problems in indian forests, Prioritisation of fire risk

zones is a practical concept and an aid to fire management and planting (Jam at @I.
1995) (F'ig. 8), Such an approach provides information on fire prone areas and helps
in predicting forest fire. Forest managers can make use of this infonnatioll to create to
construct watch towers or fire lines. Satellite remote sensing can provide valuable

infonnation to identify and detect of forest fire. Optical remote sensing is unable to
map forest fire damage below the smoke of the fire. The temporal resolution in
presently available satellites does not provide capability to provide information to
combat forest fire. However, thermal data of NOAA-AVIERR are being used to map

hot pixels on daily basis, European Commission is compiling this infonnation as a

part of global biomass burning monitoring mission. New initiative with higher spatial
resolution and eiglit bit thennat imaging is planned by Gennan Aerospace Agency
(DLR).

10. Controlling Land Degradation

in appropriate landuse, erosion, sannisation, water logging and chemxical
degradation are undoubtedly the major causes of degradation of land environment,
Desertitication is the ultimate result arising from recurrent droughts, deforestation,
and overexploitation of agricultural lands and severe erosion of dry lands. Ready and

timely availability of satellite observations of existing landuse pattern from various

satellites enable all nations to take appropriate measures for improving the land
environment and recover degraded land for productive use, Wastelands criticalIy
affected districts in india were mapped with approximate village boundaries on a 1:50,
000 scale. Geographic futonnation systems has been used to arrive action plans for
sustainable rural development planning, considering production and conservation
(Anonymous, 1997).

11. Environmental Impact Assessment (EIA)

Environmental Impact Assessment is an activity designed to identify and

predict the impact of an action on the biogeophysical environment and on man's

health and well being and to interpret. and communicate infonnation about the impacts.
It is essentially a dimension of the planning process rather than a specific

product. It bears a specific function i. e. 'to influence the decisions to be made for the
environment' and its values.

Remote Sensing by virtue of its capability to provide synoptic and frequent

coverage plays a potential role in both rapid and comprehensive EIA. It contributes
sigyificantly to number of biophysical and cultural factors for impact analysis.
Methods of impact analysis are adhoc methods, overlays, checklists, networks,
matrices, quantitative or index method and models.

I - 20

Fire History
Ground D"!"
on Fuel

Satellite
Derived

I^^^^-

Fire Frequency
Vegetation Stinta

( Fuel Vane )

Vegetation
Type Map

;^^",,.*:*,,,,
a. .*,;?,*-.*,*,,*!*I*};
_ ~*~- ..,~ ,, *,*I,
^

gr\*5. *e. ,. ' ,

A, ,. ._* * *.,* .*

**#**.,

.:,**!;,,,*:*,':$~. . .,$\ *4~.,

;*^!24!Jot^:^z^4^a
^*^,^^. 4^*;:**$

.. f

*'*

.,.-'T"ms ~,:,,.,.,

' ~ ',;;Q
*;**
$14*r.
;.*-:e.-t;.*:~.
**.~':';;10'41;*;,***;
.

~~,

. ."~ ... . . ~ ,.~.~

nigiril
Terrain Model

Road Network

I^^^^

Slope

Ignition Sources
Landuse &
LandCover

Fig. 9. Approach for Fire Risk Modelling using Remote Sensing grid
Geogrqphtc11^formatio, I System

12. Environmental Information and Environmental Democracy

People leaving in the modem democracies are essentially safeguarded from
harm by free flow of information. Unfortunately, there are serious problems of a dual
society. imformation rarely reaches to those people who end up suffering the impact of
environmental degradation due to lop sided development. Hence 'Tight to
infonnation' should be a fundamental right, enshrined in a legally binding entity so
that any company, govenunent agency or multi- lateral institutions can be forced to
release any information relating to its own activities or of its subsidiaries or associated

institutions, which timeatens peoples health or environnient. Tile effective use of

remote sensing and space technology will provide timely, accurate and duly canbrated
information which can lend requisite edge to every sector.
is the most
The riglit to survive with a certain mothcum of dignity
fundamental of all human riglits, but unfortunately, this is not enjoyed by a large part
of human kind today, The world .remains starkly divided between the rich and the
poor - between those 'who enjoy the resources of the earth and can worry about its

future and those who have to scrape the soil for less than bare survival. Economic

security today can become the very basis of ecological security tomorrow. All
internationally guaranteed 'right to survive' backed up with appropriate programmes
like space technology and remote sensing technology will only assure ecological and
environmental regeneration.

I - 21

13. Information needs in Forest Management

The information on the quality, distribution and the growing stocks of the
forest areas are required to prepare working plans. The working plan data is gathered
for this purpose every ten years, using ground-sampling techniques at divisional level.
The coinparinient level data relating to various factors is also maintained in the form
of maps and tables. The growing stock by species and stands of different site qualities
are also mentioned. Detailed forest cover type maps are needed for laying of
individual felling coupes, PIaiming of roads, firetines, wildlife management etc.

The present practice of manual data collection and storing in registers, history
sheets, field formats, periodic reports etc. is comborsome and not easily Tetrievable

and manageable because of sheer size and non-uniformity. Ground surveys take a lot
of time, money and labour to prepare stock maps. Satellite data combined with latest

data processing technology holds potential to provide above information and update
in desired interval.

13.1 Wildlife Management.

The factual infomiation on habitat parameters i. e. food, water and shelter is

required to evolve plan for protected area. The data on human interventions and
impacts on wildlife are becoming increasingly important. Remote sensing provides
reliable data on three vital parameters viz. , shelter, water and food, which when
integrated with other field data in GIS, can help in wildlife habitat evaluation and
management.

13.2 Issues On Operationalis ation of Forest Management

For optimismg data selection for vegetation applications following aspects

need to be considered.

. Consideration of phenological stage of vegetation types.

. Phytogeogi'aphical region under investigation.
. Landuse practice.
. Sun elevation and azimuth,

. Cloud and snow cover.

The areas where satellite remote sensing has potential to be used are
elaborated subsequently.
*

13.3 Vegetation Type Mapping.

, I

. Phonological types as a function of leaf duration (e. g. Evergreen, Semievergreen and Deciduous),

. Major communities and gregarious species (e. g. Sal, Dipterocarpus, Pine,

Teak and Bamboo).

. Vegetation types of unique environmental set up (e. g. Mangroves, Sholas,

riveram and alpine Zones).

I - 22

,:

. Canopy closure expressed as forest density (e. g. biotic interference like

encroachinents and shifting cultivation).
. Middle infrared band is ctitical.

CryPG mapping accuracy in the range of 82-90 %).

13.4 National Forest Cover Mapping and Monitoring
. Robust digital approach in classification.
. Statistical approach for accuracy evaluation,

. Change areas need to be monitored on larger scale i. e. I :50,000 (or larger).

. Creation of permanent sample points for obtaining training sigyature.
. Total mapping approach for change detection is not necessary (recommended
= 5yr cyde).
.

13.5 Forest Density Stratification and Analysis of Human Interventions

.
.

' TITO (ADOn*, 1993) approach using biophysical spectral response modelling
approach is probably the most comprehensive method available today,
. With the advent of high resolution PAN data of IRS-1010 at 5.8 in now it is

possible to obtain to obtain better information on stand density.

. Most accepted approach is data fusion technique wherein LISS 111
mumspectral data set is combined with PAN data to obtain forest stand density
at 20% interval.

. Texture analysis of PAN data and subsequent merging with UsS 111 data has
also provided > 87 % accuracy in digital classification for density
stratification.

13.6 Estimation of Growing Stock

. Identification of forest cover type, density, encroachment status, cultivation

patches available through. remote sensing.

. Timber volume estimation only through ground studies. Forest maps derived
..~,.

from satellite data provide basis for sampling.

. Multi~phase sampling design has been found to be very useful in gi'owing
stock assessment for large areas (example Lido-FRG Phase-rrl project on
timber volume inventory in Karnataka).
. Some studies show ' usefulness of Microwave L-band data in volume
estimation.

. Fused data set, Larger scale and biglier spatial resolution enhance the accuracy
of estimation and level of stratification.
..

13.7 Areas where Forest Mama^ers are looking for more information
. Species Identification,

. Plantation Moulinng (at early sing^).

. Abandoned Shifting Cultivation Stage.
. Forest Density Stratification.
. Real Time Forest Fire information.
. Merchantable Timber Estimation.

I - 23

Regeneration Status.

14. Summary of Achievements and Future research areas.

Areas where methodologies have been tested or validated are as below:
. Forest type and broad density stratification possible.
. Biomass/growing stock estimation possible using two/three phase sampling
design.
. Assessment of encroachments, deforestation or degradation possible.
, Mapping of shifting cultivation areas, including current and abandoned
shifting cultivation possible.
. Wild. life habitat characteris ation and habitat suitability .analysis possible
together with ground measurements,
. Afforestation/ecodevelopment and plantation monitoring possible, using
higher resolution data.
. Post-fire damage assessment, affected area and fire-prone areas mapping
possible.
. Estimation of area under grassland types possible.
. Biodiversity characterisation at landscape level.
. inventory and monitoring of wetlands possible.
. Land and water pollution assessment and moultoting possible to a large extent.
. Detection, monitoring of environmental hazards and hazards zoriation possible.
. Assessment of environmental impacts of major development activities
possible.
L4, I The researchers in the field of remote sensing applications could address
following issues in their research efforts
. Forest type classification through integration of physiognomic, climatic and
edaphic data
. Forest density mapping with narrower density grades
. Understorey characterisation, community analysis and species identification
. Illicit cutting, over-gi'azing and minor forest degi'adation assessment (in terms
of density reduction by illegal felling)
. Forest fire prediction and modelling, active fire mapping
. Non"timber forest product assessment and merchantable timber
. Site quality, age-class assessment, regeneration status and physiogiiomic type
. Ecological analysis such as biogeocb. Ginical cycting, carbon flux etc
. Disease detection and monitoring,
. Air quality and water quality assessment and monitoting.
. Detailed environmental impact assessments.
' Wetland conservation and management.

CGobotanicalindicators and mineral prospecting.

Joint Forest Management Action Plan development and their monitoring.
I - 24

15. Conclusion

Apart from direct benefits, space technology has clearly demonstrated the

interconnectivity of both natural and arithropogenic phenomena occurring any where

in the earth, through weather, climate, geosphere and biosphere, mextricably linking
the fate of each country with that of the world as a whole. Unless sustainable

integrated development strategies based on space remote sensing and biotechnological

inputs consistent with sovietal demands and cultural practices are adopted in a world
wide scale, it will not be possible to ensure enviromiiental security in the coiniri
decades.

The fundamental aspect of long-tenn sustainable development is based on the

paradigni of technblogical innovations, economic dotennin. ism and ' h siCal

constraints imposed by the environmental imperative of striking a judicious balance

,

between ultimate exploitability and regenerative capacity.

Evolution and implementation of well thought out, long-tenn national policies

and creation of a promotive international atmosphere alone can lead to such a

development across the world, enabling the entire human human}ind to share the
benefits of satellite remote sensing.

I - 25

References

Anonymous 1983. Nationwide mapping offorest and ,'01/10rest areas using Lands"t
False Colour Composite for the period 1972-75 and 1980-82, National
Remote Sensing Agency, Hyderabad, Technical Report, Vol. I, pp. I-36.

Anonymous 1985. The State of Forest Re;port, Forest Survey of india, New Denxi, pp
1-82.

Anonymous. 1993 , Rehabilitation of logged over forests in Asia/Pacific region. Final

report of Sub Project 11, International Tropical Timber Organisation-Japan
Overseas Consultants Association, pp I-78.
.

haoriynious, 1997. Geographic information System for Wasteland Mapping in

Punilia district (West Bengal). Pilot Project for Bandwan Block. indiari
institute of Remote Sensing, Dentadrui.

Abhineet Jam , S, A. I^avan, R. . K. Singli, K. K. Das and P. S. Roy. 1995. Forest fire
risk modelling using remote sensing and geographic infonnation system.
Current Science, 7000):928-933.

Botkin D. B, BStes I. E. , MCDonald R, M. and Wilson M. V. 1984. Studyimg the Earth s

vegetation from Space; Bioscience 34508.
Cannon, M. G. R. , Milne, R. , Shoppard, L. I. , and Unsworih, M. H, 1987. Radiation

interception and productivity in Willow. Journal of Applied Ecology, 24,261278.

Chacko, V. I, (1964), A manual on Sampling Techntques for forest surveys , pp 55-65

& (Sth Sampling), pp 94-110 CDouble Sampling), Govt, Publications.

Champion H. G and Seth S, K. 1968, A revised survey of forest types of India 0.16w
Dellit: Govt. Publications)*

Goward, S. N. and Dye, D. G, 1987. Evaluating North American net primary

productivity with satellite data: Advances in Space Research. 7,165-174.

Hougliton, R. A and Woodwell, G. M 1,981. Biotic contributions to the global carbon
cycle: the role of remote sensing; Proc. Seventh international Syinposium on
Machine Processing of R^roote Sensing Data, West Lafayette, indiana, pp
593-602,

1<. urnar, M. , and Monteith, I. L. ,1982. Remote Sensing of plant growth; In: Plants and

the day light Spectrum, adjted by H. Smith q. ,0"don : ACademi. press) 133144.

I - 26

.,

Legg, C. A. (1986): Monitoring of open cast coal mirinig & reclamation works in the
U. K. using Mss & TM imagery, Proc. of 20th int, symp. on Remote Sensin
of

Environment. , kiln Arbor, Michigan 2,931.

Linder, S. 1985. Potential and actual production in AUStrailian Forest stands. Res h
for forest management: Proceedings of a conference of division of forest

research held at CSU;<0, 0m 21-25 May 1984, edited by IJ. Landsber and W.
ParSons.

Lieth, H. 1973. Primary production; Terrestial Ecosystems; Human EC010 . I, 303322.

Madhavanunni I\I. V. , Roy P. S. , ladhav R. N. , Tiwari A. K. , Suchakar S. , Ran anath

B. K. and S. L, 1991. IRS-IA application in forestry; Current Science 61 189192.

Mating1'6au I. P. , 1991. Remote sensing for Tropical Forest Monitoriri : an overview;

in Remote Sensing grid Geographic Information Systems for Resource

M"nageme, It ill developing countries (eds) A, s. Belward and C. R. Valeiizuela

coordrecht : Kluwer) pp 253-278.

MCCloy K. R, 1980. Application of Landsat to mapping grassland condition in th

South East Region of South Austrialia; Soil Conservation Br@, zch Re on,
Sir80, S. A. Department of Agriculture, Adolaide pp I-28.

Mukhe^66, K. T. , T. K. Bandopadhyay & S. K. Pande(1991): Detection & delineation

of depth of subsurface coal mine fire in a part of Iharia coal field,
Photogr'arumetry Engineering & Remote Sensing, 57,1203,

Monteith, I, L. 1972. Solar radiation and productivity in tropical ecos stems; Jou at
of Applied Ecology. 9,747-766.

National Remote Sensing Agency, 1983. Nation-wide mapping of forest. 'and nonforest area using Landsat False Color Composites for the period 1972-75 and
1980-82. Project report. Vol I, Hyderabad.

Padhar, I. S. and Nageswara Rao, P. P. , (1997): All integrated Approach Towards

Crop Production Forecasting, myRMS Bulletin, 20, Jan, 1997.

Ponyal M. C. , Dabral S. L. and Rqy P. S. 1994. Revision and updatin of stock in a s

using remote sensing and' Geographic infomiation System (GIS); Proc. LSRS

Silver Jubilee Symposium*. Deftrad"12, pp 334-342.

Prince, S. D, 1991. Satellite remote sensing of primary production: Coin anson of

results for Sahelian Grasslands 1981~1988; flit. I. Rein. Sens, 12,1301-1311.

Raghavaswainy V. , at a1. , (1992): Natural resource management a new perspective

(ed, ) F. L. , unRMS, Department of Space, Bangalore, '
I - 27

Rao. U, R. et. a1. , (1991): Environmental impact of the Persian Gulf Oil Spill, 01rrent
Science, 60,486.

Rao. U. R. Space Technology for Sustainable Development, Tata MC Graw-Elm

Publishing Company Ltd. New Denxi, 1996,

Roy, P, S. and Abhineet lain, 1998 . A Satellite Remote Sensing Based Approach To
Estimate Regional Net Primary Productivity Of Forest : Global Change
Studies, Scientific Results from IsRO Geosphere Biosphere
Frograriune ,1998. (ed. ) Prof. B. H. Subbaraya, Dr. D, P, Ran, Dr. P. S. Desai, Dr.
B Manildam, Mr P Rajaratnam.
Roy, P. S and Shinsh, A ,Ravani, 1996 *Biomass Estimation using satellite remote

sensing data- An investigation on possible approaches for natural forest. (ed. )R

Urna Shankar.

Roy, F. S (1993), Reinot^ Sonsi, g for Forest Ecosystem analysis and Managem^it in
Environmental Studies in india (ed. ) M. Balakrishnan ,New Dellit: Oxford and
IBH) pp 335363.

Roy, P. S. (1998), Assessment of forest density- a review on available methodology

and significance of TIT0~10FCA project on forest canopy density model.
Roy, S. A. Ravan, N* Rajadnya, K. K. Das, Abhineet Jam and Samam Singli, 1995.
Habitat Suitability analysis of NGmorhaedus gotal - a remote sensing and
geographic information system approach. Current Science, V01.69, No. 8, pp
685-691,

Roy, P. S, , C. B. .S. Dutt, R. N. ladliav, B. K. Ranganath, M. S. R. MurtlLy, B, Gharai, V.

Udaya Lakshmi, A. K. Kandya and P. S. Thaldcer, 1996. ms- IC Application
for Forest and Enviromnent. Current Science, V01.70, No. 7, pp 606-613.
Running, S. W. and Coughlan, I. C, 1988. A general model of forest ecosystem process
for regional applications J Hydrological balance, canopy gas eXchange and
primary production processes. Ecological Modelling 42,125-154.
Patel. J. G. et. a1. , (1997) Remote Sensing and GIS Techniques in siting an industrial
Park in Khatrej Area, Aimed abad District, Gujarat, Proc, of National
Symposium on Remote Sensing for Natural Resources, November, 1,997.

Todd, D. K (1980): Ground Water hydrology, John Winly & Sons, New York,
Whinekar, R. H. and Likens, G. E. ,1975. Primary Production: the biosphere and man,
in primary productivity of the biosphere. ( H. Lieth and R. H. Wlitttekar, EDS),
Sopringer-Verlag, Bermi. 305-2-328.
UNESCO, 1973. international Classification and Mapping of Vegetation. Series 6
Paris.
&:^

I - 28

I. 2 Concept of FCD Mapping Model and Semi-Expert System

,\
.

1-29

Coincept of FCD Mapping Model and Semi-Expert System

A. Rikimaru

JOFCA

111bstr@at

Forest canqpy density is one of the most use:litl parameters to consider in the planning and
implement@tto, I of rehabilitation progr@in. nits sindy is development of bio:!, hsyc"! analysis
modeljbr obt"millg of Forest Canony Density orcD),, singL/^. N05;4TTMd@to image analysis.
rite components of FCD model are four I^ctors ; vegetation, bare soi4 thermal and shadow,
This work is implemented under the research project; FD32/93Rev2(:F') of International
Troyic"Ininber Organization (1270),
I'mtroduction

As generally applied in forestry, conventional RS methodology is based on qualitative.

analysis of information derived from "training areas" (i. e. ground-truthing). This has certain
disadvantages in terms of the time and cost required for training area establishment, and the
accuracy of results obtained. in response to these problems, new methodology was developed
during InO Project FD 32193 Rev. 2 (F), "Robabititation of Logged-over Forests in AsiaPacific Region, Sub-project in', in this new methodology, forest status is assessed on the
basis of canopy density. The methodology is presently identified as the Forest Canopy
Density Mapping Model, or for short the FCD Model. Unlike the conventional qualitative
method, the FCD Model indicates the growth phenomena of forests which is quantitative
analysis. (Shown in Fig. 1, Fig. 2) The degree of forest density is expressed in percentages: i. e.
10% FCD; 20%; 30%; 40% and so on. FCD data indicates the intensity of rehabilitation
treatment that may be required. The method also makes it possible to monitor transfonnation
of forest conditions over time including degi. adation. Additionany, it can assess the progress
of reforestation activities.

The Forest Canopy Density (FCD) Mapping and monitoring Model utilizes forest canopy
density as an essential parameter for characterization of forest conditions. FCD data indicates
the degree of degradation , thereby also indicating the intensity of rehabilitation treatment that
may be required.
The source remote sensing data for FCD model is LANDSAT TM data, The FCD model

comprises bio-physical phenomenon modeling and analysis utilizing data derived from four(4)
indices: Advanced Vegetation index(AVI), Bare Soil indexCB^, Shadow index or Scaled
Shadow index(SI, SST) and Thermal index(To. It datem, in^s FCD by modeling operation and
obtaining from these indices.

The canopy density is calculated in ^arentage for each pixel, The FCD model requires less
infonnation of ground truth. Just*for accuracy check and so on.
FCD model is based on the growni- phenomenon of forests. Consequently, it also becomes
possible to monitor transfomiation of forest conditions over time such as the progi. ess of
forestry activities
The application test were implemented in these area. The evergreen forests are in the islands

of Luzon (F'irtlippines) and Sumatra (hidonesia); and for monsoon (subtropical deciduous)
forestsin Ching-Mai(Thanand) and Tent 0.16p^I).
I-30

Actual

Ground Condition

,^^^^,.
Conventional
RS Method
Forest
Barren Grassland

TypeA

Type B

Type C

,^^^^..
Assessed
\

Ground Condition
Type A
..,, @
.

Type B

Type C

-:"I, .

. .

Fig. I Analysis by Conventional RS Method

InOIJOFCA 1999

1-31

Actu al

Ground Condition

,^^^.
New RS Method

(FCD Model)
9,.
Forest Canopy Density

Vegetation Density

(FCD)

,^^^^,.
Assessed
Ground Condition

Fig. z Analysis by" FCD Mapping Model

ITrO I JOFCA 1999

I-32

2. Characteristics of Four(4) Indices

The indices have some characteristics as below. The Forest Cano Densit Mod I b'
data from the four (4) indices. Fig. I must rates the relationship between forest conditions and

the four indices 071, Bl, SI and To. Vegetation index response to all of vegetation items such
as the forest and the grass land. Advanced vegetation index AVI reacts sensiti I f th
vegetation quantity compared with NDVl. Shadow index increases as the forest d 't

increases. Thennalindex increases as the vegetation quantity increases. Black colored so'I
shows a high temperature. Bare soil index increases as the bare soil exposure degrees of
ground increase. These index values are calculated for every pixel. Fig. 3 shows the

characteristics of four indices compared with forest condition.

Note that as the FCD value increases there is a corres on din increase I th SI I . 111

other words where there is more tree vegetation there is more shadow. Concurrentl , 'f th
less bare soil (i. e. a lower Bl value) there will b'e a corresponding decrease in the n value. It

should be noted that Vl is "saturated' earlier than SI. This simply means that the maximum W

values that can be recorded appear Gather in the analysis, This happens because the W c t

data from the total bio-mass, regardless of the density -of the trees or forest, On the th h d,

the SI values are pimiarily dependent on the amount of tall ve etation such as t h' h
a significant shadow.

Table. I shows combination characteristics between fourindices.

-I

..

C^ ~ ~

High

...........

...

.,

./
,%

,
Low

'.

0.
'..

I
,

."^"

Vegetation

Bam Soil
Shadow

^^
....

.,.

^^^^
..

, .

Fig. 3

The charactertstics offour indices for forest condition

I-33

Temperature

Table. I Combination Characteristics between Fourlndices

Hi- FCD

Low-lFClD

Grass Land

Bare Land

AVI

Hi

Mid

Hi

Low

Bl

Low

Low

Low

Hi

SI

Hi

Mid

Low

Low

Tl

Low

Mid

Mid

Hi

3.0btaimimg of rollr(4) Indices

a. Advanced Vegetation index (Aun
When assessing the vegetation status of forests, the new methods first examine the
characteristics of chlorophyll-a using a new Advanced Vegetation index (AVI) that is
calculated with the following foamulae.
Bl~B7: TM Band I~7 data

B43=B4-B3 after normalization of the data range.

CASE-a B 43 < 0 AVI= O
CASE-b

B 43 > 0 AVI = ((B 4 +I) x (256-B3)x B 43)1/3

b. Bare Soil Index (BD

The value of the vegetation index is not so reliable in situations where the vegetation covers
less than half of the area. For more reliable estimation of the vegetation status, the new
methods include a bare soil index (Bl) which is forumlated with medium infrared
information. The underlying logic of this approach is based on the high reciprocity between
bare soil status and yegetation status. By combirting both vegetation and bare soil indices in
the analysis, one may assess the status of forest lands on a continuum ranging from lit
vegetation conditions to exposed soilconditions,
Bl= ((BS+B3)-(B4+Bl))I((B5+B3)+(B4+Bl)) x 100 +100
O < B1 <200
9

The range of Bl is converted within 8 bits range

c. Shadow index (SI)
One unique characteristic of a forest is its tree dimensional structure To extract intorination
on the forest structure from RS data, the new methods examine the characteristics of shadow

by utilizing (a) spectral infonnation on the forest shadow itself and (b) thennal infonnation on
the forest influenced by shadow.
The shadow index is fonnulated through extraction of the low radiance of visible bands.

SI =((256-B I)x (256-B 2) x (256- B3))us

d. Thermal Index (To

, .

Two (2) factors account for the relatively cool temperature inside a forest. One is the shielding
effect of the forest canopy which blocks and absorbs energy from the sun. The other is
evaporation from the leaf surface which mitigates warming. Fonnulation of the themIai index
is based on these phenomenon. The source of thermal information is the thennal infrared band
of TM data.

I-34

4. The Procedure of FCD Model

The flowchart of the procedures for FCD mapping model are illustrated in Fig. 4. linage

rocessed results corresponding to the flow chart shows in Fig.

LANDSAT TM data
. ,

Noise Reduction Process

..

Scan line noise , Aimosphetic noise ,

Cloud area, Cloud shadow area, Water area, etc.

Range Normalization of TM data for each bands

Advanced Vegetation Index

Shadowlndex

Bare Soil index

Thermal Index

Black Soil Detection

Vegetation/Bare soil
Synthesis Model

Vegetation density %

Advanced Shadow Index

Spatial Process

Scaled Shadow Index

ShadowPercentage

for Forest

Integration Model
Forest Canopy Density Map

Fig. 4 Flow Chart of FCD Mapping Model

a. Noise Reduction ;Clouds, cloud shadow and water area d'

clouds have a higher i" 'an hit , , black or combinations of different

formed at the ground level. This is one t OILg Pa^a G fir ed, water

,loud area. ate' ' " k d a histo am of TM Band4.

b. Vegetation Density ;VD 'ncj al component

tatisSiS.
G PBecause
., 11 , vjessentially,
and Bl have
highBlcorrelation
of negative0. .neg'
After.'that,
Vl and
have high correlation.
96 ' set e

scaling of zero percent point an a un re percen

c. Black Soil Detection . . I b d. Where the soil is black

SIa data
is be
extrac"
I h.tod.recent
born,slash.
low itradiance
confuse
ears to
black "due
and. bum.data
, low may
itta ianCG
a y ,,shadow
h , hi h

tore duedue
to its
absorption
rate ofrate
sun oenergy.
.^ut s a ow
tern Gramre
tohigh
its high
absorption
sun. energy.
, -d d. o Ia s are
soil temperature. By overlaying ata an a
I-35

LandsaiTM dab 8.1.100-1992 Path-1.1 R0"-041 Nepal

False Colour Composite Image

V I rungelaiton trader)

B I (Bare So" "xiex)

Lo,

High

Low

S I ishachw mmx)

Low

T I finermnl Index)

High

Hid.

,,^^

^;^
VD(Vegeta, ,on Density Mapj

SSl (Scared Shadow Index)

Integration odel
I

.*}'!^"'

<61,

^
,.
,

,$P ^t^. it $1,

a

FC D (Forest Canopy Density Map) I

Fig. 5 Procedure of FCD (Forest canopy Density) Mapping Model

In'O JOFCA Rinmaru 1997

also useful when evaluating the relative madiance of different parcels of land characterized by
various shades of black soil.

d. Advanced Shadow Index ;Asl

When the forest canopy is very dense, satellite data is not always be able to indicate the
relative intensity of the shadow, Consequently, crown density Thight be underestimated. To
deal with this problem, the new methods include those described below for dotennining the
spatial distribution of shadow information. Details in (A. I^. ikimaru. 1996)
e. Scaled Shadow Index ;SSl

The shadow index (SI) is a relative value. Its nounalized value can be utilized for calculation
with other pararrieters. The SSl was developed in order to integrate Vl values and SI values.
In areas where the SSl value is zero, tins corresponds with forests that have the lowest shadow
value (i. e. 0%). In areas were the SSl value is 100, tins corresponds with forests that have the
highest possible shadow value (i. e. 100%). SSl is obtaliied by linear transfonnation of SI.
With development of the SSl one can now clearly differentiate between vegetation in the
canopy and vegetation on the ground. This constitutes one of the menor advantages of the new
methods. It significantly improves the capability to provide more accurate results from data
analysis than was possible in the past.
.

f. Integration process to achieve FCD model

integration of VD and SSl means transfonnation for forest canopy density value. Both
parameter has no dimension and has percentage scale unit of density. It is possible to synthesize
both indices safely by means of corresponding scale and units of each.
FCD =(VD X SS1+I)112.1

5. Development of the Semi"expert System for the FCD Mapping Model

At present, RS assessment of forest conditions is a task reserved for personnel who have
received special trading in remote sensing analysis. Forthennore, the images derived from
conventional RS analysis have rarely been simple enough to be readily understood by
practicing forestets, plan^ling officers and other concerned individuals.
As a follow-on phase of the previous project, the Japan Overseas Forestry Consultants
Association (10FCA) has been implementing ITTO Project, FD 13/97 Rev. I (F), and
developed the Semi-expert System to make the new methodology more available to persons
who are not RS experts. The system is now available on CD-ROM fomiat.
.

\, L I

6. Principal Features of the Semi-expert System

The Semi-expert System, or named the FCD-Mapper, is a computer software package
compatible with Windows-type personal computers. The FCD Mapper contains the
algoritlrrns and other fonnulas utilized to compute values of the several indices contained in
the FCD Model for the analysis of satellite imagery data.
The system will upgrade the planting capacity of decision-makers in forest management and
increase the infonnation collecting ' capability of foresters, while reducing costs and saving
\

time.
.

Use of the Semi-expert system will facilitate production of "user-friendly" satellite imagery
that accurately portrays forest conditions. Access to such imagery will increase the capability
of forest ors to provide accurate, unbiased data on the status of forests in a format that is easy to
understand.

I-37

7. The Concept of FCD-Mapper

The concept of FCD-Mapper is the useful system for the forester, and not for the remote

sensing engineer. The Practical Processing System for the Forester which implements FCD
Mapping Process as if the Remote Sensing Professional would operate. FCD-Mapper has the
Ianowledge of remote sensing specialist, for analisys of FCD model.

Conventional Processing System

System

RS expert

Forester

Semi-Expert System

System

Forester

(RS e>^it monlad"

The Concept of FCD-Mapper

8, Specifications of the Semi-expert System

Data Source:

Landsat TM data

Optional Made:
011^litResiitts:

ERS-IC; ID; SPOT4, eto.

Eon Map (Color I^^ Copy); Area Counting Sta^CS of FCD

Stratification(CommentsofFCD Condition)

E^orc F^e Fom^t:

Computer
CPU:

Bin^IP

DOS\IN17iridows 95, 98 or 1.11' type

Pentium 133 ^^I^I^ or above

Memory
32 un or above (641^^ recommended)
Hard Disk Capadty: Fte^space 500 ^ or ahave

(for process^ of TM full coene, IGB or alum recommende(I)

Ifu^ge^^lay
Flintsr:

1024X768pi>^s, 641^colorvatiati. on orahave

Windowsco^. pathletype (colorim. ODD^ome)

reD Madel:

into^active (^erationwith. A^^nm of Semi^t ^^in

Sth-memory Madin: MO-Disk230^^IB (640^^IBr^Drumende^

1.38

9. Comparison of the Semi-expert System with Conventional Methods

Semi-expert

Work Station

Personal

Coinpi:Iter

S stem

Processing Time

0.5 days

4~5 days

5~6 days

Operator

Forester

Remote-sensing
IE;xpert

Remote-sensing
Expert

2~3 months
Re quired. 2~3 days
for Training
Hardware Cost
$ 3,000~4,000 Us $ 20,000 ITS~
Time

Software Cost

Provided for the

ITTO Members

$ 10,000 ITS~

2~3 months

$ 3,000~4,000 Us
$ 4,000 ITS~

10. The Core. members of the Development Team

The FCD Model and FCD Mapper were developed by Dr. At sushi Rikimaru. This work was
carried out in collaboration with Dr. P. S, Roy, Dean of the indian institute of Remote Sensing,

Dr. Surachai Ratanasermpong, senior research officer of the National Research Council of
Thailand, un. Ruandha Aging Sugardiman of the Directorate General of Forest and Estate

Crops inventory and Land Use of indonesia, and Mr. Virgino Basa and Ms. Alma ATquero of
the National Mapping and Resource Information Authority of the Philippines* Computer
software prograinming of the FCD Mapper has been accomplished by Mr. Minoru Narajima.
Continuing improvement in the FCD Model and the Semi-expert System is a collaborative task
of all those just mentioned.
Re:fere, ,ces

JOFCA, 1991. Classification system on log!^ed-over forests. A workshop for the ITl'O project
FD 2187(F) Sub~ project U. Rehabilitation of logged-over forests In Asia/Pacific region.

Manila, Philippines, December 1991

10FCA. 1993. Rehabilitation of logged-over forests in Asia/Pacific region, final report of Subproject rr. Prepared for TFTO, March 1993. Part ill pp. 1-41
10FCA. 1995. Proceedin s of the workshop on utilization of remote sensing in site assessment

and Ianning for Febabi it ation of logged-over forests. Cisarua, Bogor, hidonesia, September

25-2^, 1995. ' ' '

10FCA. 1996. Proceedin s of the workshop on utilization of remote sensing in site assessment

and
Ianning for robabi^tation
of logged-over
forests.eBangl!,
Thailand, July
30- August
I^,^Ian'ing
Ta
ii n gg.
re ok,
, aian
, uy
U ,I,

A. Rikimaru. 1996. LANDSAT TM Data Processing Guide for Forest Canopy Density Mapping

', onrig o e, ,. wor s ^11 On -I d^ I 3 - A I,

199^ pp, I-8
10FCA. 1997. Utilization of remote sensing in site assessment and planing for rehabilitation of

logged-over
forests.
Project
report onin.FD32/93
Rev.for^;^I,O.Rehabilitation
of logged-over forests
in Asia/Pacific
region,
Sub-project
Prepared
January 1997,
1-39

I ,

..

,. 3 Algorithms used in FCD Mapping Procedures

\.

I - 40

Algorithms used in FCD Mapping Procedures

ARikimaru

JOFCA

Definition of Technical Parameters

Vegetation Index 6/1)

NDVl ; Normalized Differential Vegetation Index (!, OAAoriginal)

NDVl = (NIR-R) I (NIR+R)

AVI ; Advanced Vegetation Index JIT'01JOFCA Original)

AVI = (NIR x (256-R) x (NIR-R) + I)" , 0< (NIR-R)

Awl ; Advanced Normalized Vegetation Index

Awl ; Sunthesized' Index from NDVl and AVI ,

with Principal ComponentArialysis
Bare Soil Index (Bl)
Bl = ((SunR+R) - (B+NIR)) I ((SunR+R) + (B+NIR))
Thermal Index in)
Tl; Calibrated Value of Thermal Band Information

Shadow Index (SI)

SI= ((256-B)x(256-G)x(256-R))"
B; Blue Band, G; Green Band, R; Red Band
NIR; Near Infra-Red Band, SunR; Short Wave Infra-Red

Advanced Shadow Index (Asl)

Forest Gap Detection; IF Vl value is smaller than vegetation threshold level ,
the subject pixel look upon Nori Forest area. Then Asl value is zero (0).
Black SoilDetection; IF rivalue is smaller than Thermal threshold level,
the subject pixel look upon Black soil area. Then Asl value is zero (0).

This process is implement for avoidance of mistake as black soilto be

forest shadow area.

Spatial Process; The case of Forest Canopies are closed condition, Shadow
area is few observed from satellite. Then three (3) by three (3) pixel area
are searched ,surround of subject pixel. Then maximum SIva!ue input as
SI vale of subject pixel .

I - 41

Vegetation Density (VD)

VD ; Vegetation Cover Density (%) for each pixel
Calculate from 1st Principal Component of Vl and Bl,
and Calibiate of Minimum and Maximum Vegetation Cover Level.

Scaled Shadow Index (SSl)

SSl; Calibrated Shadow Index for Forest Area

Forest Cluster (FC)

FC; Indicate a tendency level of Forest Area,
High value shone high possibility of Forest Area.

FC=(Vl x SI x (25681) +I)in

' I
,

Forest Canopy Density (FCD)

FCD; Forest Canopy density (%) for each pixel

FCD= orD X SSl +, )" -,

\
.... . ,

.
..

1.42

The out line of FCD mapping procedure flow chart

For the first time, it mm^s Eon,
Registration. of the New project

Reprocessing of
the existence project

Import image

FCD Map making preparation work

Set Water area, Cloud area, Cloud shadow area
Normalization of jinaue

Selects of the Vegetation Index 6/1), as following

Advanced Vegetation Index (AVI)
Normalized Differential Vegetation Index 0.10Vl)
Advanced Normalized Vegetation Index (AN\71)

Calculation of the Thermal Index orl)

Calculation of the Shadow Index (SI)

,

*. ..*

Calculation of the Vegetation Density GID)

Calculation of the Scaled Shadow Index (SSl)

b

Producing the. Forest Canopy Density Map

1-43

(, ) Designation of the project name, for the first step

Tlie first step in implementation of FCD Mapping is to desi^I, ate the project name.
T}Ie project records the processing procedure, the relfi^Tence e>^pert sheet name and the
name of the person performing the analysis.

Each project is arranged or managed by project name. When the project is registered
by its designated name, this makes is easier to conduct the analysis, produce the FCD
maps, do the necessary processing, change the processing procedure if required, and
resume processing after stoppages due to rest breaks, or for other reasons.
T^Ie e>^pert sheet is the recording file. It contains the knowledge built into the semi-

e, ;pert system by a remote-sensing e>, pert. 111e e, <pert sheet includes the processing
procedure, the various computations and formulas, the parameters for triplit by the
operator of the system and so on.

a. For the first step to make FCD Map, it requires to register the Project name
For the first step to make FCD Map, sets the project name, the expert sheet, the
analysis person name and so OIL
b. The case of Reopen the project

When you resume processing after any pailse or interruption of a previotisly

registered project, the work begius at t}lis place (i. e. at the file name). Here is where the

name of the project will appear, along with the map as it appeared at the time of pause
or interruption. This is also the place to make changes in the processing procedure if
necessary.

(2) Designation of the image and the. analysis area

Tliis means designation of the format, the file 11a;me, and the analysis rarea of the
input image.

a. Designation of the fomat of the input image

T}lis means selection of the format to be use^. (i. e. BSQ, BIT. ,, or B^!ip), the size of the
image, and the band of the inputinnage.

b. Designation of the file name of the input image

.
..

c. Designation of the analysis area , .

It designates an input range and the sampling interval Alternutively; the range to input
can be designated from the display image, Procedures are as follows:
d. Confim, ation of the input image

These procedures display the input image and confinns the amIysis range and the band.
I - 44

Expert Sheet

Project Sheet

FCD Method
PCA Select :
AVI
SI Select :

Setting up of the New Project

Reference Expert Sheet

FCD Method

Normal Equation
Asl Arrangement
Gap, Black Soil,
and Special
NO Execute
,.

Project Name

Threshold (ALL O)
linage Format File

linage Input Process

,

linage Format, Band

Analyzed Area Info.
Pixel & Line Start,
End, and Interval

Cloud
Noise reduction and Normalization

Equation Parameters

Noise & Reduction Info.

Mask Threshold

Water, Cloud, and

Cloud Shadow

**a,

Figure of the relation between Expert Sheet and Project Sheet. (412)

I'.

,..

...

Project Sheet

Expert Sheet
AVI, NVl, ANVl, Bl, SI

Equation Parameters

Index Process

Made Index Info.

Average, S. D.

TI Calibration Coef.

Made VD Info.

Vegetation Density Process

..

Average, S. D.
VD Threshold

I'

.,

Clustering Parameters
Initial Cluster No.

Scaled Shadow Index Process

Stability

Made SSI Info.

Average, S. D.
Forest Class Info.

Made FCD Info.

FCD

Equation Parameters

FCD Process and Map Statistics

Average, S. D.

Figure of the relation between Expert Sheet and Project Sheet. (212)

(3) Preparatory procedures for making maps, and normalization of the

.

linage
Tlie normalization of the image is the process through which all data is adjusted to
conform to a common standard, It is necessary because of vanati. one in shade on
different parts of the image due the time period and site spedfi. c foatul, es of the place.
a. Procedures for masking bodies of water
These procedures set the value of those portions of the image where noise exists
because of water, and remove (i. e. mask) the area.
b. Procedures for masking clouds
Tliese procedures set the value of those portions of the image where noise exists
be catise of clouds, and remove (i. e. mask) the area.
c. Procedures for removing (masking) the cloud shadow area
Tliese procedures are automatically synchronized with the procedures for masking
clouds. As e>;plained above the aloud mask responds to Drag & Drop movement of the
left mouse button. The semi-e><pert system produces point information (i. e. the extent of
movement on the image ) of the doria shadow, which of course relates to movement of
the cloud. This point information is used to create the cloud shadow mask A maximum
of three (3) information points can be derived to delineate the area of the cloud shadow.
Basically; the right btitton. meriti of the cloud shadow mask is made by adding IPOint

Selectl, IPOiii. tinitializel, I^^11askl to the right button menu of the ^Ise color image.

b
..

I-47

Original
False Color linage

,^
V

Noise Recognition

Water

Cloud Shadow

Cloud

.^

^,

Removed Area

Noise Masking
\ ,.- ,^

J.

! \ 11. I^- .

\ I"
\

I<.\'h-v -e
\

.<.

, ., IQ^.* ,..,
Noise Reduction linage

P ocedu e of No s

Redu tion

Removed Area

Noise Masking

Range Normalization
*

,,~

j\;*,
\

Noise Reduction

+.
L
..
.

linage 7\

, I"

Normalized Data Range

220

M:Mean
S :Standard
Deviation

,20

20

Original Data

M. 2S M M+2S

Histogram
M-2S

L-- M+2S

I \;^i, '

.\
\4, ,
,

oc dur

I'*
=

Normalized
.

False Color linage

a I iza o

Original Data
Range

(4) Designation of the Vegetation Index on)

Tlie FCD Model chooses a vegetation index from the relation between I^I\71, NDVl,
Awl and Bare soil index (Bl).

(5) Calculation of the Thermal Index in)

(6) Calculation of the Shadow Index (S^
Calculation of the shadow index involves a choice among of two kinds of normal
equation or a log equation, both of which are included in the semi-e, :liert system. The
system also revises forest gap and bladc soil information. Additionally the system can
calculate the Advanced Shadow index (As:D.
a. Correction of the forest gaps

Quite often, there are Forest Gap areas in the analysis range. Tlie semi-expert system
delineates the boundaries of these gaps by calculating their value through use of a
Vegetation In dex
b. Procedures for dealing with black soil
In many instances, the area being analyzed contains patches of black soil. These

patc}Ies call be confused with shadow. Tl, .e following steps are taken to elm^late this
confi. 1ston.

c. Calculation of the Advanced Shadow Index (Asl)

When revising the Forest Gap or dealing with Blacl< Soil, the semi-e, :Tiert system
calculates Asl.

I - 50

.I

Asl (Advance Shadow index) Option

3. Spatial Process

2. Black Soil Detection

, . Gap Detection
~,

Threshold
of N o n-

Vl

SI

Threshold

of Temperature

Vegetation

Nori-Vegetation

I It!

123

3x3 Ma jinum

Nori-Shadow

Nori-Vegetation

Spatial Filter

Area ^

Area ^

200 ,80
,40

200

90 200 90
200 180
a40

200

90 200 90

Asi-3

Asi-2

Asi-I

" **

" lit}
,

\I

I\
.

4-

I.

*'
,

Asi=SI .

Asl=0 .

Asi=SI

Asi-, -- Asi-2 -- As1-3

.~::~ . ,

, ,, .

I'".":'., ',:;I*;i,",' ,'

,

Asi

roced re of Asl

(7) Calculation of the Vegetation Density orD)

TTLe system calculates the density of vegetation lay the percentage.
a. Designation of the vegetation area

1111e system designates the naniimuin. value of the vegetation area byVl.
. ,

b. Designation of the Bare Soil

. .

TTLe system designates the mittm. urn value of the bare soil by Bl.
c. Designation of the range which calculates the VD

It calculates VD in the designated bare soil area and designates that ve tation
density at'that part of the image is O %.

Next calculates VD in the designated vegetation area and it designates where the
density 100%.
d. Calculation of the un

I " 52

Bl

o%

255

g ation

Den ty
100%

^
Ist PCA axis
255

I Pixel

Vegetation

9""o

3%

7%

00%

79^^"'"

3 91"

o%

Ba re Soil

^O

Vegetation

ensi y

odel

AV

DV

AV

Principal Component Analysis - ( Bl, Vl )

1st PC +

sP

S
Bl

Bl

Bl

"I.

..^^

"

^,"^,.
.,,

Lit

",.."
*t

ANVl

NDVl

AVI

Selection of Best Vl - ( AVI, N Vl, ANVl )

I^--

Vegetation Cover
Low

High

I st PC linage

P 'nc'pa Coinpone t Analysis of Bl and V

(8) Calculation of Scaled Shadow Index (SSl)

T}Ie semi-e>^liert system fillst alassifi. es a duster using Vl-Bl-SI and the Tl image. It
then identifies a forest area.

It next adjusts the range of SI in the forest area from O to 100 % and calculates SST.

a. Confirming the conditions for processing SSl from in-BISl false color display
The system changes Vl-Bl-SI Into GRB and it displays in ^Ise color image. It also
confirms the mutual relationships of Vl, Bl, and SI. The following foatures appear on
the ^Ise color image:

- the area of Ingh Deusi^, Forest is displayed in the cyan.

- the area in grass, agricultural crops and so on is displayed in green
- the area in bare soil is displayed in red.
b. Designation of the fores. area

The system performs cluster erassifi. cation using the Vl- Bl-SI and TT images and the
result is displayed. This designates the classification category of the forest and inputs
the forest area.
c. Calculation of SSl

T}lis function of the system adjusts the range of SI in the forest area from O to 1.00 %
and calculates SST.

I - 56

Vl linage

Bl linage

Lo*

F1^h

I-

Threshold of

Threshold of
Baresoil

Vegetation

I-

I-

Vegetation

Bares oil

Area

Area

,^.

I
1st PC of Bl. Vl

1st PC of Bl. Vl

Bares oil

Vegetation

Area

Area

Histogram of 1st PC
. Bare Soil

Calibration of Vegetation
in and Max Level
167

^. VD O%

2311

. Vegetation

VD , 00%

50

100%
.

VD ( Vegetation Density ) linage

Procedu e of V

(9) Takes statistics of Forest Canopy Density (FCD), Implementation of

the accounting area
This part of the system performs the following functions:
- Takes statistics FCD every Grid or polygon. area.

- Presentation of the datairinlimber of hectares per'^'CD category*

- Saving of results in an Excelfile format.
- Saving of images in a Biim. ap file (B^^IP) format.

a. Producing the FCD Map

TTLe ROD map is produced using data from the Vl and the SST.

b. Accounting the area in each FCD category and saving the FCD Map
This furiction of the system calculates FCD in every Grid or polygon area. When it
takes statistics in the polygon area, 'inputting the mask file. Tlie mask file is a bitmap
file (256 colors) which is prodlined using existing imagery software. TTLe same palette

~\

value can be used to aistingiiish. one FCD category area from another. Tlie results can

be presented in tables and saved as an Excel file. The image can be saved as a himnap
file.
c. Presentation of the results

Results showing the number of hectares in each FCD category can be based on either
the whole Grid or polygon area.
d. Designation of the partition which take statistics of FCD Map and,
Designation of the display color

The de^. ult is IT pieces of partition in all. (0% and I partition per 10% from I%)
\.*

Also, display color is suitably arranged.

e. Accounting of the FCD Map
It takes statistics of the area accounting.
f. Saving of the results

When the TABLE OF MAPPING STATISTICS window is displayed the results can be
saved for the first time. Tables ate s^I. \!. edin E, ECel file format and the image is saved as a
bibnap file.

(10) Close of the project to be processing

I - 58

...

,*.* ..'!!. "

v<;

.,;a. ,, ... . , Jh' . , .*. *",

*,.,
~
*

f:,
t43,
- ,.,*,,

*,
*

.,,.

$:*;t

', tvt ~'~'

*,

,*.. * :*. . '.. "~. '* ~** *..! J4, ,*h.

-.~--

"--*h-,-,, , , ,.. _.,, I'S**,

,--

*.\,'. t. ',;\.';$;,* ..;*., r? . :.? <5

~~,,.~-.,

,ng;; a, ' ,L:*L3:, -;*.~

*"""

.**, t*>* .;^;* - $*;^:* ,

.*,*. .'*. * 4<44L _' A, .

I
I
I

I *I t, :** *' 1<';$* T"~

-*.*. ; **\* ;*'
<1*, <:1;-.

';-?t';

e;e;

4;.

I
I
I

,..

^^^. ^

*IF:;
~

. _,, L:**; '

I
I
I
I
I
I
I
~
I
^^^. ^

41. ,,

4;:;:,,,,,

a'

I
<9.

I
,?

-^..^,.~

^.

~.-

^...

~----~ s, ^. .~~

I'

a
I
^~

45,

^^.,.
I ~$"

Grass

<2"

t.

,,,,, I one Pixel Size

$9. ^^^

a~~^

3, ,,,,,, If Bare Soil

Ground Condition and RS Data

.

..

ITrO/JOFCA 1999

FC D (^'.)

Forest
+ Bare Soil

Forest
+ Grass and Brush

Forest + Bare Soil

+ Grass and Brush

000.0
o00.0
100%

..,...

^-^^;,
^Q,
.

70%

.
.

00,000

. 0'0

.
.

8
o'^p
o,
000
;""^"
, .
000.
o0
.^ '

co Co
o0
o0 .

o0

50%

.
20%

co

00

~
"
.

o%

;^?,

,, .. *,. ,, b
,,;,;!;. $4;- . .

One Pixel

^ Forest

"

'^! ';. IC'^.^^' a:..

b
-.

q*,

^^;

.I

I^^^ Grass&Brush I^I Bare Soil

The Situation of Forest Canopy Density for One Pixel

ITTOIJOFCA 1999

Forest Canopy Dens^ty/ Mapping Madel'

Forest Canopy
<Case I >

Density

<Case 2>

80%
VD=80%

VD=80%

VD=60%

VD=80%

60%

40%
11

VD=40%

11

VD=80%

20%
in

VD=20%
Tree

VD

\
Bare Soil
\

litOIJOFCA 1999

Grass
^,
11

Shadow
*"

VD=80%

SSl

integration Model

Forest Canopy Density

^ I!

~^^

I'^'^;^~'~^^^
"
I,
,

>

, . .- , 41^,"

^,,

-4

Forest Class of

SI (Shadow Index)
Low

Cluster Process

High
*,^?,

SI Scaled Forest Area

SI Scaled Forest Area Histgram

M : Mean
S : Standard Deviation

M-2S - SSl = 0%

Itr
M

- M+2S-SS1 = 100%

Coo%

SSl ( Scaled Shadow Index ) linage

Procedure of SSl

...

~.

,,..

,..,

,.

#:I^-",^^.*^.
- ,, ...,*,
* . *,-,*^$i FF. f;;!;:*^i^;:,;-!',
." ,>j^'^:,
. ,^:!^,! j^
.- A

',4.' ,,,,,
''' '",' j^. 4%-<;;;^
. , . .I-,^.^;,
.";;, t:'}
*' '*,$ ,/ :,,.>. ,
*.'
11 "^^;:
*!,**,,.
,,,*, : .45
*..;.,!'
"';'^:"I^ '.t~ *.I r. ..*,!.
6- ',,*

. ,.*,?:::^*;;;,-*:-- :I^if '1' ' - '~-^;,~; 11!;^'^ ^;j:^!;

^,/:"4', " ,.. , , -- .:^ ..*, , j-,., ^,;^^.,!.^,, e^ , ' I! I^' ' ' '^ ' -

" "' I% $17^^^-.',, ,- , '>,/?. -:$!4:* *"' ' I.

' I " "~-j '- "", 2, ^., ",,, ,,,;i

it, :;'::,:-., A^:,,,. I -:.,,:,',', -:- ^!;*- '6:41^;,{ .?'

-^".; '14'1. '\' ;y!;^'il :{'.. ';'- :; , I '^;!\^^;!,, ' .,; ,'^

, , ,,. , . , *.,^;,-, 't*,-^;... , , .,,.

I "'^' "" P *' ':;' . "~ ^' . *. I^. .,, ,

S

VD

100%

100%

VD
^,

.
.

. veso
a. .

..
.

SS
50

too%

,,..,. ,,. ,.,.,,, . . * . ,.#:.!. - ^ @j^.,*. , - -.. j

JP. =.\. \;: .t \ " ' ^

~ ~ "" ' ' ,-.* e. ' I- " '
, ",.,':.*'I'c'" ' ;" ' "" ' " ' " 11 * ' ,

141t .',^ ;.. .\-.* ..,, ,. ,

t'lt ,..'. '. J e, . ^^,, ' I, "rip^, 351, ,

b;..;**, . '$!#:; {"' - "'i" '^;;' ,

*"*.
\
* *,.. ,

FCD

,00%

FCD Integration Model

Effect of integration Model

VD(%) SSl(%) FCD(%)
90

90

Land Cover

90

Hig h Forest
90

50

67

Young Forest
90

25

47

90

10

30

50

50

50

^,^

Bush + Tree

Bares oil + Tree

50

90

67

20

90

42

90

21

90

01

I
^"^

Tree + Slope Shadow

Grass land
ITrO I JOFCA I 999

..

2. How to Use the FCD Ma'pper

2.1 Introduction to the FCD Mapper
2.1. , The calculation procedure of FCD Mapping
The flowchart of the calculation procedure anim^trotsdin Fig. 2.1, I.
For the first time, it makes FCD.

Reprocessing of
the existence project

Registration of the New project

Import image

FCD Map making preparation work

Set Water area, Cloud area, Cloud shadow area
Normalization of jinaqe

Selects of the Vegetation Index 011), as following

Advanced Vegetation Index (AVI)
Normalized Differential Vegetation Index (NDVl)
Advanced Normalized Vegetation Index (ANVl)

Calculation of the Thermal Index (Tl)

I
,

Calculation of the Shadow Index (SI)

Calculation of the Vegetation Density GID)

Calculation of the Scaled Shadow Index (SSl)

.

Producing the Forest Canopy Density Map

Fig. 2.1. IFlowc^t of the calc^tonproced. mm

2-1

(, ) Designation of the project name, for the first step

The ^,, st step in implementation of ^'CD Mapping is to designate the project name.

The project records the processing procedure, the ref^Tence e>:pert sheet name and the
name of the person pertonning the analysis.

Each project is arranged or managed by project name. When the project is registered

by its designated name, this makes is easier to conduct the analysis, produce the FCD
maps, do the necessary processing, change the processing procedure if required, and
resume processing after stoppages due to rest breaks, or for other reasons.
The e, ,pert sheet is the recording file. It contains the bowledge built into the semie, <pert system by a remote-sensing e>:^, ert. The e, ^, iert sheet indrides the processing
procedure, the various computations and formulas, the parameters for imptit by the
operator of the system and so on.
a. For the first step to make FCD Map, it requires to register the Project name
For the first step to make FCD Map, sets the project name, the e, ,pert sheet, the
analysis person name and so on.
b. The case of Reopen the project

When you resume processing after any pause or interruption of a previotisly-

registered project, the work begins at this place (i. e. at the file name). Here is where the
name of the project will appear, along with the map as it appeared at the time of pause
or interruption. Tliis is also the place to make changes in the processing procedure if
necessary.

(2) Designation of the image and the analysis range

T}lis means designation of the format, the file name, and the analysis range of the
input image,

a. Designation of the format of the input image

Inns me am selection of the format to be used (i. e. BSQ, BIL, or Bin^IP), the size of the

image, and the band of the trip^t image.

b. Designation of the file name of the input image
..

c. Designation of the analysis range.

It designates an input range and the samplirig interval. Alternatively, the range to triptit
can be designated from the display image. Procedures are as follows:
d. Confirmation of the input image

These procedures display the input image and confirms the analysis range and the band.
2-2

(3) Preparatory procedures for making maps, and normalization of the

.

linage

Tlie normalization of the image is the process through which all data is adjusted to
conform to a common standard. It is necessary because of variations in shade on

different parts of the image due the time period and site specific fleatul, es of the place.
a. Procedures for masking bodies of water

These procedures set the value of those portions of the image where noise exists
because of water, and remove (i. e. mask) the area.

b. Procedures for masking clouds

These piocedul, es set the value of those portions of the image where noise exists
be callse of dollds, and remove (i. e. mask) the area.
c. Procedures for removing (masking) the cloud shadow area

TTLese procedures are automatically synchronized with the procedures for masking
clouds. As e>CPIalned above the cloud mask responds to Drag & Drop movement of the
left mouse button. The semi-e, <liert system produces point information (i. e. the extent of
movement o11 the image ) of the cloud shadow, which of course relates to movement of
the cloud. This point information is used to create the aloud shadow mask. A maximum
of three (3) information points can be derived to delineate the area of the dolld shadow.

Basically the right button menu of the cloud SI. adow mask is made by adding IPOint
Sele^l, IPOint Initializel, INlaskl to the right button menu of the ^Ise colorimage.

.~

, .

2-3

(4) Designation of the Vegetation Index (Vl)

The FCD Model chooses a vegetation index from the relation between A\71, NDVl!,
ANVl and Bare soil Index (B^.

(5) Calculation of the Thermal Index (Tl)

(6) Calculation of the Shadow Index (SI)
Calculation of the shadow index involves a choice among of two kinds of normal

equation or a log eqnation, both of which are trialuded. in the semi-e, :pert system. The
system also revises forest gap and black soil information, Additional15^ the system can
calculate the Advanced Shadow index (ASO.
a. Correction of the forest gaps

Quite often, there are Forest Gap areas in the analysis range. The semi-e><pert system
delineates the boundaries of these gaps by calculating their value through use of a
Vegetation Index.
b. Procedures for dealing with black soil
In many instarices, the area being analyzed contains patches of black soil. Tli. ese

patches call be conftised with shadow. The following steps are taken to elm^mate t}lis
confusion.

c. Calculation of the Advanced Shadow Index (Asl)

When revising the Forest Gap or dealing with Black Soil, the semi-e>:lierb system
calculates Asl.

.
..

. .

2-4

(7) Calculation of the Vegetation Density (VD)

The system calculates the density of vegetation by the percentage.
a. Designation of the vegetation area

The system designates the minimumo. value of the vegetation area by Vl.
b. Designation of the Bare Soil

TTLe system designates the minimum value of the bare soil by Bl.
c. Designation of the range which calculates the VD

It calculates \ID in the designated bare soil area and designates that vegetation
density at that part of the image is O %*

Next calculates un in the designated vegetation area and it designates where the
density 100%.
d. Calculation of the VD

(8) Calculation of Scaled Shadow Index (SSl)

The semi"e>:pert system first classifies a cluster using Vl-Bl-SI and the T!I image. It
then identifies a forest area.

It next adjusts the range of SI in the forest area from O to 100 % and calculates SSl.

a. Confirming the conditions for processing SSl from Vl-Bl-SI false color display
The system changes Vl-Bl-SI into GBB and it displays in ^Ise color image. It also
confirms the miltual relationships of Vl, Bl, and SI. Tll. e following features appear on
the false color image:
- the area of High Density Forest is displayed in the cyan.

- the area in grass, agricultural crops and so on. is displayed in green.

- the area in bare soil is displayed in red.
b. Designation of the forest area

The system performs duster classification using the Vl- Bl-SI and 111 images and the
result is displayed. nits designates the classification category of the forest and inputs
the forest area.

c. Calculation of SSl

This function of the system adjusts the muge of Slim the forest area from O to 1.00 %
and calculates SST.

2-5

(9) Takes statistics of Forest Canopy Density (FCD), Implementation of

the accounting area
T}lis part of the system performs the following franctions:
- Takes statistics FCD every Grid or polygon area
- Presentation of the data in number of hectares per FCD category.

- Saving of results ill an IBxcelfile format.

- Saving of images in a Bitmap file (B^^IP) format.
a. Producing the FCD Map

The FCD map is produced using data from the Vl and the SSL.
b. Accounting the area in each FCD category and saving the FCD Map
T}lis function of the system calculates FCD in every Grid or polygon area. When it
takes statistics in the polygon area, trip Litfuig the mask file. The mask file is a bitmap
file (256 colors) which is produced using existing imagery software. The same palette
value can be used to distinguish one FCD category area from another, T'}Ie results can be
presented in tables and saved as an Excel file. TTLe image can be saved as a bitmap file.
c. Presentation of the results

Results showing the number of hectares in each FCD category call be based on either
the whole Grid or polygon area.

d. Designation of the partition which take statistics of FCD Map and,

Designation of the display color

The default is 11 pieces of partition in all. (0% and I partition per 10% from I%)
Also, display color is suitably arranged
e. Accounting of the FCD Map
It takes statistics of the area accounting.
f. Saving of the results
When the TABLE OF MAPPING STATISTICS window is displayed the results am be
saved for the first time. Tables are saved in Excel file format and the image is saved as a
.

bitmap file.

, .

(, 0) Close of the project to be processing

2-6

2.12 The menu list of FCD-Mapper

A The pull down menu of the main window
Main Menu

Sub Menu

File

New Pro 'ect

O en. Pro^Cb

O eniri the existence Project and Pro^Ct file

Close Pro 'eat

CIOsiri the Pro^Ct and Pro^Ct file

Project Pro e

Editor of current roject To er^r

Menu Pro e

File maintenance of Pro^Ct files

Load False IDLa e

False color dis lay of Bib^IP file

Pseudo color dis lay of B^^IP file
Tiern^Iation of FCD-Mapper

Function

Creation of New Pro^Ct and New Pro^Ct file

load Pseudo In. a

Exit
Window

Help

FCD Menu Show

SemiAtito Mode

ON/OFF switch of the FCD menu dis lay

Tile Windows Hor^ontall

Tile Windows VernalI

Tuna e windows ti^I with}10rizontall

ina e windows tihii with vertically
ina ewindowsdis Ia with inn u
All of jina es icons dis Ia min^d size)
HeI menu

01,110E'E' switch of the SemiAlito Mode

Cascade Windows
ATran all
FCD Hat
About

The version information of FCD MAFF:I^R

,
.. .

. ,

2-7

B.

FCD Menu

Menu
Grou

Main Menu
Sub Menu

Image

Irua e Tin. ort

Function
T}Ie source satellite jina e data jin orb

Tm ort Format
Tm ort File
11n ort Area Select
mora e Confi.
Noise

Index

un

SSL

Desi

atton of the source jina

format.

Desi

atton of the source image file name

Tin. ort area selection

T^Ie selected area dis in and confirmation

Ched^Redtiction.

Deceleration of the noise condition

and selection of the noise reduction menu

Water Set

The water area masltin

rocess

Cloud Set

Tli. e cloud area masking process

Shadow Set

The cloud shadow area mask^I

Normalization

Radiometric normalization rocess

AVI
NDVl

Advanced ve tation. index anal sis

A1. rvl

Advanced normalized ve etation. index anal sis

Bl

Bare soil index analysis

PCA Select

Selection ofo tonum Vl from AVI, NDVl, ANVl

TT
SI Select
SI

Thermal index analysis

Gap Set

Designation of the threshold for forest gap

Blad^ Soil Set

Asl

Desi

Advanced shadow index anal sis

Ve etation
Bare-Soil Set

Desi
Desi

VD Range Set

Designation of the minimum and maximum level in the

un 67e etation density) analysis

VD

VD (ve station densi ) jina

rocess

Normalized difference ve tation i:odex analysis

Selection of o tional shadow index

Shadow index anal sis

Vl-BT-SI

Toness

atton. of the threshold level foie blacl^ soil

atton. of the minimum level value of Vl
atton of the minimum vanie of Bl

anal sis

Confi. mmation of interim results by ^Ise color composite

ofVl/G, Bl/R, SIIB

FCD

Cluster Select
SST

Semi-automatic classification for the forest area

FCD

Trite ation anal sis for FCD ina

Ma Statistics
Grid

Statistics account for each FCD d. asses

Mask

SST jina

analysis

FCD area account by

.
..

FCD area account b

2-8

'd

o1y on mask

2.1.3 File Specification

A Expert Sheet
The e>;pert sheet is the recording file* It contains the knowledge built into the semie>;pert system by a remote-sensing e>cpert. The e>;lierb sheet inclTides the processing
procedure, the variotis computations and formulas, the parameters for input by the
operator of the system and so on.

1111e Expert Sheet has 11 Sections as below.

1'10
I

Section Name
E>^ERT

INDEX

3
4

CLOUD
AVI

NDVl

Bl

TIT

8
10

SI
SSL OLD
F1

1.1.

FCD

Description of Comment and author

Defo. ult Selection of FCD Method
General e uation. arameters of ina^I
Generalequation. arameters of makiri
General e uation arameters of maim
Generale uation. arameters of makiii

OLD file
A\71 file
NDVl file
Bl file

Calibration coef^. dent of makin I^ file

General e uati. on arameters of makin SI file
Parameters of Cluster Anal sis
General e untion. arameters of makin. IFT file
Generalequation arm^eters of makin FCD file

One Section, INDEXis copied to project file whenis generated, thereafter will not be IISed.
These sections are only on the position to record de:t^,. ult values.

Other Section except INDEX is not copied to project file. An e, :pert sheet file name is
.

record on a project file by full path name.

.
..

. .

2-9

B. Project Sheet
The project sheet saves the reference expert sheet name, the processing parameters and
processing logging and so on. .

TTLe Project Sheet has 1.5 Sections as below.

1.10
2

Section Name
PROJECT
FORMAT

BAND

ASSIGN

5
7

unAGE
Inn^X
TIERESl-{OLD

FILE

NOISE GnuCK

Noise information

10

MASK

information of Mask settin

Descri tion of Comment and alithor

( e ual to the section in an tina format file )

(e ital to the section in an inna e format file )
(e 11al to the section in an jina format file )
Information of Tina eArea to be Analyzed
( e ual to the section in an e ert sheet file
TliresholdVanies

File name of e, :pert sheet file, BIT. ,, BSQ or BMP file,

mina e file

11

CLUSTER

Choice cluster number in cluster anal sis

12

INPUTSTT

In ut o erati. on state

13

ErriBSTl'

14

AVE RESULT

011 ut o eration state

Avera e values of enerated jina files

15

SD RESULT

Standard Deviation values of enerated image files

, .

2 - 10

C. linage Format File

Image format file has the information of BU. ,, BSQ or B^!IP file structure. This file can be

made by linage Import Dialog. it'other Projects want the same information of BIT, , BSQ
or BMP file structure, Image format files can be utilized again.
A sample of the Image format file as below.
IFORIVIAn
COMMEl\fr=Nepal
AUTHOR=OMDEC

FORMAT=,

; I:BSQ, 2:BIL, 3:BMP

SEPARATE=O

; This data is BSQ only

HEADSIZE=512

; File header size

PIXELS=3600

; Data Pixel Size

UNES=2499

; Data Une Size

RECORD=3600

; Data Record Size

OFFSET=O

; Data Record Onset

XDISTANCE=20

; P^el Distance (in)

YDISTANCE=20

; Une Distance (in)

IBANDl

; Eficacy Band

BANDl=YES
BAND2=YES
BAND3=YES
\

BAND4=YES
BAND5=YES

BAND6=YES
BAN 07=YES

, .

2 - 11

O:Continuous type, I :Separate type

IASSIGNl
TM, =I

TM2=2
TM3=3
TM4=4
TM5=5
TM6=6
TM7=7

All information of Bn, , BSQ or B^^IP file structure is copied to the Project file.

. ,

2 - 12

Expert Sheet

Project Sheet

FCD Method

Setting up of the New Project

Project Name
Reference E:coerL Sheet
FCD Method

Threshold (ALL or etc

Image Format

Image Input Process

Cloud

Noise reduction and Normalization

Equation Parameters

Image Format, Band

thinly red Area hito. etc

Noise & Reduction Info.

Mask fineshold

AVI, 1.1Vl, Awl, BT, SI

Index Process

Equation Parameters

Made Index Info.

TI Calibration Coef.

PCA Parameters

Vegetation Density Process

Made un info,
VD TIEeshold

Clugtering Parameters

Scaled Shadow Index Process

Made SSI Into.

Cluster Result Info.

FCD

FCD Process and Map Statistics

Equation ParameterB

Made FCD Info.

Accounting FCD Map

."

..,

,,
.,

,,
.

b
..

Figure of the relation between E>cperl Sheet and Project Sheet

2 - 13

2.1.4 Operating environment

Computer:

DOSV/windows'95, 98 or NT type

CFIT:

Pentium 1.33 MHz or above

Memory:

32 I^^ or above (64^IB recommended)

Hard Disk Capacity:

Free-space 500 MB or above

(for processing of TM full scene, IGB or above

recommended)
Image Display:

1024 X 768 pixels, 641< color variation or above

Printer:

Windows compatible type (color/monochrome)

Sub-memory Media:

MO-Disk 230^^IB (6401i^IB recommended)

..
..

..,

2 - 14

2. , .51nstallation Guide
I. Starts a Windows.

It is strongly recommended that you exit all Windows programs before rumxing this
Setup program.
2. Sets the CD-ROM disk of FCD-Mapper in the CD-ROM drive.
TTLe Setup window is displayed

\
\

2 - 15

3. 011cl< Next

This User Information dialog box is displayed

Type the Your Name, Name of the Company and the Product Serinl number.

Torn. Forester

JOFOA
ABODE123456789

Example:
Name : Tbm. Forestsr

->

Your Name

Company : JOECA

.>

Your Company

Serial : ABODE123456789

->

Your Product Serial Number

2 - 16

4, GIId< Next

The Choose Destination Location dialog box is displayed

Tb install to a different directory

(1) GIIcl^ Browse
The Choose Directory dialog box is displayed.

^a c\

^ Program F1
;^a fed
I^ ExTe$11

I^I
^C

(2) Select another directory and click OK.

.

, .

2 - 17

5, Confirmed install directory and click NexL

TTLe Select Program Folder dialog box is displayed

^,

Adob
ATOK9, !Ii
AutoD^D R14
Borland C++ 4.0
ER M
5.5

FCD-Mapp
GPSil. >,, b BR-5000

6. Confirmed program folder and click Next.

TTLe program icon of FCD-Mapper is registered on the installation folder name

Especially; there is not a change of the program folder, being necessary.
The Start Copying Files dialog box is displayed.

\^!.*.

2.18

I, ^I, .

7. Click Next

The installation of FCD-Mappei: is begun and a progress situation is displayed. When

the installation completed, the Setup Complete dialog box is displayed.

8, Click Finish.

In t}lis, the installation of FCD-Mapper is endirig.

L
\

2 - 19

2.2 FCD Mapper Operations Guide

2.2. , Start-up and Registration Process
A. How to Start the FCD Mapper
There are two ways in the starting.
I) Storts from the start menu.
. Clid^ Start in the task hat' The Start MeniL is displayed.
. Adjusts a mouse pointer to the Program. The Program Menu is displayed.
. Adjusts a mouse pointer to the FCD. The FCD Program Name is displayed.

. Clicl< FCDMenu.

FCD Mapper stsrts and an opening window is displayed.

2) Sin^ from the Icon. .

. Clicl< FCD Mapper Icon on Disktop.

. FCD Mapper storts and an opening window is displayed.

,. ,

2 - 20

Menu Bar

"..*^. ,~.,,,,~,**

s~,~,

-.

~.,,^

'

'

a
..

un. ,,,"~\,,...^",

FCD
Menu

"
*

a
,

;a

,.

I11
.,

linage display window view area

Indicator

NO Exist (Black)

Exist

Invalidate (Yellow)

Deleted

(Blue)

(Red)

b
..

, .

2 - 21

B. Registration of the expert file

When an e, ^liert file and a project file aren't registered on the system after installation.
If you execute in New Project, the following message is displayed.

!"

",

It registers the e, <pert file in a following way

(1) GIIdc OK on the dialog box
The File Choice dialog box is displayed.
I.

<:^ Fad
. Dind

r==

exp

E, IPeit FilesF. expl

(2) Choice the e>^pert file of the Installfold. er and Ond^ Open.
.

2-22

C. Setting up of the Project Name

C. I Designation of the project name, for the first step

The first step in implementation of FCD Mapping is to designate the project name. The
project records the processing pronedtire, the reference expert sheet name and the name
of the person performing the analysis.

Each project is arranged or managed by project name, When the project is registered by
its designated name, this makes is easier to cond. 11st the analysis, produce the FCD
maps, do the necessary promssing, change the processing procedure if required, and
resume processing after stoppages due to rest breaks, or for other reasons.

C. ,.'I For the first step to make FCD Map, it requires to register the Project name

For the first step to make FCD Map, sets the project name, the e>;pert sheet, the
arialysis parson name and so on.
(1) 011cl^ New Project in the File menu.
The New Project dialog box is displayed.

(2) Choose the expert sheet to ref^13 to the new project from the Base Expert list.
(3) Type the name of the new project in the New Project Name box.
(4) Type the project file name in the File position box.
(5) Type the user name in the Author box.
(6) T;, pe any specific comments in the Comment box
(7) Click OK.

.
. .

2.23

Opens the project

and closes this dialog box
Lists base expert files

Closes the dialog

without doing anything

Type the new project name

EC

OS\, CODA"A\PCD\PCO. Pro

~' ~ , **, ., .-, ...." *.***~'."'"~~~ " " '

r . . ' .,~~~~-",' ' " ' ' ' ' ' ' ' ~ ' ,,' '.. : ' ' - '~"' t . ~ ~-; ..,. !

*,.,~ ,,

Property of selected base expert

~. - -. . -*

Selects required or type author of project

Type the comment of project

Type the project file

2 - 24

C. ,. 2 The case of Re-Open the project

When you resume processing after any pause or Interruption of a previously-registered

project, the work begins at this place (i. e. at the file name). Here is where the name of
the project will appear, along with the map as it appeared at the time of pause or
interruption. 'IIJis is also the place to make changes in the processing procedure if
necessary.

(1) Onek Open Project in the File meriti.

Tlie Open Project dialog box is displayed.
(2) Choose the project from the Select list.
(3) Click OK

.
..

2 - 25

Lists project

Property of selected project

Opens the project and closes this dialog box

Closes the dialog without doing anything

.
..

, .

2 " 26

2.2.21mage Input Process

A How to Input Satellite linage Data
This means designation of the format, the file name, and analysis range of the input
Image.

co Click linage Importiri the FCD menu.

nthe linage Import dialog box is displayed.
(2) Because the linage Import window is displayed, input the format, the file name, the
analysis range of the image.
(3) Click Import Execute.

2 " 27

8Z ' Z

- *."~~ ~,~~".~.~
,&

313
Is

it

13

as

,
I

$;
^g:
E

o
^

a
g
,
,*.,.

".

L
*

der ago. r
re

^!;

,.~~,~...~~~.*

A1 Designation of the format of the input image

This mean^ selection of the format to be used (i. e. BSQ, BIT, , or B^^IP), the size of the
image, and the band of the input image.

(1) GIIcl< Import Format tab.

(2) Choose the ^rinat of the data from BSQ, Bll, or B^^"P.
TTLe data is the format which separates each band. in the TM data, general15^ there
are 7 files.

Note:IIJ BSQ format, image data is composed of I file. Therefore, chedc One File.
In the Bit^ format, data from each band appears on a separate line. Every set of
data is stored in one (1) separate file.

BlvlP is the standard image format of the Windows program, Only bitmap files
with 8 bit format can be loaded. Till. e himerap file includes a color palette.
However, ignore this and only input the image data.

(3) hi the Header Size box, input the number of bytes to the head of image data from the
head of file.

(4) input the number of bytes in I record to the Record Length box.
(5) Input the number of pixels in the image data to the Pixel Max box
(6) Input the number of hales in the image data to the Line Max box.

(7) In the Record Offset box, input the number of bytes to the head of the image data
from the head of record,

(8) Input size per pixel to the Distance/13ixel box by the meter unit.
(9) Check the image data number of the bands to the 1-7 tilthe Import Bands.
(,. 0) Procedures for saving the set format information.
a) Input comments to the Comment box
b) input the name of the person doing the analysis in the Author.
c) 011d^ Save.
d) Triptit the format template file to the File name box.
e) Click Save.
(11) Procedures for inputting format information that was previously saved
a) Click Load.
by Type the format template file to the File name box.
c) Click Open.
.

, .

2 - 29

Input the saved fomat information

Save the set

format information
512

3600

3600

2.99

aq

Closes the dialog without doing anything

Imports image data and closes this dialog box

Exsample Import linage Data

Record Lengh1(3600byles)
Header

(542 bytes)
^

-.

Lines

I^I

Pixels(3600)

Record Offset (0)

(Pixel Max)

(2499)
(Une Max)

Distance I Pixel : Ground Resolution IFOV

,,

43, , _

.,

I^ Pixels(3600) ^
(Pixel Max)

2 - 30

Is ~ Z

, .

..

..

'p^eq tjot?a Joy emui:I 819 aq. Trim. aretl^'SIIT:IL

'(!, co I '8'1) SJOqTLrnt:I (!,) troAos I{'11/6. arueT:I o19: B q. noseJd ATt8011^rungnB TIP^ SII^

'81!tj: emus ^10/10 nan^ pun arugtI o19: at10 ^rider! co ,. s^:g s! SIq^ op co KB^. oldtrr!s axilL
,xoq Lows. rowa

aqq. un oareu o19: aq, . adjC, . 'B, .lap punq. tt0120 30 q. Emuoy I^Sai Z0 <1/1^:81 10 aspo all, . LIT (8)
'Uad0 >10TtO PUB Xoq. arueu all^ all^ co emutr

o191 ,. ridtr! '- att^ ^10/10 itOA nett^. JPedd^ rip^ Adoptr!/!,. elec unduj pales atI, L:alON
'xoq. Lowg att^ co acnetr eTg: ^ridtr! 'o19 I aq^10 t^sai zo TLB:10 egg0 111 (^)
'qe^ all^ uodLL11 ^101t0 (1)
abetL^! undu! aqj!o arueu ajg. aLjj 10 uo!Ieu6!Sao aV

Selects required import file

CE\, CODATA\.^a

The Select Input Data

dialog box is displayed

Imports image data and closes this dialog box

The file name has numbers ( I - 7 ) automatically

( Case of BMP or BSQ of the each band data )

2 - 32

B. Selection of linage Area to be Arialyzed

It designates an input range and the sampling interval Alternatively the range to imptxt
mm be designated from the display image. Procedures are as follows:
(1) 011cl^ Select Area tab.

(2) In the SLat Pb, ell>DJ^ type the number of the begi^pixelin the mmy^s are^
(3) Also in the Start Line box, type the number of the begi^ line in the analysts area.
(4) T\, pe the input interval of the data reduction to the Interval box.
(5) After these steps have been completed, the ending position and the size of the
analysis area are calculated and the results are displayed in each box.
(5)-I Procedures for inputting the ending position of the processing object area
a) Type the number of ending pixel on analysis area in the End Pixel box.
b) TTLe number of pixels in the analysis area is displayed in the Pixel Size box.
c) T;7pe the number of ending line tilthe analysis areairithe End Une box.
d) The number of lines in the analysis area is displayed in the Line Size box.

(5)-2 Procedures for entering the size of the analysis area

a) T^. pe the number of pixels in the Pixel Size box.

b) The number of ending pixel of the analysis at^^a is displayed in the End Pixel box.
c) Type the number of inIes in the Line Size box.
d) The number of ending fine of the analysis an^ais displayed^Ithe End Unebox.
(5)-3 Procedures for displaying the image, and setting the analysis area by visual
observation

a) Click All Area View.

'111e Check linage & Select Cutting Range dialog box is displayed.
b) Clro^the at^^areaby, ^rigftiemoi^elefrbtmmto^g&^pont^d^layi^.
c) Tlie value of the analysis area is displayed at the Start Pixel, End Pixel, Pixel Size,
Start Une, End Line, Une Size box.

(6) The size of I. pixel is displayed in metric units in the Distance Pixel box.
(^ When the amIysis area exceeds 1280 pixel x 10241ine, ched< the Thai Mode,
The data is redticed to be stored in it, and inputted to it,

(8) Ond< the Select Area View, the amIysis area will appear.
(9) The free space of the current drive and e>:^)ected. space to be used for conducting the
FCD calculation ending are displayed in the Disk Free Space Information,
.

CTlie current drive is the drive which stores the project file and an all kinds of image
processing files. )

. .

(10) Click Semi Auto. The processing is coined out semi-automatically according to the
contents of the process file, consistent with the meriti specified by the author after
inputting the image data. If any marinaloperation is to be conducted along the way;
semi-antomatic processing will continue until reaching the point where manual
operation is designated.
2 " 33

The select analysis area is displayed

(5)-I, (5)-2
Type in

aLa

2.09

2097

L5,9

(7)
1280pixels I 0241ines
(Sampling Full linage)

(5)-3

Import image data and closes this dialog box

The Check linage and

Select Cutting Range window is displayed

\
at

. ..

321

2289

ex

$2

48

e;

Choose the analysis area which Drag & Drop by the mouse left button
2 - 34

2.2.3 Noise Reduction and Normalization Process

The normalization of the image is the process through which all data is adjusted to
conform to a common standard. It is necessary because of variations in shade on

different parts of the image alle the time period and site specific fleatures of the place.
(1) Oncl^ ChecklReduction in the FCD menu.
'11^e Noise Check and Reduction dialog box is displayed.

(2) Ched^ the related item responsible for the noise existing in the input image.
(3) Clied< the item that has to be masked in order to redlice the NOISE, such as water
area, aloud, cloud shadow and so on. Tb process, click ON.
(4) Click OK

2 - 35

a
, ~~~~~"~*~ .

a
*

$
,
S Chechinedubct, a

<6*%I
~

a:

; Sag

a^^;

^ #

; #

! ^

;;*;

.~ ,,,. .._,,__._ ;$;

'..^!...$^I

Checks the noise and sets the reduction to Noise menu

and closes this dialog box

Check and ON

~~"...,~,~~
.~~~,"~"~

;
I
} I:beck/Red!, CF*on *

;
^

11
1:,".~ .~ "J

2 - 36

Active
mode

A Procedures for masking bodies of water

TTLese procedures set the value of those portions of the image where noise exists because
of water, and remove (i. e. mask) the area.
(1) Click Water Set in the FCD menu.
The Set Threshold of Surhace Water Mask with Histogram dialog box is displayed.

(2) Drag & Drop the cursor bar on the histogr. am graph of the Set Threshold of Sunace
Water Mask with Histogram window with the left mouse btitton and move it to the
thL. es}101d of the water area.

(3) TTLe portion. of the image below the t}Ireshold of the water an, ea designated in (2) above
is displayed in the black ( the mask area ). Everything else is displayed in white.
(4) GIIcl< OK

, ,

2.37

FCD Menu

Uace^ Sec
,

Drag & Drop and it move a cursor to the threshold

ex. Landsat-5 TM

Normalized linage

Band 4 Histgram

Surface water mask Threshold

Pixel count

Band 4 statistics

Makes the surface water mask and closes this dialog box

~..~.

~~,~.~..,

..--.. ,

d
*,,

^!^

.
..

Black : Suitace water mask

{Pixel Value ^Threshold(ex. 11n

2 - 38

B. Procedures for masking clouds

Tliese procedures set the value of those portions of the image where noise exists becallse
of clouds, and remove (i. e. mask) the area.
co Click Cloud Set in the FCD menu.

file Set Threshold of CLOUD Mask with Histogram dialog box is displayed.
(2) Drag & Drop the cursor bar on the histogram g, :, aph of the Set Threshold of CLOUD
Mask with Histogram window with the left mouse button and move it to the
tm. eshold of the dolld area.

(3) T}Ie portion of the image below the threshold of the dolld. area designated in (2) above is
displayed in the black ( the mask area ). Everything else is displayed in white.
(4) Click OK

2 - 39

FCD Menu

Cloud Set.

^:^

Drag & Drop and it move a cursor to the threshold

Cloud linage
(Made from Normalized linage)
Histgram
Cloud mask Threshold

Pixel count

Cloud linage Statistics

Makes the cloud mask and closes this dialog box

,^^
.
,.

. .

Black : Cloud mask

(Threshold(ex. 149)^ Cloud Value)

2 - 40

C. Procedures for removing (masking) the cloud shadow area

These procedures are automatically synchronized with the procedures for masking
clouds. As e>CPIained. above (B. ) the aloud mask responds to Drag & Drop movement of
the left mouse button. T}Ie semi-e><pert system prod. Lices point information (i. e. the
extent of movement on the image ) of the cloud shadow, which of course relates to

movement of the doud. . T}lis point information is used to create the cloud shadow mask.
A maximum of three (3) information points can be derived to delineate the area of the
cloud shadow. Basically; the right button menu of the cloud shadow mask is made by

adding 1/9'oil^t Seleatl, neoint Initializel, D^Iaskl to the light button menu of the false
colorimage.
(1) 011d^ Shadow Set in the FCD menu.
T}Ie Set Threshold of CLOUD Mask with Histogram dialog box is displayed.

(2) Drag & Drop cloud mask ( displayed in white ) on the Set Threshold of CLOUD
Mask with Histogram window with the left mouse btitbon.
Tliis removes the cloud shadow in the area encompassed by the information points.

The points to be set and a cloud shadow position (i. e. the extent of movement on the
image) are displayed at the status bar, .
(3) Choose the point of clottd shadow mask from either Pointl or Poin12 or Points in the
Point Select display sub meriti.
(4) When the^e an. e interrtiptions or pauses in the analysis, it will be necessary to return

to the point whe^e work stopped (i. e. Tetui. .rimg to the position which was loaded from
the project file. ). Tb do this, choose Point Initialize.
(5) When confi. Linertrig what cloud shadow mask was made at setting point, choose the
Mask. The cloud shadow mask is displayed in the black.
(6) Click OK
i

.
..

2 - 41

Point : Set Point number and the moved position from the image
Zoom : Magnification of display
^

r~'~~

a
&.,,,..~

9h.

a,

;
g
>,

g;
.A. .

Makes the cloud shadow mask and closes this dialog box

^
be

iai
*. J

shado. sec !
*

; I ~ t, ' ' " '

*

i*^

62

in

".
,

^..

White : Cloud shadow mask

2 - 42

us

22.4 Index Process

A Designation of the Vegetation Index 6/1)

The FCD Model chooses a vegetation index from the relation between A\71, NDVl, AN\71
and Bare soil Index (BID.
(1) Click PCA Select in the FCD menu
AVI, NDVl, Awl and Bl are made.
The Select Vl dialog box is displayed.
(2) Choose the Vl from AVI, NDVl and ANVl. Then chcl^ OK.

2 - 43

Select Vl and Bl distribution

.*.~,,.~,~~

~.^~-~..,,,~,^

I
4,

I:

~ mm ~.. ,

a
,
a
*
*

PCA', e, EC^
a

nat

Ib. ,

1:8,
~

.~*.~.~~

a, Iai

Sets Vl and closes this dialog box

Bl
255

J .^

~*

e-et o- D' ns
-,

,~,

00, o
e"r

.,.

*,

~*, a

,*~,
*
a
\';,.;:: ~ ' * *by ~J ' . $1 ' ..:* . . *

'''*t'***^:';;'I."'*"z:*>'~"'
;:*?*'**~ * '*'"*'*'*";$;$
*Fit*:,,:.;*:*. -v:t;,>.\\,**14:39t, *:,,,*. -.*"%\* . .As'* '*-\*,-;r-~#;..;;a
4:<: ;-*:..*L*:"-, ./:!**;15, ::'-*$~, r:,, t"*:***;t, .*,*,.., ,. **,,\ %.. 4*r:,*{**,,,
'

..

-.

_-*.

-...,

,..,

,\*,

,J,

.*.,

..

"'

'~

Vl

255

2.44

B. Calculation of the Thermal Index troy

(1) Cinch n in the FCD meritx.
(2) Tl is calculated and the False Select Window dialog box is displayed.
Choose the pref^rred. method for displaying the data and clid^ OK.

2.45

^-..,".-.. .....~.."

3
*
re
. ~"~..~,.".,..,,~.*~ *

is

"
*I
,
t.
.,

"

"

!;

,:
";
-,

,~~

~.~

"

a
,
e

,,
*

S
TZ

,
51
,
g

F;

*
I
*

*
,

.-:a
~

.,,,

;**
~;S^

..
..

2 - 46

C. Calculation of the Shadow Index (SI)

Calculation of the shadow index involves a choice among of two kinds of normal

equation or a log eqtiati. on, both of which are included in the semi-e>^pert system. T}Ie
system also revises forest gap and black soil information. Additionally the system can
calculate the Advanced Shadow index (As^.

(1) Click SI Select in the FCD menu,

The SI Select dialog box is displayed*

(2) rib calculate SI, choose either the Normal Equation or the Log Equation.
(3) rib calculate Asl, choose either the Gap Detect, Black Soil Detect or Special
Process.

(4) Click OK.

.
. .

2 - 47

g a ec
~~~~".~

a
a
,

*,,~
3

*
*

,-

~a

"
a
*,

*:
e

;^!
:^I
*~

ST SeL

;;^^
11:3
;*
,+-.**~* ~_

it^
Sets the SI generation method and the Asl generate option to SI menu,
closes this dialog box

I'

I
,

2 - 48

C. I Correction of forest gaps

Quite often, there are Forest Gap areas in the analysis range. The semi-e><pert system
demieates the boundaries of these gaps by calculating their value through use of a
Vegetation Index.

co Click Gap Set in the FCD menu.

The Set Threshold of Gap with Histogram dialog box is displayed,
(2) Drag & Drop the cursor bar on the histogram 'graph of the Set Threshold of Gap
with Histogram window with the left mouse btxtton. and move it to the threshold of
the Gap.

(3) Tli. e portion of the image above the threshold of the gap designated in (2) above is
displayed in white. Everything else is displayed in black.
(4) GIIcl^ OK.

2 - 49

FCD Menu

;
Drag & Drop and it move a cursor to the threshold

I'- GAP Sec

^~.^,^."^,^

^:^

Vl linage Histgram
Forest gap mask Threshold

Pixel count

Vl linage Statistics

Makes the forest gap mask and closes this dialog box

.\

...

...

,...

..

;*,&
'/' , ,^,,
,*;' -.,,;'
,t^ ' ;;:,
"": ...""
:;^j, ,;;; :,.^?-.,:;^"t,
',,'.;{.;!,..^:j
Fi, .t. #,1'^
;*:;&,^a;6,.;6,. ^::;'
.' ,$" .,;;^11;:-,
I. ,j ;*-:
if. .;!'i:2",!?r::*I;.
\

^.,* ' '~:*'

^..,*
.;*. .^^$^^^-..,.
.;#it?^ ~'^...,.*j,-,,,--, ,jI-*.-*. - ,^.,{-.
,,^,,,-*------.*- ~ .-* ,:-.,^...;^:
-*..^^,..;^.* $314
^#;4

,.,'.:^;:..$13:*., ,^"-;;.;.*f
^j. <;,':,'^^;;'*,^,:^^;:31".
". a>!,;,.,,$,
-*~*-;,-31.*, .,-.,;^^,j^?"4it^' #;!*-!' ~^ ,^^!;^3,
ai' ";^'r; ":#"'^^::'^,' '^*;.,.""'--;':^",^^;^;-.*..*;^'^*\;;4.
**>--- *,- , ..*. j-.,-*';:-*,,;,!*
.",.,.
. J . . . ... .;" *:.*..";'*"%:,', ,* ^^ * ' ; ' " ' '^ "'I' ' '3"" *"' ' ' ' '^"'

I : " ' ' ,!;:.-:.- .-!,*- . .y :$:I :,, " ;,;!';;b-^,^;jin-'s ,.. e :

; ..-;,I -f,,,,:
.^^.,.;.^..'+,'*~\::,
I;;Ai^I~:~\ -',^z;t
~~' "' -"'
";" "" ....
...*,,.11-.*'^
1<:-.}_,,
.,,,^::;,:.
* ~ 14/1^:,#*-

1.1.:"-.. '"~?;I'
I.
I I , '/! ""I 'i,
" '; ;^^;',";'^-;i;'-,;,,!,;+.!^^.'
,-;;,!;^; ";': ,
.,

,j

I,

4',

./.,.,
:
,..
..

..

4".

%
\,

..

,.

*
$ t

White : Forest gap mask

(Threshold(ex. 114 ) ^ VI Value )

2 - 50

I'

C. 2 Procedures for dealing with black soil

111 many instances, the area being analyzed contains patches of black soil, T^lese patches
can be confused with shadow. TTLe following steps are taken to eE^late this collfi. Ismn,
(1) GIIcl^ Black Soil Set in the FCD menu,
It'}I. e Set Threshold of Black Soil with Histogram dialog box is displayed.
(2) Drag & Drop the cursor bar on the histogram graph of the Set Threshold of Black
Soil with Histogram window with the left mouse button and move it to the threshold
of the Black Soil,

(3) The portion of the image above the of Black Soil designated in (2) is displayed in the
white. Everything else is displayed in black.
(4) Click OK

2 - 51

FCD Menu

B, .86kSo, .I. Sec

Drag & Drop and it move a cursor to the threshold

\^:^^
Tl linage Histgram

Black soil mask Threshold

Pixel count

Tl linage Statistics
,
.

Makes the black soil mask and closes this dialog box

^.

.-

..

, *>.,* L*,:,^i, ,: ' " " * . .

.

_ I, ^,!/\'
,, ,., -'^;;:'j'
'Iv^;!!,-'
,, .,-,
: ;:.,. .,.' -,~. ,." "'

..

...

..

.-

..

, ,.*
": "~
' '.
" ,\
'^;.' .' .' ..'
'If.

*,,, , \ ,,

,
.

..

., , * ^- "" '; . .' f

.,

L.

. I

,.

\,

.,
.,

..

,*

4.1.
".

. .

^..

C
,

.^

;r

*'

.* ,

,.

,,

L,

.. r' -. ?^,, .. \*'

, .
,.

~:

.,

>.

White : Black soil mask

(Threshold(ex. Ian^ in Value)

2 - 52

,.

~., . :.

C. 3 Calculation of the Advanced Shadow Index (Asl)

When revising the Forest Gap or dealing with Blacl< Soil, the semi-e>:pert system
calculates AsT.

(1) Gild< Asl in the FCD menu.

(2) Asl is calculated and the False Select Window window is displayed
(3) Choose the pre:fi^ited method for displaying the data and chik OK.

2 - 53

a
.",~~..,,~.~~.~..

*... . -.~ .~" ~ .,,,. , ,

~-..-"~ "-~,,., ~

:
I

%
$

*
*
,
4

e
,

a;!,
I^.

AsT
^..^..

aa

\.

2 - 54

22.5 Vegetation Density GID) Process

Tlie system calculates the density of vegetation by the percentage.

A Designation of the vegetation area

Tli. e system designates the minimum value of the vegetation area by Vl.
(1) Click Vegetation Set in the FCD menu.
TIT. e Set Threshold of Vegetation with Histogram dialog box is displayed.

(2) Drag & Drop the cursor bar on the Instogram graph of the Set Threshold of
Vegetation with Histogram window with the left mouse hiltton. and move it to the
threshold of the Vegetation.

(3) T^Ie portion of the image above the Vegetation threshold designated in (2) is displayed
in the white. Everything else is displayed in blacl^.
(4) Click OK

2.55

FCD Menu

vege"actor! Sec

^:^

Drag & Drop and it move a cursor to the threshold

^

Vl linage Histgram
Vegetation mask Threshold

Pixel count

Vl linage Statistics

Makes the vegetation mask and closes this dialog box

*t;- ' <. 27

I;

.-

. . 49'F, : " '

?$
^

a.
.~

I
,

**,

.*,,

,,

\:^;^}

I*E

*^*; , , , ,.

a\;

I.:
- .".
' ^*...
... ^,^,,,
I
.~~.
", ..

b*

-.*

'; ;;^^#I.

",

63e,

,,$,;:I :t'. '. ,, t' "%;" . L I

" * ,f:I .I ,,.,

43j:^^;!,^,^;^
$3^. ^
'";.,:;,\; " I

,,

..

"..
,

if~., I- 1.1 !' '

I^;:,*., .I^ ^!If

*.

,.,. ~ I, , , ,. I

' . . '. ' - 1.1, '-." "' , '..

J, .,

;;,,.. . ,,. , .F1^'

?.
' 1:1^*?:;.-$, ^,:
I

! . ,. a ;. \' .., I.

I ".!'* i' ;' t FF ;:';! - ; '

41

'. ..*.. - ,....

. . *::^^t^, 4. -, *.
I
If^

,; atI, ' ,',.. ' *.:'

at"^*
\"J , ~>. . ~

."

.J, ... J

?z, ,--.. :. -..

I
I

White : Vegetation mask

(Threshold(ex. 125)^ VI Value)

2.56

B. Designation of the Bare Soil

The system designates the minimum value of the bare soilby Bl.
(1) Click Bares oil Set in the FCD menu.
The Set Threshold of Bares oil with Histogram dialog box is displayed.

(2) Drag & Drop the cursor bar on the Instogram graph of the Set Threshold of Baresoil
with Histogram window with the left mouse button and move it to the threshold of
the Bares oil.

(3) The portion of the image above the Bare Soil threshold designated ill (2) is displayed in
the white. Everything. else is displayed in blacl^
(4) Click OK.

2 - 57

FCD Menu

Bai:e, So, .L See I

Drag & Drop and it move a cursor to the threshold

us

Bl linage HistgTam

Bare soil mask Threshold

Pixel count

Bl linage Statistics
.

Makes the bare soil mask and closes this dialog box

*
*

:I^^,11. .

,,, ' *$1

I .'

t *. -,~

,, t*:: ;. I\

~.
,,

C.

I:$;-.*4 , ,'<. , .e ,. . ,. .,.,, ..,.

'""'^:$;\^;^*,,
..
,
444
,,.,,,,,
'
'
'
'
!$,$^;^^^;,^^;;;^^;^;*;^.^^^i^^;..-,^i^'^-,:#""
. ';,**t;I^;^. " *\!:;, f;^!'-,.^'^,,,'^;\:*;^:?^.*':$,;^^:^^;'^^^{,^ .#^;^"^^. '\^;:~'

'^;:'"'-\'.*, .' $.-:'.\I. , .,^*,,.?$!,:'71.5!;:^; ^^!34:1:^.'^;^'i:^^^^;;,;^;^:^: \;.,;^*,;-** I

White : Bare soil mask
(Threshold(ex. 130)^ BI Value)

2 - 58

:,

C. Designation of the range which calculates the VD

It calculates VD in the designated bare soil area and designates that vegetation density
at that part of the image LS O %.

Next calculates VD in the designated vegetation area and it designates where the
density 100%,

co Click VD Range Set in the FCD menu,

nine Set Threshold of VD with Histogram dialog box is displayed,

(2) Drag & Drop the cursor bar on the histogram graph of the Sei Threshold of VD with
Histogram window with the left mouse button and move it the minimum value of the
VD.

(3) The portion of the image above the minimum value of VD designated in (2) is displayed
in the white. Everything else is displayed in blacl^.
(4) Click OK

(5) Drag & Drop the cursor bar on the histogram graph of the Set Threshold of un with
Histogram window with the left mouse button and move it to the maximum vante of
the un.

(6) Tlie portion of the image above maximum value at the VD designated in (5) is
displayed in the black. Everything else is displayed in white.
(^ On(,^ OK.

..,

2 - 59

FCD Menu

Drag & Drop and it move a cursor to the threshold

" ;
un Range Set

^:^

Set the minimum value of VD dialog box

Set the maximum value of VD dialog box

be

0<

Vl:Green

Bl:Red

VD Minimum Threshold

Vegetotion pixel count

VD maximum
Threshold

Bare soil pixel count

I^^
:
e.
",

It move to set the maximum value of VD

},,
-*^I

Makes the VD mask and closes this dialog box

A
. ,.:
,.: a
all,,.
!,- },
}, '\
\ r,
C
*;,'

''~" "" ' "'^:;\,- \{^^:^';;\,'.:$-.;;

,,,

,
.*

,;i;' -, u, **.*^ >. ,

;.*^:t-^;,',* ;:::$*; , "a", ' ' ' ""

$

>t
I

;!,,*,^;^"'."-"'*}";;$*'
'. ' a~\, , ,
;. ;; 'iff;;iE:' ,!,*?*j** *- .?<,

...

\
'*us

-,..4;;-.,,
,: ., t. . .*,,,-^^:;i
. , I' I':--'?'*^1:4f
. '^. .' -t?'
A; :'I'
' ' ' I. -,- .. 7-1, . "

. ^.!"

3:0'..,,

3, 'I.

' ', , , -"'!'in I, ., , *,

.. . ... .. ./ .,

!^@

"" $,;,? *.- . .,",.".

'*;*"
' - "-'*-^*:6"*V, * " ' ' ';I
A.

' " " .*#,,,, ,?! *. . , , ^;^;: .'

,..

.,

-..

~*

,.

I'
"

-I
.

. Case of set the minimum value of the VD

White : Above the minimum value at the VD
(Threshold (ex. 159) ^ VD Value)
Case of set the maximum value of the VD

Black : Above the maximum value at the VD

(Threshold (ex. 214) S VD Value)

2 - 60

D. Calculation of the VD

(1) 011dc VD in the FCD menu,

(2) un is automatically calculated and Pseudo Select Window dialog box is displayed.
(3) Choose the prefi^rred method for displaying the data and chik OK.

2 - 61

e a be
:.

"

$5;

in, ,

-*

*;

.?

13, ,,~..-,..,~,-~.. ~

~A

.,
+
a
*

HD

"

g
*

~." ." . , , , , .~ . ~

,,
.."~~ . .

' ' "'**,*,"' '"'1-<4

~

..*Ticti=Cast*;;',,*.;::a:

2 - 62

2.2.6 Scaled Shadow Index (SSl) Process

The semi-expert system first classifies a cluster using Vl-Bl-SI and the Tl image. It then
identifies a forest area,

It next adjusts the range of SI in the forest area from O to 100 % and calculates SSl.

A Confirming the conditions for processing SSl from Vl-Bl-SI false color
display
The system changes Vl-Bl-SI into GBB and it displays in ithlse color image. It also
confirms the mutual relationships of Vl, Bl, and SI. The following :foatures appear on
the false color image:
- the area off{i. gh Density Forest is displayed in the cyan.
- the area in grass, agricultural crops and so on is displayed in green.
- the area in bare soil is displayed in red.
Tb confirm the conditions and move ahead on processing, the following steps are
required:
(1) 'Ib display the false color image of Vl-Bl-SI, dick Vl-Bl-SI in the FCD menu.
TTLe False Select Window dialog box is displayed.

(2) Choose the pref^Trad method for displaying the data and clicl^ OK
TTie Vl-Bl-SI fo. Ise color image is displayed.

2 - 63

,"

*
"

* .-.
~.

*
*

a
*
*
.,.. ..~"~,,,~.~~,"*

; ^
a

~~~~~~~

! or-BE-SL ;
a

, ,,~~~",~~~~
J,

"

I.,
~'*^*^^"
-^!.:^,.
, ;fi*txt;63E,

5.3^j;I;^^:;\;i; ;ii^

,
\

.
. *

2.64

B. Designation of the forest area

TTLe system performs cluster classification using the Vl- Bl-SI and T, : images and the
result is displayed. This designates the classification category of the forest and imp tits
the forest area.

(1) Click Cluster Select in the FCD menu.

(2) Cluster classification is done and the statistical result ( Clustering of Vl-Bl-SITl to
identify Forest Clusters(FC) dialog box ) of the classification of ass and the cluster
image are displayed.

(3) On the Clustering of Vl-Bl-SLn to identify Forest Clusters(FC) dialog box, dick the
classification class of the forest area with the left mouse button.

(4) Tlie chosen classification category is displayed in the green.

(5) Tb cancel the choice dick the left mouse button.

(6) When all classification categories of the forest area have been chosen, click OK.

, .

2 - 65

FCD Menu

I^

FC : Forest Clusters

It thinks that it does a forest area as much as the big value

Average of class

IIS

106.5

1.37. O

co

L33.9

L2

147

2.21. .

147.5, }}^:}I

L52

I
ISO. 6

163.2

L48.0

LSI. 9

LSD. '

I
\

Makes the forest mask and closes this dialog box

Chosed forest class is displayed in the Green

\
\
\.

$.. t

,
A*

,*

.F, .
' .~*.. .,
;,., *!;.:

~,.

*.~ ,*;,

*&, .;.',>. , *,

*;
.-.*

* ~.*

;*;;.

Green : Forest mask

(Ex. Class number 2 - 4)

2 - 66

A*'*$'*\,* ' . -.
-

-.*

C. Calculation of SSl
This function of the system adjusts the range of SI in the forest area from O to 1.00 % and
calculates SST.

co Click SST in the FCD meriti.

(2) SST is calculated from the designated forest area and Pseudo Select Window dialog
box is displayed,

(3) Chose the preforred method for displaying the data and dick OK

2 - 67

a
*
,

;^;

*
".

;;^

.^.,

*,,,.-.- - -,-.-,~. -f :'

,*;;^
,.

*!

DC

$
~~~..~~.

,"~

a
\

SST

I:

,*,

g
~

*., A. ~~~-- . ~,,- -~

,.

=L::****.,*: :,**-*.:

~ r 9j~ .*,*3< 31

';.;^j*-:$a**?;!
in ' ~<3:1:,.-:;..;; ;: ! ;.

a, ,' ":;:. :**:'::;* "

-~

.,.

83n^*.\.;;4*;;:*

..

. .

2 - 68

2.2.7 FCD Process and Map Statistics

This part of the system performs the following functions:
- Takes statistics FCD every 'Grid or polygon area.
- Presentation of the data in number of hectares per FCD category.

- Saving of results in an Excel file format

- Saving of images in a Biim. ap file (B^^lip) format.

A Producing the FCD Map

'11/1e FCD map is prod. Liced using data from the Vl and the SSl.
(1) Click FCD in the FCD menu.

(2) FCD is calculated and the Pseudo Select Window dialog box is displayed.
(3) Chose the pref^rifed format for displaying the results and dials: OK.

. .

2 - 69

..

_ a PC
a

air
*'

.,,

~ .. "
, .,.
.,. .
. " . ~. , ,

a
;"

~~ "-,.- ,,,,*.*"..<:I

"$1
*
a;
"*^;,,-~ -

_ __. J*-., ,.
,

*
a

$1

;.-

I ';$"~\- *,, ;;
0.16, ;ICt, ^;;t*,*:

.,,

.
..

..,

2 - 70

B. Accounting the area in each FCD category and saving the FCD Map
TTLis function of the system calculates FCD in every Grid or polygon area* When it takes
statistics in the polygon area, inputting the mask file. Tlie mask file is a bitmap file (256
colors) which is produced using existing imagery software, Tlie same palette vanie can
be used to distinguish one FCD category area from another. The results am be

presented in tables and saved as an Excel file. The image can be saved as a bitmap file.
(1) 011dc Map Statistics ill the FCD menu.

The Mapping Statistics dialog box is displayed.

B. , Presentation of the results

Results showing the number of hectares in each FCD category can be based on either
the whole Grid or polygon area.
Case-I: Designates the area by Grid,
(1) Click Grid tab.

(2) Chooses one milt of Grid at the pixel or the distance.

(3) hip\It one piece of bigiiess of Grid.
Case-2: Designates the area by mask file.
(1) Click Mask tab.
(2) GIId^ . . . in the MaskFile.
(3) input file name in the File name box.

B. 2 Designation of the partition which take statistics of FCD Map and, Designation of
the display color

The default is 11 pieces of partition in all. (0% and I. partition per 10% from I%)
Also, display color is suitably arranged.
.

(1) Tb designate where the division into FCD categories begins and where it ends, use the
Drag & Drop function of the left mouse button on the color bar and displayed at the
box under the Default button.

(2) When (1) above is completed, the Color dialog box is displayed.
(3) Chose the preferred colors for presenting the data and click OK.

2 - 71

FCD Menu
"^,,.,^.

Selects of the partition which totals FCD Map and,

the display colors

fop SCOCi. Sties

.~:

Selects of the partition value

\^

^"

100

.*-

is;
11
42.85 "

:;*:2:1

0.75

L. I

L. ?6

1.1- 20

3.2L

21- 3

6.67

31. I

L2.29
6 L4.20
I:. ?.:..' 8.70

;^* 6.12
.

41- s 14
51- 601

*' is.
, . ,, ,,,,, a. :?
h, . , ,... . ,>

7J- Gol

a, ,:*

{.^j;,

..$:g:.., ^\,$t

2.96 eL~ 90

,.\

,;;!

.$1e '*.-. ^^;, .

' '~ "t;,*.^!:

0.49 9L-JOD

:^

!^!,

. ,^;,.;^$^;,

-a,

""^;;;;

A.
\

Closes this dialog box

Accounting of
the whole analysis area

Saves the image and accounting result

Totals FCD Map every in putted accounting area

"

. ,*

$4*
*.

42.85 I 0- o
---.-^, --^

o. 75 I L. to
*2, . 1.76 I IL. 20.

3.2L zl- 301

6.67 I aL- 40,

TZ:55'' 4L. ~3^, ': a

;^^

"'!, *I '~

7 8.70 : 6L. 70.1 -'

.*'.;4,
,"

*'t
,

*.

q. ;;^^, j

a'~',..
g:1:3:; t.

..

~~~'!'-----!

.,..

*
.".

..

..

* -t-,*! 3<4* ,

0.49 I 91-100:'

0,49, 91 100,

.,>. q
^,

,.
*,

.*^ ',

I ', ', g, -* . *
-.

>: ^

..,.

. I^
*-' 14

-- *

_Gal ' ' ^

Ra

C^,

".

The select mask file dialog box is displayed

2 - 72

B. 3 Accounting of the FCD Map

It takes statistics of the area accounting.
(1) 011clc Calculation.

(2) TIT. e result are displayed in the TABLE OF MAPPING STATISTICS window.

B. 4 Saving of the results

When the TABLE OF MAPPING STATISTICS window is displayed the results can be
saved foil the first time. Tables are saved in Excel file format and the image is saved as a
bitmap file.
(1) Clicl^ Save.
(2) The system first saves the FCD Map Tm age.
It alitomatically saves under the ^Ie name to be preserved and clicks OK.
(3) After saving the image, the system then automatically saves the tables

..,

2 " 73

23.11 Aa, se 124.3 10 166.4a 132.4 121.87 16.48

,, '::_*'.. 59.94 130.78 ,L?0.2 IJ6.2 20L. 69 176. A4 ;55:ET~ IL6,2

LSS. 52 165.61 1534.6 ;23.49 1332.36 IL63.62 j294.84 '38.8
^-^- .-^---. -~. .-^---- .--..- -..-.. J - ^ - ^ - --J ~. -

$06.25
1861. o3 12.47.82 16 1720,9 1962.28 !27,749 f6. I
_L^;^_-^!.^-__^,. L_
~ ^-- ---------- ~---^-- -- . - - ---. . . - 4 -. . .. . . .- . ... . - .. _. , . . . _ ^ .

842.4 11637.82 11250.64 10 ILO3S. to 12383. BS 12313.36 12.43

:'---.. ' Jan6.9 IISL8. ?S I~T~^~"i I I ~ ,~
$5.71=;, L776.33 In1,639 12.43 10 4254,69 1780.03 14.86 O
1598. Ls 413.5a 10 ID 1603. ,s ?6.95 io o

us-,* I, -*, " to jus~" J! " '

^:*!*e^'*,*-*at^'^#%?$^:%^^;**&^?^aL^,;,^

. .

2.74

2.3 FCD-Mapper File Menu

2.3. , Close of the project to be processing
co Click Close Project in the File menu,
(2) Return to the first window that appeared at the start of the FCD exercise.

2 - 75

_D

P
e;q

:I

*;^
#,{

*"

**I

^1.1

^ti, r&~ , 615 ' . *, j

~.~.~

,.

"
Ie

' <4 A
, \' <4"'
' ~
,\ '*,.

,*

*,

g
'3

*;^;!* ,:,.%*?,.: ,'. - $35

13
^,*

'f;$*,*'

**;P, is*43^*$*:

a*9

-. ,

..

.-

2 - 76

2.3.2 Changing the expert file, the name of the person conducting the
analysis, and comments on the project
(1) Oncl^ Pro^CI Property in the File menu.
Tlie Project Properly dialog box is displayed.
(2) 'I^, pe the name of the e><pert file in the Expert File box.
(3) T^7pe of name the person conducfuig the analysis in the Author box
(4) input comments in the Comment box.
(5) 01/131c OK

2 - 77

Seiecs required or type author

Type the comment

A. Rtk, ita

The select expert file dialog box

is displayed
Reglstes project property
and closes this dialog box

...

2 - 78

2.3.3 Adding to or deleting from the project file, the expert file,
or the name of the person conducting the analysis
(1) Click Menu Property ill the File menu.
TTLe Menu Property dialog box is displayed.
(2) When ending, cfick Close.

A Deleting the name of a person in conducting the analysis

(1) OnCIE Authors tab.
(2) Choose the name to delete from the list.
(3) Click Delete. ..

B. Adding and Deleting of the project file

co Click Pro^CI Files tab.
(2)-I When adding:
a) 011dc Add Files. .,
The File dialog box is displayed.

by T^pe the project file name in the File box.

(2)-2 When deleting:

a) Choose the project file name to delete from the list.

b) GIIcl^ Delete. ..

C. Adding to and Deleting of the expert file

(T) Gild< BCPeit files tab.
(2)-I When adding:
a) Click Add Files. ..
T}Ie File dialog box is displayed.

b) T^pe the name of the new e, :liert file in the File box.
(2)-2 When deleting:

a) Choose the e>:pert file name to delete from the list.

by Cliclc Delete. .,

2 - 79

Lists authors

Closes this dialog box

Deletes of selected authors

SE

.a^
,

Usts project files

Closes this dialog box

Adds new project file
Removes of selected project file

.. .

\ .....

UsIs expert files

Closes this dialog box

Adds new expert file
Removes of selected expert file

2 - 80

2.3.4 False 'color display of the processing image

This function of the system allocathig various processing images for RGB such as FCD
and displays them hatse color.
(1) Click Load False linage tilthe File menu.
TTLe Load False linage dialog box is displayed.

(2) Choose the image to display (Tb a maximum of 7 images)

a) Click Add.
The File box dialog box is displayed,
b) Type the file name to the File box.
c) A file name is displayed in the list.
d) When cance^Ig a chosen image, choose the file to amcal and click Demove,
(3) Click OK
The False Select Window dialog box is displayed.
(4) Allocating the images for RGB, it displays in ^Ise color.

2.3.5 Pseudo color display of the processing image

This frillction of the system displays various processing images such as FCD in pselido
color.

(1) Grid< Load Pseudo linage in the File menu.

The File dialog box is displayed.
(2) 115, pe the file name to the File box.
The Pseudo Seleel Window dialog box is displayed*
(3) It designates a color and it displays in ^Ise color.

. .

2 - 81

C:\E'CDPATA\POD\POD. Asi. FBI

C:\FCDDATA\FCD\POD, AVI .FBI
CS\FCDDATA\POD\ECD, Bi. FBr
C :*C'CDDATA\FCD\Feb. CLD. EEL
CS\FCDDATA\ECD\FCD. Pep, FBI

Displays False Select window

and closes this dialog box
Demoves of selected image file
Adds display image file

.
-..

2 - 82

2.41mage display window operation

Click the right monse button on the image display window, and the sub menu of the
various processing is displayed

2 - 83

".

bag
".

BOL
*

i I

"

"

^ i
~~,.~~,,*~

a
*

,-

"

.*!

"

\.

*I

, 193*,>"

*^;-;^
False color display sub menu

( Right clicks )

$
_B"

t.

~~--- -

at*.,,

u ,,,. -<!,

!a
,.

* .*;,.*.; "

.^

.^

- --*-,. .

^,

Its\

,
;E

\,

Fir^at6', 66';,::;:'

!-,:,&:^r. ^t, !A
~

a~.,*

b$*

""*'"

I~' - ?*1.1:-;:.!:,,,_

I;,:I by"""

kg",^,%.%, a

I, .;i, I, *^

Pseudo color display sub menu

( Right clicks )

2 - 84

A False Setting
Changing the RGB band and threshold.
(1) GIIcl< False Setting in the display sth menu,
The False Select Window dialog box is displayed.
(2) Click band.

(3) Chi!c on the following buttoi^ as required,

Linear Stretch : linear stretch transform.

Histogram Equalization : Instogram Equalization transform.

User : Designates the minimum value and the maximum vanie of the inPITt by
Drag & Drop of the left mouse biitton. .
(4) Click OK
False color image is displayed.

B. Pseudo Setting
Changing the pseudo band and color

(1) Click the Pseudo Setting in the display sub menu,

The Pseudo Select Window dialog box is displayed.
(2) Click on the following buttons as required.
Unear Stretch : linear stretch transform

Histogram Equalization : IEli. stogramIE;q. nanzation transform.

linear 0-1 00 : The range to display is image value from O to 1.00.
(3) Click on the following buttons as required* '
Default : It does suitable coloring.

User : Designates the minimum vanie and the maximum vaine of the color partition
by Drag&Drop of the left mouse btxtton. .

a) Color partition is displayed at the box under the Default button.

b) When designating color partition, the Color Select dialog box is displayed.
c) Select new color.
d) Click OK.

A color bar is renewed.

. .

2 - 85

Selects start and end value

Histogram of selected band

PR

Displays color bar

Displays image and closes this dialog box

Strechs start and end value

Selects start and end pseudo color

96

Start value of pseudo color

End value of pseudo color

Displays image
and closes this dialog box

Seiecs of the partition which pseudo color and,

the display colors

2 - 86

C. Save Bitmap
Saving the image as a bitmap file.
(1) Click Save Bitrnap in the display sub menu.
The File dialog box is displayed.

(2) Type the name of bitmap file (, B^^IP) in the File box.
(3) Gild^ OK.

D. Printer Setting
Setting the property of the printer.

E. Print

Printing the image,

F.

Zoom

Zooming the image

(1) Click Zoom in the display sub menu

(2) Select one of the following sub menu as required.
Iru : magnification is 11n^
*I, . : magnification is 1.1.

G. Histogram
(1) Click Histogram in the display sub menu.
Tlie Instogi, am window dialog box is displayed,
.

111 case of fo. Ise color, it chooses Bliie, Green, Red.

2 - 87

H. Ruler

(1) Click Ruler in the display sub menu.

Displaying a ruler in the TIPper and left side of the window
The display of the ruler is a 1.00 pixel unit

I. Scale

(1) Click Scale in the display sub menu.

Displaying the scale bar which is shown in the lower right of the screen by 100 pixels.

. .

2 - 88

..

Appendix
A List of Abbreviation
Awl

Advanced Normalized Vegetation index

Asi

Advanced Shadow Index

AVI

Advanced Vegetation index

Bl

Bare soil Index

BIT,

Band Interleaved by Title

BlvlP

Bitmap

BSQ

Band seqtiential

FC

Forest Clusters

FCD

Forest Canopy Density

ITTO

internationalT^, opicalTimber Organization

JOECA

Japan Overseas Forestry ConsultantsAssociation.

NDVl

Normalized Differential Vegetation. index

PCA

Principal ComponentArialysis

SI

Shadow Index

SSl

Scaled Shadow index

Tl

Thermal index

VD

Vegetation Density

Vl

Vegetation Index

. .

2 - 89

. .