Escolar Documentos
Profissional Documentos
Cultura Documentos
Catena
journal homepage: www.elsevier.com/locate/catena
A R T I C L E I N F O A B S T R A C T
Keywords: Many surveys with differently sized rainfall simulators have been conducted in the past to study runoff and
Rainfall simulator erosion processes on the plot scale. Concerning the understanding of runoff processes on the slope scale,
Surface runoff comparative studies have shown that sprinkling devices with large plot sizes (≥40 m2) produce more
Runoff formation representative results than small-plot rainfall simulators (4 m2 and less). Nevertheless, the latter ones are
Land use
frequently used due to their portability and the reduced effort and water demand compared to large-plot
Grazing
simulations. In this study the results of torrential rain (100 mm h− 1) experiments using a small (1 m2) dripper
Torrential rain
device and a large (40–80 m2) spray device are compared regarding to identify site characteristics which allow
estimating the representativeness of surface runoff coefficients measured with a small-plot rainfall simulator. We
conducted 39 small-plot (1 m2) and 14 large-plot (40–80 m2) rainfall simulations at eight selected experimental
sites in the Eastern Alps with differing land cover, land use, intensity of grazing and antecedent soil moisture
content. At sites with intensive grazing, due to topsoil compaction and the shortened flow path, the small-plot
device delivered significantly higher runoff coefficients than the large-plot device. At mainly mowed grassland
sites with at most a short grazing period in autumn both sprinkling devices showed similar results or even lower
runoff coefficients compared with the results of the large-plot simulator. The measured surface runoff
coefficients strongly depended on antecedent soil moisture content and grazing intensity.
⁎
Corresponding author.
E-mail addresses: gertraud.meissl@uibk.ac.at (G. Meißl), klaus.klebinder@bfw.gv.at (K. Klebinder), bernhard.kohl@bfw.gv.at (B. Kohl), gerhard.markart@bfw.gv.at (G. Markart).
http://dx.doi.org/10.1016/j.catena.2017.04.009
Received 22 August 2016; Received in revised form 10 March 2017; Accepted 7 April 2017
Available online 26 April 2017
0341-8162/ © 2017 Elsevier B.V. All rights reserved.
C. Mayerhofer et al. Catena 156 (2017) 184–196
Fig. 1. Comparison of the results of rainfall simulations with a small (4 m2) and a large device (75 m2) at Löhnersbach catchment (modified after Markart and Kohl, 1995). The small
device was a simulator with swayable nozzles, carried out by Peringer (unpublished), the large simulator was equal to the large device in this study.
(Guggenberger, 1980; Markart and Kohl, 1995). In Sharpley and saturated. On the contrary, the results of 1 m2 sprinkling experiments
Kleinman (2003), the higher surface runoff values on a small plot may also be regarded to represent the lower limit of possible runoff
(2 m2) compared to a larger plot (32.6 m2) were attributed to the fact coefficients as water may rapidly percolate in the not irrigated areas
that a greater area of the smaller plots, namely > 75% of area, became outside the sprinkled plot (Dobmann, 2009). Thus, in order to reduce
185
C. Mayerhofer et al. Catena 156 (2017) 184–196
side effects, plot sizes > 40 m2 (Auerswald et al., 1992b) or > 10–50 detailed description see Markart and Kohl, 1995; Markart et al.,
m2 (Schindler Wildhaber et al., 2012) are often recommended. 2011). As the plastic tarp, guiding the water into the drain, is
In general, it can be assumed that the larger the plot is, the more positioned about 5–10 cm below soil surface, the collected discharge
representative the results are with respect to the understanding of also includes near surface runoff. Soil moisture content is measured
runoff processes at the slope scale. However, the realization of large- at two profiles in about the middle of the experimental plot in four
plot rainfall experiments is associated with a considerable effort. depths (5 cm, 15 cm, 25 cm and 40 cm) with TDR sensors.
Demand of time and staff as well as financial and technical issues b) The centrepiece of the small-plot simulator is a commercial micro-
(e.g. availability of water) are increasing with the size of the plot. This drip irrigation system constructed with drip heads and connecting
is why small-plot rainfall simulators – which are suitable to investigate pipes by Gardena attached to a 1 × 1 m aluminium frame. The drip
initial runoff generation and splash-, interrill- and rill erosion (e.g. system has been preferred as the irrigation of an exact plot size of
Iserloh et al., 2013a, 2013b) – have also been used very frequently to one m2 is very difficult with spraying nozzles. The device is
study runoff and erosion processes on the slope scale (e.g. Dobmann, positioned on telescopic sticks which are adjustable in height in
2009). Thus, it is necessary to compare differently sized rainfall order to adapt the simulator to the experimental hillslope site. A
simulations according to their sensitivity to local characteristics, second aluminium frame with a fine-meshed grid attached below
influencing factors and hydrological parameters (Dobmann, 2009). the drip system distributes the drops in an irregular pattern to the
This could result into a better understanding of the interaction between surface. Surface runoff is collected in a similar manner like at the
the environment and their influence on surface runoff generation at the large-plot simulator (Fig. 2). Soil moisture content is measured at
plot-scale and hence allow a more reliable assessment and application the upper edge of the drainage trench with two FD sensors.
of the results found with small-plot sprinkling experiments. Therefore, a
systematic approach to determine the correlation of hydrological According to the large set of previously realized rainfall simulations
parameters and surface runoff at different experimental scales during (Markart et al., 2011), a constant rain intensity of 100 mm h− 1 and an
rainfall simulations is preferable, but still missing. The presented survey experiment duration of 1 h was chosen for each experiment. Fuel-
refers to this subject and aims at answering the following research operated water pumps and different types of hoses were used for the
questions: water-supply of both rainfall simulators. Pump capacity for each
sprinkler was chosen due to the required amount of water. Valves were
• How do the results of small-plot rainfall simulations agree with used to generate a consistent flow rate and thus a constant intensity of
those of large-plot simulations? What is the variation between the the applied artificial rainfall. A variable area flowmeter by Krohne
results of the two simulator types? (Model: DK32; measuring range: 40–150 l/h; measuring accuracy:
• Does the variation of the results follow any specific pattern 4.0%; directive VDI/VDE 3513-2) was used to generate an accurate
depending on local characteristics? Can a statement be found to and constant flow rate during the small-scale rainfall simulations. At
explicate the variation? the device for the large plot, precipitation input is also controlled by a
• How can effects of antecedent soil moisture content and mechanical pressure gauge, an automatically registering flow meter with a data
impact (grazing) on surface runoff be described with different sized logger.
simulators? Both rainfall simulators differ regarding to the drop generation:
while rainfall is sprayed at the large plot (maximum drop size 5 mm,
In order to answer these questions, we performed rainfall simulation approx. 5% of atomized spray), the small simulator generates drops
with two different rainfall simulators (small plot device: 1 m2, large with a size of 3–5 mm. Due to constructional reasons of the small-plot
plot device: 40–80 m2) at eight different sites in Alpine catchments device, it is not possible to irrigate outside the investigated plot in order
(grassland and forests). The comparison of the results is supported by to reduce side effects. For the large-plot device, a lateral co-irrigation of
laboratory analysis of soil physical properties of the sites. the neighbouring areas was not conducted within the presented study
due to limitations in water availability.
2. Material and methods Regarding the uniformity of rainfall distribution over the whole
plot, the size and the construction of the small-plot device supports
2.1. Rainfall simulators homogeneous conditions. Newesely et al. (2015) tested a smaller, but
similarly constructed (20 m2) device like the large-plot device pre-
Two rainfall simulators were used: sented here with a satisfying uniformity apart from the upper and lower
boundary of the plot where the rainfall intensities were too low. This
a) The transportable spray irrigation installation for large plots (width problem does not occur at the large-plot device used for the presented
5 m; variable length up to 20 m, typically irrigated plot size study due to the usage of nozzles with different apertures. A systematic
40–80 m2) uses quadrant (0.18 m3 s- 1), semi-circle (0.36 m3 s- 1) distribution of rain gutters within the plot showed good results.
and full-circle commercial nozzles (0.72 m3 s- 1). The application of
different nozzles allows a more exact coverage of the test plot. Thus, 2.2. Study area
in the edges and along the upper boundary line of the irrigation unit
quadrant circles are used. Half circle nozzles are used for the The rainfall simulations were conducted at eight experimental sites
segments in the line of the greatest slope. If water supply is not located in Tyrol and Vorarlberg in Western Austria (Fig. 3 and Tables 1,
restricted, the outer sides in the line of the greatest slope are 2). At each site one plot was irrigated with the large-plot simulator
irrigated too to equalize losses due to wind and side outflow. In this once, partly also repeatedly in order to simulate the influence of
case semi-circle and full-circle nozzles are used, raining half part increased soil moisture content. Plot size had to be chosen according
into the investigated plot and half part outside. The nozzles are to the site's topography and water availability between 40 m2 and
PIPELIFE products (BW-STDF-series). As this special type is not 80 m2. At each site several plots were sprinkled with the small-plot
produced any longer, a larger quantity of theses nozzles was bought device (1 m2), also partly repeatedly. At some sites (Brix, Lutz, Steg),
in time to secure the operation of the spray irrigation installation for the large- and the small-plot simulations were done simultaneously side
the future. by side. In order to increase the number of experiments, small-plot
The surface runoff resulting from the applied artificial rainfall is rainfall simulations were also carried out at sites where the large-plot
collected in a drain which is placed in a trench at the lower end of simulation had been accomplished in the past within the last three
the irrigated area and led to calibrated measuring tanks (Fig. 2) (for years (Rech, Rech F, Tal 1, Tal 2 and Watt). At those sites, the small-plot
186
C. Mayerhofer et al. Catena 156 (2017) 184–196
Fig. 2. Rainfall simulators used in this study: large-plot rainfall simulator (left); small-plot rainfall simulator (right; here at Brixenbachalm Brix).
simulations were carried out within exactly the same plots like the large plots and six small plots at Brix were irrigated in May prior to
large-plot simulations. At Rech and Tal 1, brilliant blue was added to the grazing season and in September after the cattle had left the
the sprinkling water in order to visualize the infiltration and percola- pastures. All experimental plots were marked allowing a repetition
tion paths in the soil. On the day after the rainfall simulation several of the rain simulations at exactly the same spots.
vertical cross sections of the irrigated soil were photographed (Schader,
2013). The results of the experiments were compared by means of three
A systematic experimental set-up was selected to reveal factors characteristic numbers: the total surface runoff coefficient Ψtot, the
influencing the surface runoff generation and to discuss the compar- surface runoff coefficient at runoff constancy Ψconst and the time to
ability of the large-plot and the small-plot sprinkling device: runoff. The latter is the time span between the start of the rainfall
simulation and the first occurrence of surface runoff in the drain at the
• Rainfall simulations were conducted at sites characterised by lower end of the experimental plot (Fig. 4). The time to runoff depends
different types of land cover and land use (Table 1). on the amount of interception and water storage in soil depressions
• To assess the influence of antecedent soil moisture content on (Kohl and Markart, 2002; Markart et al., 2011). After the start of
surface runoff, experiments at Rech, Lutz and Steg were repeated surface runoff, during most of the rainfall simulations a constant level
after a short break. The break between the first and second of surface runoff is reached, which marks a kind of equilibrium state.
experiment was about 1 h for the large-plot and 20 min for the Thus, the dimensionless surface runoff coefficient at runoff constancy
small plot (as the flow length is shorter at the small plot, a shorter Ψconst is calculated by dividing current surface runoff by the actual
break was regarded to be justifiable and allowed the execution of applied precipitation during the time in which equilibrium state is
more experiments). At Brix the repetition experiments at the small reached (Fig. 4). The dimensionless total surface runoff coefficient Ψtot
plot were conducted on the next day within 24 h due to logistical is the sum of surface runoff divided by the applied rain sum at the end
reasons. of the experiment. According to the definitions, Ψtot has to be smaller
• Most experiments were performed at grassland sites differing in than Ψconst.
vegetation cover and in grazing intensity. To demonstrate the After the rainfall simulations loose soil material and undisturbed
influence of intensive cattle grazing during the summer, one of the soil cores (200 cm3 cylinders) were taken from the irrigated sites for the
Fig. 3. Location of the experimental sites in Western Austria (Eastern Alps): 1 Ruggbachtal, 2 Brixenbachtal, 3 Wattental.
187
C. Mayerhofer et al. Catena 156 (2017) 184–196
Lower Austroalpine
moraine material
Quartz phyllite
Talus material
organic matter, the latter measured after loss on ignition at 430C. The
saturated vertical conductivity was measured at the undisturbed soil
Geology
3. Results
Cambisol - based on calcareous unconsolidated
runoff
Cambisol; loamy sand – sandy silt
The surface runoff coefficients Ψtot and Ψconst achieved in the field
Cambisol; silty – loamy sand
Cambisol; sand – silty sand
experiments are displayed in Fig. 5 for all first rainfall simulations (for
unconsolidated sediment
the results of the repeated experiments see chapter 3.2, Figs. 7 and 8).
At the experimental site Brix, all small-plot rainfall simulations showed
a higher surface runoff than the large-plot experiment prior to the
grazing season (Brix bg), the Ψconst-values from five out of six small-plot
sediment
experiments exceeded the results from the large-plot after the grazing
Soil
season (Brix ag). Also at the pasture sites Rech and Tal 1 and the forest
site Rech F, the small-plot simulations delivered significantly higher
runoff coefficients compared with the large-plot experiments. At the
pasture site Tal 2, however, both sprinkling devices produced almost
Mowing; short grazing
period in autumn
period in autumn
Intensive grazing
Intensive grazing
Intensive grazing
Intensive grazing
meadow Lutz ranged between 0.01 and 0.53, with eight out of nine
small-plot rainfall simulations below the result from the large-plot
simulator (0.45). At Steg (hay meadow), two small-plot experiments
Land use
Forestry
ψconst 0.52 at Lutz, of ψtot 0.35 at Watt) and may over- (18 out of 32
Grassland
Grassland
stricta L.)
1160 m
1490 m
1450 m
1240 m
m a.s.l.
880 m
Tyrol, Austria
Tyrol, Austria
Tyrol, Austria
Tyrol, Austria
Vorarlberg,
Vorarlberg,
Austria
Austria
with the large device. However small-plot experiments at Steg and Rech
resulted in similar measurements.
Abbrev.
Rech F
Tal 2
Rech
Watt
Lutz
Steg
Brix
Brixenbachalm
Talkaseralm 2
Brixenbachtal
Rechentalalm
Ruggbachtal
Lutzenreute
Wattenberg
Wattental
188
C. Mayerhofer et al. Catena 156 (2017) 184–196
Table 2
Rainfall simulations.
Number of irrigated plots Among them: repeatedly irrigated plots Plot size (m2) Number of irrigated plots Among them: repeatedly irrigated plots
1.0
saturated runoff
precipitation coefficient Ψconst
0.8
runoff coefficient Ψ
measured runoff
coefficient Ψ per
0.6 time step
time to runoff
begin of experiment
end of experiment
0.4
0.2
0.0
15:30 16:00 16:30 17:00 17:30
time
start/end precipitation
measured runoff coefficient smoothed runoff coefficient
Fig. 4. Hydrograph from a rainfall simulation (modified after (Kohl and Markart, 2002)). Fig. 6. Time to runoff for all first rainfall simulations with the large-plot device (square)
and the small-plot device (triangle), repetition experiments with increased antecedent soil
moisture content are not shown. No surface runoff for the large-plot experiment at Rech F.
(Abbreviations see Tables 1 and 2).
Fig. 7. Surface runoff coefficients Ψconst in relation to antecedent soil moisture content
from all double sprinkling experiments. Grey-first experiment, black-repeated experi-
ment, ӿ Brix bg plot 1,+ Brix bg plot 3, x Brix bg plot 4 (explanation of abbreviations see
Table 1).
189
C. Mayerhofer et al. Catena 156 (2017) 184–196
Fig. 8. Results from the experimental plots at Brixenbachalm before and after the grazing season: (left) Ψtot and Ψconst, time to runoff (right, value for Brix_2 missing).
general, both simulators react to changes in antecedent soil moisture in texture classes silty loam, loam and sandy loam. The coarse material
a comparable manner. was enclosed by the soil matrix. Fine and medium pores had a major
At Brix, the results of the experiments showed the effect of intensive proportion of the total pore volume.
grazing by cattle on the runoff reaction (Fig. 8). With exception of two Knowledge on the physical characteristics of the site Wattenberg is
small-plot simulations (Brix_1 and Brix_2) all experiments delivered a unfortunately poor due to the loss of the soil samples of the two upper
significant increase of surface runoff after the grazing season. Apart layers. The existing data showed a low bulk density and a medium
from Brix_6, the time to runoff was reduced in autumn. Both, small-plot hydraulic conductivity (about 50 cm d− 1).
as well as large-plot experiments showed the impact of grazing on the
surface runoff, the large-plot simulations indicated a higher increase
4. Discussion
after the grazing season than the small-plot simulations.
In general, rainfall simulations are approximation techniques to
determine runoff generation at experimental sites. The representative-
3.3. Soil physics
ness of the found information is increasing with the plot-size
(Auerswald et al., 1992a, 1992b; Schindler Wildhaber et al., 2012;
The soil at the Brix site was characterised by a very high content of
Wainwright et al., 2000). Here it is assumed that the results of the large-
skeleton (> 50%, Fig. 9), predominantly gravel from talus material.
plot sprinkling experiments produce a realistic estimation of the runoff
The grain distribution showed a high percentage of sand (> 60% of fine
reaction of the investigated sites. Thus, the surface runoff values
soil material), the soil texture was classified as sandy loam. Due to the
measured at the large-plot experiments are regarded to be the reference
soil material and the content of coarsest pores (> 20%), the saturated
values. In the following, deviations of the results of the small-plot
hydraulic conductivity increased towards the deeper layers. The top
device therefrom are discussed.
layer had a limited infiltration capacity due to the reduced coarse pore
volume, which is a result of the intensive trampling.
The sites Rech and Tal 1 showed also a high content of skeleton 4.1. Runoff behaviour at the experimental sites
(> 30%). While the fine soil fraction at Tal 1, situated in the geological
Schattberg Formation (paleozoic sandstones), was characterised by the 4.1.1. Brixenbachalm before grazing season (Brix bg) and after grazing
high content of sand (60%–70% of fines, texture sandy loam), the Rech season (Brix ag)
site (porphyroid) showed a content of silt between 50% and 60% At Brix bg (May 2012) the surface runoff coefficients (Ψtot and
(texture silty loam). The rapidly drainable coarsest pores increased Ψconst) of all small-plot experiments exceeded the result of the large-
towards deeper situated soil layers, which is a result of the decreasing plot experiments. The small-plot experimental site with the lowest
mechanical compaction due to trampling. The values of the saturated vegetation coverage during the whole survey (approximately 65%)
hydraulic conductivity were low (1.2–21.2 cm d− 1, Table 3) at Tal 1. showed by far the highest Ψconst (0.78). This corresponds with a
In contrast, the forested site Rech F showed a high infiltration previous study (Dadkah and Gifford, 1980) in which an increased
capacity (mean 93 cm d− 1). This can be explained by the loose packing runoff potential was noticed for areas covered by < 70% of vegetation.
(bulk density < 0.5 g cm− 3) and the very high percentage of very The low vegetation coverage could be an explanation for the extremely
coarse pores. Deeper layers had a higher content of skeleton while the high surface runoff coefficient Ψconst compared to the remaining small-
content of silt increases (45%–55%, texture loam/silty loam). plot rainfall simulations. Regarding Ψtot, this plot showed also the
The Tal 2 site had an unfavourable distribution of the pore volume. highest value within the measured range at Brix bg (0.6), however the
While fine and medium pores were sufficient, the content of drainable same value was also produced at another plot with a higher vegetation
pores was poor. Thus, values of the saturated hydraulic conductivity coverage (75%). This might lead to the conclusion that the Ψconst might
were low (about 28 cm d− 1 till a depth of 20 cm). In a depth of reveal a stronger dependence on vegetation coverage than Ψtot.
20 cm–30 cm, the hydraulic conductivity decreased to < 10 cm d− 1. In However, as no other sites with < 70% vegetation coverage were
this depth, percolation was hampered which explains the dammed-up included in this study, no reliable conclusion can be drawn.
water during the snowmelt season. The small-plot experimental sites with the lowest and second lowest
Due to the geological/geomorphological situation of the experi- runoff coefficients at Brix bg also offered a specific site characteristic.
mental site Ruggbachtal, which was characterised by small-scaled Massive cattle trampling in previous years led to the forming of a small
sequences of glacial sediments and Fresh Water Molasse, the soil at depression of 1 m length and 25 cm height at the small-plot with the
the sites Lutz and Steg showed a high amount of silt (45%, 54% of fine lowest surface runoff coefficients (Ψtot = 0.37, Ψconst = 0.42). Thus,
grained soil material) and clay (9%, 6%). Thus, the soils belonged to the retaining of water in the reservoir and a delayed runoff could be
190
C. Mayerhofer et al. Catena 156 (2017) 184–196
Brix Rech
5 5
15 15
Depth [cm]
Depth [cm] 25 25
40 40
60 60
0% 10% 20% 30% 40% 50% 60% 70% 80% 90% 100% 0% 10% 20% 30% 40% 50% 60% 70% 80% 90% 100%
Volume Volume
RechF Tal1
5 5
15 15
Depth [cm]
Depth [cm]
25 25
40 40
60 60
0% 10% 20% 30% 40% 50% 60% 70% 80% 90% 100% 0% 10% 20% 30% 40% 50% 60% 70% 80% 90% 100%
Volume Volume
Tal2 Lutz
5 5
15 15
Depth [cm]
Depth [cm]
25 25
40 40
60 60
0% 10% 20% 30% 40% 50% 60% 70% 80% 90% 100% 0% 10% 20% 30% 40% 50% 60% 70% 80% 90% 100%
Volume Volume
Steg Watt
5 2,5
15
Depth [cm]
Depth [cm]
15
25
25
40
60 35
0% 10% 20% 30% 40% 50% 60% 70% 80% 90% 100% 0% 10% 20% 30% 40% 50% 60% 70% 80% 90% 100%
Volume Volume
coarse frac. sand silt clay organic matter Very coarse pores coarse p. medium p. fine p.
Fig. 9. Grain and pore size distribution of the experimental sites (due to loss of soil core material, the data for Watt are incomplete).
observed. At the other site a high content of skeleton in the soil, loosely 4.1.2. Rechentalalm (Rech), Talkaseralm 1 (Tal 1) and Stegen (Steg)
embedded in a sandy soil matrix, was observed (see Table 3). This The experimental sites at Rech, Tal 1 and Steg had a more or less
favours infiltration and decrease of surface runoff since soil skeleton homogenous structure than the Brix site. Constant vegetation coverage
boundaries can serve as preferential flow paths (Scherrer, 2006; of approx. 95% and uniform topsoil characteristics led to a small
Spreafico et al., 2003). variance of the surface runoff coefficients for the small-plot sprinkling
After the grazing season (September 2012), the small-plot runoff experiments. However, due to logistical reasons only two small-plot
coefficients still exceeded the large-plot results. Again, the experimental experiments each were conducted at the plots, thus it cannot be ruled
site massively trampled by cattle in previous years showed the lowest out that a larger number of simulations would not have led to a broader
surface runoff coefficients of the small-plot experiments due to retain- range of results. At the sites Rech and Tal 1 (Fig. 10), the surface runoff
ing of water in a reservoir formed by cattle trampling. After intense coefficients (Ψtot and Ψconst) measured with the small-plot devices
grazing, the difference of the small-plot and large-plot Ψtot and Ψconst overestimated the values gained with the large-plot device. At these two
was smaller than prior to the grazing season. Obviously intensive sites, the small-plot simulations with brilliant blue showed matrix
grazing leads to a homogenization of the whole slope, thus the infiltration in the compacted top soil and macropore flow in the deeper
characteristics of the small plots don't differ so strongly leading to layers with decreasing interaction with the matrix (Fig. 11).
more homogeneous surface runoff coefficients. At the site Steg (Fig. 10), both the small-plot and the large-plot
rainfall simulations showed intensive runoff contributions from the
surface and additionally a pipe flow caused by preferential flow in
191
C. Mayerhofer et al. Catena 156 (2017) 184–196
Table 3 with the small device occurred without delay and reached a high level
Results of the soil-physical analysis of core samples at the experimental sites. quickly. Increased surface runoff at the small device is attributed to the
shortened flow path for surface water compared to large-plot systems,
Plot Depth Bulk Density Saturated Hydraulic Conductivity [cm/d]
[cm] [g/cm3] as there is less time and area for surface water to infiltrate into the soil
Num. of Values Mean (Markart and Kohl, 1995). In contrast, a higher infiltration rate at the
samples large device is to be expected due to a higher overland flow depth
(Wallach et al., 1997). The length of the large sprinkling device results
Brix 0–10 0.92 3 13.7; 23.9;47.2 28.24
10–20 1.20 3 13.7; 54.4; 60.85 in a longer time of concentration. Consequently, the level of more or
114.5 less constant runoff occurs later at large device experiments. However,
20–30 1.66 1 335.7 335.7 both sprinkling devices respond to the differing soil characteristics of
30–50 – – – – the sites in a similar manner. The higher runoff at Tal 1 can be
Rech 0–10 0.95 3 34.5; 38.9; 41.30
explained with a significant higher bulk density and a reduced
50.6
10–20 1.05 2 80.0; 168.4 124.20 hydraulic conductivity in a depth of 10–30 cm.
20–30 1.09 2 103.2; 175.1 139.15
30–50 0.89 – – – 4.1.3. Talkaseralm site 2 (Tal 2)
Rech F 0–10 0.49 3 6.8; 75.5; 92.78
At Tal 2, very high and almost identical surface runoff coefficients
196.0
10–20 0.67 2 19.3; 36.9 28.11 for the large-plot and the small-plot simulation were found. Since the
20–30 0.94 3 8.2; 11.2; 22.1 13.85 site – also quite intensively grazed - was covered with dense mat grass
30–50 – – – – vegetation (Nardus stricta L.), organic matter can act like a thatched
Tal 1 0–10 0.84 3 4.4; 9.2; 21.2 11.59 roof and cause high surface runoff (Markart and Kohl, 1995; Markart
10–20 1.29 3 1.2;6.2; 16.8 8.04
20–30 1.18 3 2.9; 5.2; 14.5 7.53
et al., 2011). The low infiltration capacity at this site appears also in
30–50 – – – – spring after snowmelt, when the organic layer is nearly saturated. The
Tal 2 0–10 0.69 3 8.9; 18.6; 55.3 27.58 good correspondence between the results of the large-plot and small-
10–20 1.06 3 4.0; 27.4; 51.9 27.75 plot devices at Tal 2 support the assumption expressed above, that at
20–30 0.99 3 3.3; 10.0; 11.3 8.2
sites with very low infiltration potential the influence of the plot size of
30–50 – – – –
Lutz 0–10 1.04 3 15.4; 20.0; 35.11 the sprinkling device diminishes.
70.0
10–20 1.31 3 6.1; 9.7; 76.3 30.71 4.1.4. Rechentalalm Forest (Rech F)
20–30 1.33 3 5.7; 13.7; 94.5 37.97
At the spruce forest site of Rech F, no surface runoff occurred at the
30–50 1.34 5 1.7–121.2 45.19
Steg 0–10 1.03 3 4.8; 6.2; 17.7 9.55
large-plot mainly due to the loosely layered forest soil, intensively
10–20 1.34 3 5.3; 6.3; 8.1 6.57 rooted and rich in macro-pores. At the small-plot, surface runoff started
20–30 1.46 3 2.2; 26.8; 57.1 28.70 quickly due to hydrophobic effects of the litter layer at the beginning.
30–50 1.40 3 17.0; 26.0; 37.43 About 10 min after starting the rainfall simulation, the temporary water
69.3
repellency ceased and surface runoff decreased. Such hydrophobic
Watt 0–5 0.64 1 14.4 14.4
10–20 1.00 1 50.4 50.4 effects are known from other studies (Burch et al., 1989; Butzen
20–30 1.20 1 52.1 52.1 et al., 2015; Markart and Kohl, 1995; Kohl, 2011; Scherrer, 1996).
30–40 – – – The detected low runoff coefficients (Ψtot = 0.11, Ψconst = 0.12) in
contrast to Ψtot = Ψconst = 0.00 of the large-plot simulator can be
explained with these initial hydrophobic effects of the surface com-
mouse holes during the experiment. In comparison with the large-plot,
bined with the short surface flow path of the small-plot. Surface runoff,
small-plot experiments led to relatively similar or a slightly under-
which also occurred at the large-plot due to hydrophobic effects,
estimated results regarding Ψconst. On the contrary the Ψtot-values
shallow rocks and uncovered roots, could infiltrate in the soil at
gained with the small-plot device overestimated the value of the large-
numerous small sinks, along rotted wood fragments and at zones
plot simulator.
covered by dense ground vegetation (grass, herbs, moss). The high
However, the chronological development of the runoff differs
amount of large soil pores (> 50 μm) (Fig. 9) allow an instantaneous
between both devices at all sites. While the large device produced a
percolation of the rain to the deeper soil layers, which are loosely
delayed beginning and gradual increase of runoff, runoff generated
bedded and capacious at this site.
0
00:00 00:00
Time [hh:mm] Time [hh:mm]
Fig. 10. Experiments Rech, Tal 1 and Steg with the large-plot (black) and the small-plot device (dark and light grey) showing an overestimation of runoff (both Ψtot and Ψconst) and a
shorter time to runoff with the small-plot device.
192
C. Mayerhofer et al. Catena 156 (2017) 184–196
Fig. 11. Result of a small-plot rainfall simulation with a dye tracer (brilliant blue) at Rech. Photograph (left), colored pixels (middle), histogram of colored pixels (right), medium dotted
line 50%, right dotted line 100%. While matrix infiltration takes place in the compacted top-soil layer, macropore infiltraton and exfiltration into the matrix dominate in the lower layers
(bar for depth in 10 cm steps, modified after Schader, 2013.) (For interpretation of the references to color in this figure legend, the reader is referred to the web version of this article.)
193
C. Mayerhofer et al. Catena 156 (2017) 184–196
c) We identified one site (Tal 2), at which the surface runoff and after the grazing season (September 2012) at Brix showed a strong
coefficients delivered by the small-plot device was nearly equal to influence of grazing to the generation of surface runoff (Fig. 8). The
the values measured by the large-plot device. High surface runoff antecedent soil moisture contents were similar at both times despite of
coefficients result at large plots, if a high proportion of the area differences in the three days sum of antecedent precipitation (48.7 mm
shows surface runoff producing characteristics (e.g. impermeable in May; 0.9 mm in September). The strong surface runoff increasing
root felt, compacted top soil, saturated top soil). Thus, the site is effect of the grazing seasons, measured at the large plot and four of the
quite uniform and there a balanced ratio between influencing six small plots, confirms other studies (Kohl and Markart, 2002;
factors (side losses, shortened flow path) may have led to very Leitinger et al., 2010; Pietola et al., 2005). After grazing, less water is
similar surface runoff coefficients at the small and large plots. able to infiltrate into the soil due to the reduction of the vegetation
However, due to logistical reasons, only one small-plot experiment cover, the compacted soil surface owing to mechanical impact by
took place. It cannot be ruled out that a greater amount of rainfall grazing livestock and hydrophobic effects initiated by accumulated
simulations would have led to different results. Also at Brix ag, Lutz dead organic matter (Markart et al., 2011; Markart et al., 2004;
and Steg singular small-plot experiments produced surface runoff Sharrow, 2007). During the non-grazing season a recovering process
coefficients very close those of the large-plot simulations. Further of the soil (decreasing of bulk density, increasing of macroporosity and
experiments are necessary to verify these findings. hydraulic conductivity) occurs as a consequence of a number of
processes including shrinkage and swelling, frost slacking, earthworms
Summarizing it can be stated that at our investigated Alpine sites we burrowing and root penetration (Cournane et al., 2011; Unger, 1991).
found both relationships between the surface runoff coefficients Thus an increased infiltration potential exists in spring before grazing
produced by the small-plot and large-plot rainfall simulations, with season begins (Leitinger et al., 2010). This explains the relatively low
the small-plot values underrunning the large-plot values as expected by surface runoff coefficient (Ψtot = 0.14, Ψconst = 0.22) at the large plot
Dobmann (2009) as well as exceeding them as reported by several measured in May 2012 at Brix. Such effects of restructuring of the
authors (Guggenberger, 1980; Markart and Kohl, 1995; Sharpley and topsoil as well as their limitations are known from different studies (e.g.
Kleinman, 2003). As the conduction of large-plot rainfall simulations Czell, 1972; da Silva et al., 2003; Sharrow, 2007).
evolve a greater effort, it would be helpful if the results of a small-plot In order to investigate the effect of antecedent soil moisture content,
rainfalls simulator [1 m2] could be related to the more representative repetition experiments were conducted at Brix, Rech, Lutz and Steg.
results of the large-plot device by a correction factor or even a power Apart from one exception (Rech 1) all sprinkling experiments at the
law as formulated for comparative plot size studies under natural rain small plot resulted in a higher surface runoff at the repetition experi-
(Asadzadeh et al., 2012). Although some characteristics were pointed ment (Fig. 7). The increase of the surface runoff values ranged between
out which may help to identify plots at which the small-plot device will 0.07 and 0.18 at the small plots and between 0.06 and 0.17 at the large
produce too low or too high surface runoff coefficients, no evidence on plots and showed a similar trend for both sprinkling devices and both
the magnitude of the deviation can be given. surface runoff coefficients.
It may be hypothesized that increased antecedent soil moisture
leads to smaller differences between the runoff coefficients of the large- 5. Conclusions
plot and small-plot simulations because of the increasing connectivity
of the flow paths within the plots. This is true for the repetition This study targeted at providing a possibility to a better under-
experiments at Lutz, Steg and Rech (Fig. 7). However, as at all three standing of the information found with small-plot sprinkling experi-
sites, only one repetition experiment at the small plot is available due to ments (1 m2) compared to – in the context with runoff estimation in
logistical reasons, the significance of these findings cannot be evalu- hydrological catchments – more representative but labour intensive
ated. At the site Brix, where three small-plot repetition experiments large-plot simulations (40–80 m2). Similar conditions provided, the
were carried out, the above stated hypothesis was not confirmed. This following considerations may allow a better evaluation of the results of
might be because of the low antecedent soil moisture content at the small-plot rainfall simulations in relation to large-plot experiments.
large-plot, even in the repetition experiment. Further rainfall simula- If the runoff reaction shall be analysed by a small-plot rainfall
tions are necessary to get reliable results. simulation, it is highly recommended to conduct several experiments at
According to the time to runoff, with the exception of Tal 2 and each one site. Due to the small patterned site characteristics the resulting
two plots at Lutz and Watt, all small-plot experiments delivered a runoff coefficients often vary within a broad range.
shorter time span until surface runoff was measured in the drain than Runoff measurement executed by different sprinkling devices
the large-plot simulations. These findings are contrary to the results of (sprinkled area, side proportions, type of sprinkler or nozzles) will
Auerswald et al. (1992a) and Newesely et al. (2015) where runoff mostly lead to deviating results of surface runoff parameters (Markart
started later with decreasing plot size. In this study the explanation for and Kohl, 1995; Iserloh et al., 2013a, 2013b) as confirmed in this study.
the shorter initial time could be given in the spatial pattern of the However, based on the present results it seems possible to assign
artificial rain. While raindrops at the small simulator fall close to the following general characteristics:
drain, the rain intensity at the lowest part of the irrigated area is
reduced at the large simulator (Newesely et al., 2015) in order to avoid • At sites intensively grazed by cattle, the small-plot device is
direct rain into the drain. Thus the flow length and the time of first expected to deliver surface runoff coefficients higher than the
runoff are increased. large-plot rainfall simulator, due to topsoil compaction and the
Due to the flow distance of surface runoff at the large sprinkling shortened flow path. If the site additionally is characterised by low
device, the runoff generation process takes more time at the large vegetation coverage (≤ 70%) the overestimation can be signifi-
simulator. Thus it takes more time till a constant runoff level is reached. cantly higher.
Beside the area of the sprinkling device, the proportion between the • At mainly mowed grassland sites, surface runoff coefficients pro-
length and the area of the simulator must be respected comparing the duced with the small-plot device are similar or even lower than the
runoff volume. results of the large-plot rainfall simulator. Caution is required at
sites indicating high macropore content. There, the small-plot
4.3. Influence of grazing and antecedent soil moisture content on surface simulator shows a tendency to significantly underestimate the
runoff realistic runoff coefficient.
• Experiments with the small-plot simulator may result in a higher
Repeated rainfall simulations before the grazing season (May 2012) surface runoff coefficient in forested sites with low ground cover due
194
C. Mayerhofer et al. Catena 156 (2017) 184–196
to the shortened flow path and the initially hydrophobic surface. Ghadiri, H., Payne, D., 1988. The formation and characteristics of splash following
195
C. Mayerhofer et al. Catena 156 (2017) 184–196
initiation. Hydrol. Earth Syst. Sci. 10, 309–320. flow and phosphorus transport. J. Environ. Qual. 32, 2172–2179.
Rodrigo Comino, J., Iserloh, T., Morvan, X., Malam Issa, O., Naisse, C., Keesstra, S.D., Sharrow, S.H., 2007. Soil compaction by grazing livestock in silvopastures as evidenced
Cerdà, A., Prosdocimi, M., Arnáez, J., Lasanta, T., Ramos, M.C., Marqués, M.J., Ruiz by changes in soil physical properties. Agrofor. Syst. 71, 215–223.
Colmenero, M., Bienes, R., Ruiz Sinoga, J.D., Seeger, M., Ries, J.B., 2016. Soil erosion da Silva, A.P., Imhoff, S., Corsi, M., 2003. Evaluation of soil compaction in an irrigated
processes in European vineyards: a qualitative comparison of rainfall simulation short-duration grazing system. Soil Tillage Res. 70, 83–90.
measurements in Germany, Spain and France. Hydrology 3 (1), 6. http://dx.doi.org/ Spreafico, M., Weingartner, R., Barben, M., Ryser, A., 2003. Hochwasserabschätzung in
10.3390/hydrology3010006. schweizerischen Einzugsgebieten, Berichte des Bundesamt für Wasser und Geologie.
Schader, R., 2013. Untersuchungen zu den Versickerungseigenschaften von Almflächen pp. 4.
mittels Färberversuchen. Master thesis at University of Innsbruck, Austria. Unger, P.W., 1991. Overwinter changes in physical properties of no-tillage soil. Soil Sci.
Scherrer, S., 1996. Abflussbildung bei Starkniederschlägen - Identifikation von Soc. Am. J. 55, 778–782.
Abflussprozessen mittels künstlicher Niederschläge. Dissertation at ETH Zürich, Wainwright, J., Parsons, A.J., Abrahams, A.D., 2000. Plot-scale studies of vegetation,
Switzerland. overland flow and erosion interactions: case studies from Arizona and New Mexico.
Scherrer, S., 2006. Bestimmungsschlüssel zur Identifikation von hochwasserrelevanten Hydrol. Process. 14, 2921–2943.
Flächen. Landesamt für Umwelt, Wasserwirtschaft und Gewerbeaufsicht Rheinland- Wallach, R., Grigorin, G., Rivlin, J., 1997. The errors in surface runoff prediction by
Pfalz (LUWG). Oppenheim. (http://scherrer-hydrol.ch/pdf/scherrer/luwg_bericht_ neglecting the relationship between infiltration rate and overland flow depth. J.
18-2006_ansicht.pdf (5.8.2016)). Hydrol. 200, 243–259.
Schindler Wildhaber, Y., Bänninger, D., Burri, K., Alewell, C., 2012. Evaluation and Weiler, M., Naef, F., 2003. An experimental tracer study of the role of macropores in
application of a portable rainfall simulator on subalpine grassland. Catena 91, 56–62. infiltration in grassland soils. Hydrol. Process. 17, 477–493.
Schmocker-Fackel, P., Naef, F., Scherrer, S., 2007. Identifying runoff processes on the plot Wilcox, B.P., Wood, M.K., 1988. Hydrologic impacts of sheep grazing on steep slopes in
and catchment scale. Hydrol. Earth Syst. Sci. 11, 891–906. semiarid rangelands. J. Range Manag. 303–306.
Sharpley, A., Kleinman, P., 2003. Effect of rainfall simulator and plot scale on overland
196