This document evaluates an expert-based erosion model called LandSoil in its ability to map erosion in an agricultural catchment area. The model was tested on a study area from 2014-2016 and results were compared to field measurements of runoff, erosion, and deposition. The model was able to account for 92% of erosivity and showed close correspondence with measured runoff volumes. Comparisons of modeled and measured erosion and deposition using graduated rulers also showed satisfying results, though with high standard deviations. The model analysis suggested that while field boundaries had little impact on erosion amounts, they did influence sediment transport. Grass and wood strips were found to effectively deposit sediment through reduced runoff speeds.

1. Is an expert-based model able to map
erosion in agricultural catchments?
Vincent Cantreul1,Nathalie Pineux2, Gilles Swerts3, Charles Bielders4, Aurore Degré5
1Université de Liège, 2Gembloux Agro-Bio Tech, Service Public de Wallonie, 3Université de Liège,
Gembloux Agro-Bio Tech, 4Université catholique de Louvain, 5Université de Liège, Gembloux
Agro-Bio Tech
MATERIALS AND
METHODS
INTRODUCTION
CONCLUSIONS
Rainfall amount
[mm]
Selected rainfall
[mm]
Runoff
[mm]
Runoff
coefficient [%]
2014 266.6 195.2 6.3 3.2
2015 76.8 64.6 0.9 1.4
2016 283 260.8 14.8 5.7
Fig.5 Comparison between measured and modeled runoff for
point 1 (2014, 2015 and 2016)
Fig.6 Comparison between graduated rulers measurements
and erosion modeling
Tab.1 Rainfall and runoff amount for 2014, 2015 and 2016
• The rainfall selection permitted to include
92 % of the erosivity in the model (tab.1).
• Comparison between measured and modeled
runoff volumes showed very close
correspondence (fig.5). Three events (2 in
2014 and 1 in 2016) get out of the adjustment
for point1. These rainfalls had a high quotient
between rainfall amount and effective
duration.
• Graduated rulers permitted to validate
modeling at some locations in the watershed
(fig.6). The comparison for the 15
measurement zones was satisfying even if
standard deviations of measurements and
modeling remained high.
• Field limits did not seem to have big impact on
amount of eroded and deposited particles. But
it played a role in the transport acting as a
channel for small erosion rates lower than 0.5
T/event or around. It showed that land use had
more influence than landscape design in the
erosion management.
• Grass strips deposited sediment because of
infiltration in the strip and consequently
runoff speed reduction even if other processes
could interact. The process was amplified by
low slopes in addition to the grass strip.
• The wood strip deposited much more soil than
grass strips (>5 cm for some zones) and cut the
erosion channel by decreasing connectivity.
Fig.3 Position of rulers with graduated scale
Fig.4 Runoff (flume with debitmeter) and erosion
(ruler) measurements
Fig.1 Location of the study area
Fig.2 DEM and measurements points
LandSoil is a spatially distributed model based
on the STREAM erosion (Cerdan et al., 2002a,
2002b; Souchère et al., 2003; Souchere et al.,
1998) and the WaTEM/SEDEM tillage erosion
model (Govers et al., 1994). STREAM models
runoff and erosion at plot/ small catchment
scale spatially and at rainfall event timely. It is
adapted as an ArcGis template which provides
ready-to-use layouts to make interface easy to
understand. The objective with LandSoil is to
analyze topographic evolution in agricultural
landscape resulting from soil redistribution in
the catchment (Ciampalini et al., 2012). After
each rainfall or ploughing event, a new digital
elevation model is calculated taking into
account all eroded or deposited soil
transported in the runoff water flow.
For comparison purpose, rulers with graduated
scales are placed at some locations in the
catchment to measure erosion and deposition
(fig.3). 15 zones with several rulers for each are
identified in the watershed, respectively 6, 3
and 6 for 2014, 2015 and 2016 (fig.4).
Two automatic samplers are located at the two
discharge measurement points in the
watershed.
MAIN RESULTS
Sediment production from agricultural field and
transport after a rainfall event have significant
consequences on environment and population .
It underlines the need to better understand
erosion processes including production,
transport and sedimentation in order to find
solutions.
Among all different types of models, expert-
based models try to represent the watershed
taking into account only major erosion processes
in a specific situation in order to prevent from
over-parameterization .
The objective of the study is to test LandSoil
about the capacity to represent runoff, erosion
and deposition patterns in an agricultural
watershed (fig.1) with a particular attention on
spatial distributions around major connectivity
points. For that purpose, the whole watershed
behavior (fig.2) between 2014 and 2016 is
modeled to then focus on some linear features
(grass strips, wood strip and field limits). These
results are confronted to field measurements
(erosion and deposition height measured by
graduated rulers, eroded particles mass
measured by automatic samplers). This aims to
assess the reliability of the model in the Belgian
loamy context and particularly to evaluate the
capacity of the model to quantify the effect of
linear features as observed in the watershed.