Validation of erosion estimates
The increasing availability of regional scale data layers on climate,
topography and land use has recently led to the application of quantitative
soil erosion model applications at European scale
It is important to assess the accuracy of such soil erosion estimates
from models that could be used in the context of soil protection.
This is, however, rather problematic at European scale, because
it is very difficult, if not impossible to acquire direct soil erosion
measures for large areas.
Therefore indirect soil erosion data derived from sedimentation
volumes in reservoirs in Belgium, the Czech Republic, Italy and
Spain were used to assess the accuracy of the PESERA soil erosion
estimates.
Firstly, the area contributing to each reservoir was delineated
using automatic watershed delineation algorithms. Next, WaTEM/SEDEM
was applied in order to assess for each reservoir a Sediment Delivery
Ratio (SDR). Erosion estimates were than multiplied with the SDR-values.
Finally observed and predicted sediment yield values were compared.
 |
 |
 |
 |
Central Belgium
(Verstraeten and Poesen, K.U.Leuven) |
Czech Republic
(Dostal et al., CVUT Prague) |
Spain
(Avendano Salas et al., Devente, Poesen and Verstraeten)
|
Italy
(Bazzoffi et al., ISSDS, Firenze) |
The results show that the model doesn’t produce accurate
soil erosion estimates based on 1km x 1 km data for all European
environments. The model is relatively successful in predicting the
relative pattern of soil loss from agricultural areas in central
Belgium and the Czech Republic. None of the model results however
accurately predict the
measured soil erosion patterns in Italy and Spain. The main reasons
for this are:
- uncertainty involved in the indirect validation method (i.e.
sedimentation data in reservoirs)
- low resolution of the input data used for model applications
at European scale
- simplified internal model structures that do not take into
account all sediment producing and transporting processes.
However, past research has shown that some models perform much
better for areas when high quality input data are used (Van Rompaey
et al., 2003). For example, runoff is better predicted from a 250m
digital elevation model (DEM) than from a 1km DEM. In many cases,
these high resolution input data exist in national mapping, meteorological
and soil survey institutions but at present are not readily available
for the whole of Europe.
An important future role for the JRC in future could be to harmonise,
aggregate and facilitate access to these data at European level.
It is clear than some attempt must be made to estimate soil erosion
losses in Europe for the forthcoming EU Soil Protection Strategy
and this could be done best by using a standard model applied to
standard data sets. Therefore, the Services of the Commission should
put much more effort into encouraging national and regional administrations
to provide the necessary input data for soil erosion models at much
higher resolutions than are currently available.
|
