Case Study Belgium

In this project we tested and compared two landscape evolution models with simple model structure: LAPSUS and Watem LT.

Several models have been developed that aim to reconstruct landscape evolution on a millennial time scale. These models are often not spatially evaluated against observed data. Furthermore, results obtained with these models are rarely compared with each other. In this study the performance of two landscape evolution models, WaTEM LT and LAPSUS, has been tested in the Belgian loess belt. WaTEM LT has been developed in Leuven and is special in its capability to simulate backwards in time. Both models aim to reconstruct landscape evolution based on a simple model structure, although LAPSUS is a little more complex. Both models are topography-based and spatially explicit, but use different detachment limited process descriptions. Additionally, the performance of a transport limited version of WaTEM LT is defined to test whether this kind of approach produces better results in the study area.
fig3.jpg

LAPSUS Application

Model calibrations and validations are performed within the Nodebais catchment (103 ha) and the Hamme-Mille catchment (64 ha), where intensive soil augerings have provided detailed data to evaluate long-term spatial patterns of erosion and deposition along two transects. The 2D model calibration and validation are performed within Nodebais, while for the 3D model calibration Nodebais was used and for the 3D model validation Hamme-Mille. Simulations are performed forward in time on a two-dimensional transect for four resolutions, and backward in time for the whole catchments three-dimensionally at 20 m resolution.

Results

In 2D, all models perform very well on all resolutions, resulting in Model Efficiency Factors (MEF) of 0.92 to 0.99 for calibration and 0.77 to 0.96 for validation. Especially LAPSUS reproduces good patterns of erosion and deposition. For both models parameter values on different resolutions are similar, which is an indication of small scale dependency of the models.

WaTEM LT performs equally well in 3D, resulting in a class-based MEF of 0.88 and 0.68 for calibration and validation, respectively. LAPSUS performs well for the calibration (MEF of 0.87), but was unable to execute a successful validation caused by technical anomalies related to backward modelling. The transport limited approach produced unrealistic results after application in 3D.

fig4.jpg
The results of this research demonstrate that the more complex process description of LAPSUS is an advantage in modelling forwards in time, but prevents it from modelling backwards. Consequently, simpler process descriptions, such as in WaTEM LT, are advisable when modelling overall soil distribution patterns backwards in time. For the Belgian loess belt, it can be concluded that a detachment limited approach is more appropriate, but this was only obvious after 3D application.

Download

Publications

  • Peeters, I., Temme, A., Buis, E., G. Govers and Veldkamp, A., in preparation. Comparison of two landscape evolution models in the Belgium Loess belt.
Further reading about backwards modeling:
  • Peeters, I., Rommens, T., Verstraeten, G., Govers, G., Van Rompaey, A., Poesen, J. van Van Oost, K. 2006. Reconstructing ancient topography through erosion modelling. Geomorphology, 78(3-4): 250-264.

ยป more LAPSUS Publications.

Contact

Ir. Arnaud Temme

Ir. Eke Buis

Ir. Iris Peeters KU Leuven

Prof. Dr. Ir. A. Veldkamp

Prof. Dr. G. Govers  KU Leuven