Comparing landscape evolution models with quantitative field data at the millennial time scale in the Belgian loess belt

Abstract

ABSTRACTThis study compares three landscape evolution models and their ability to correctly simulate measured 2500 year landscape evolution in two small catchments in the Belgian loess belt. WATEM LT and LAPSUS both model tillage and water erosion and deposition and have detachment‐limited descriptions for water erosion and deposition. Equations in LAPSUS are more mechanistic than those in WATEM LT. WATEM LTT resembles WATEM LT, but is a transport‐limited model. All three models are DEM‐based. Calibration and validation simulations were performed forward in time on (1D) transects for four spatial resolutions, and backward in time for (2D) catchments at 20 m resolution. For transects, model outputs were compared with discretized observations of transect shape. For catchments, outputs were compared with point observations of palaeo‐altitude, averaged over landscape element classes. For transects, the three models performed well, resulting in model efficiency factors of 0.92 to 0.99 for calibration and 0.62 to 0.96 for validation. However, for catchments, simulations showed that the transport‐limited WATEM LTT model could not realistically simulate long‐term landscape evolution. Performance of WATEM LT and LAPSUS at catchment scale was similar to that on transects, although LAPSUS has problems with backward calculation. Tests demonstrate that a transport‐limited approach cannot be used to model long‐term landscape evolution in the Belgian loess belt, which is in agreement with theoretical and empirical understanding of soil erosion processes in this environment. The difference in performance between transport‐limited and detachment‐limited models is clear only when the models are evaluated in a 2D catchment. The lack of such distinction when models were applied in a 1D transect highlights the importance of evaluating landscape evolution models in a 2D setting so that effects of flow convergence/divergence can be accounted for. Copyright © 2011 John Wiley & Sons, Ltd.