| Abstract: | The response of a landslide near Barcelonnette (southeast France) to climatic factors was simulated with three slope stability models: a fully empirical gross precipitation threshold, a semi‐empirical threshold model for net precipitation, and a fully conceptual slope stability model. The three models performed with similar levels in reproducing the present‐day temporal pattern of landslide reactivation, using dendrogeomorphological information as test data. The semi‐empirical and conceptual models were found to be overparameterized, because more than one parameter setting matching the test data was identified. In the case of the conceptual model, this resulted in strongly divergent scenarios of future landslide activity, using downscaled climate scenarios as inputs to the model. The uncertainty of the landslide scenarios obtained with the semi‐empirical model was much lower. In addition, the simulation of strongly different scenarios by the fully empirical threshold was attributed to its incomplete representation of the site‐specific landslide reactivation mechanism. It is concluded that the semi‐empirical model constitutes the best compromise between conceptual representation and model robustness. Copyright © 2000 John Wiley & Sons, Ltd. |
