Benefits of a multiple-solution approach in land change models

Abstract

Land change (LC) models are dedicated to a better understanding of land use and land cover dynamics. A fundamental aspect of those models lies in the calibration of spatial parameters underlying such dynamics. Although there are many studies on the calibration of LC models, current efforts have a common goal of seeking to find a single global optimum solution, even though land change dynamics may be inherently heterogeneous throughout a given space. This article presents a calibration approach for finding multiple optimal solutions. A crowding niching genetic algorithm (CNGA) is incorporated into a cellular automata LC model. The model is applied to simulate urban expansion in Wallonia (Belgium) as a case study. Our findings demonstrate the ability of the model to locate multiple solutions simultaneously. In addition, the CNGA performs better than the standard genetic algorithm—besides, the CNGA helps to better understand the properties of land change dynamics within a given landscape.