Résumé
This work relies on DESERTICAS software, which is designed to model desertification dynamics. Spatiotemporal land changes are represented by cellular automata with continuous states. Their transition function incorporates various factors from the MEDALUS model, fundamental desertification properties, and additional factors such as land use practices, exploitability, and ownership considerations.
The main contribution of this article is the introduction of a control parameter in the DESERTICAS model to study an input-output problem related to the desertification phenomenon and the application of control theory in cellular automata models. The idea is to act directly on a predominant factor by setting its average intensity. This makes it possible to act indirectly on all the other factors in the model and slow down or halt land degradation processes. In this way, we found that the predominant factor in the study of desertification is management, and we determined its intensity using a genetic algorithm (GA) approach. The aim of combining control problem and GAs is to integrate land protection actions into the desertification simulation, i.e. the DESERTICAS software, and turn it into a decision-support tool.