Abstarct: Abstract This research has addressed the modeling of the Moroccan locust (Dociostaurus maroccanus) habitat using biophysical indicators extracted from satellite images and the multilxayer perceptron (MLP) neural network model. In this study, temperature indices (TCI, LST), vegetation cover (OSAVI), and bare soil (BSI) were selected as input variables and combined with locust presence data. The performance of the model was evaluated using two criteria: the coefficient of determination (R²=0.9851) and the Nash-Sutcliffe coefficient (NSE). The NSE values for the test and training data were 0.98 and 0.99, respectively, indicating the high accuracy of the model in predicting the locust distribution. The findings emphasize that temperature and vegetation variables play a key role in the distribution of this species, and the neural network has been able to identify susceptible habitats with high accuracy. This study demonstrates that artificial intelligence models baxsed on satellite data are highly effective for monitoring pest habitats. These models are particularly useful for managing pests in vast and inaccessible areas.