Predicting land use and land cover change is vital for managing forest dynamics, particularly in protected areas under human pressure. This study focuses on Talassemtane National Park in Morocco, utilizing advanced geospatial modeling techniques specifically, the QGIS MOLUSCE plugin and artificial neural networks to assess and forecast forest cover changes. Using historical Land use land cover maps (1995,2000,2005) and spatial variables such as slope, distance to buildings, roads, and water, the model was trained and validated, achieving an overall accuracy of 86% and Kappa coefficient of 71 %. Predictive analysis indicates forest cover will decrease from 35,522 hectares in 2024 to 31,254 hectares by 2040, a 12% loss averaging 251 hectares annually. By 2040, forests are projected to cover only 48,4% of the park’s area, down from 55% in 2024. The main drivers of forest loss include fires, illegal logging, agricultural expansion overgrazing, and infrastructure development, all exacerbated by climate change. These findings highlight the urgent need for targeted conservation strategies and sustainable land management practices. The methodology presented offers a scalable approach for other protected areas, providing valuable insight for policymakers and resource managers to mitigate forest loss and promote ecological resilience.