Evaluation of spatial and temporal changes over satellite images with machine learning using cloud-based google earth engine (Zonguldak sample)

The detection and monitoring of land cover changes is crucial for understanding the dynamics of ecosystems and managing natural resources effectively. Recent advancements in machine learning techniques, particularly through platforms like Google Earth Engine (GEE), have significantly improved the accuracy and efficiency of land cover classification. This study utilizes Sentinel-2 imagery and the Random Forest algorithm within the GEE platform to map land cover for the years 2015 and 2024. The results show a decrease in forest area from 49,628.51 ha in 2015 to 48,363.62 ha in 2024, primarily driven by the expansion of agricultural and urban areas. Agricultural land increased from 7,223.45 ha in 2015 to 7,869.61 ha in 2024, while urban areas expanded from 8,516.49 ha to 9,116.39 ha during the same period. The highest land cover transition was observed from forest to agricultural land, with a change of 2,582.14 ha. This study highlights the dynamic nature of land cover classification in recent years, thanks to the integration of machine learning and satellite-based data. The findings underscore the importance of continuous land cover monitoring for sustainable land management and environmental protection.