Integrating google earth engine and random forest for land use and land cover change detection and analysis in the upper Tekeze Basin

This study analyzed Land Use and Land Cover (LULC) change in the Upper Tekeze Basin using Google Earth Engine (GEE) and the Random Forest (RF) classifier. Remote sensing data from Landsat 5 (TM) for 1986, Landsat 7 (ETM+) for 1998 and 2010, and Landsat 8 (OLI) for 2022 were used to classify LULC classes. To triangulate the satellite imagery results, field data from 178 randomly selected households and 15 local elders were used. The study results revealed that from 1986 to 2022, water bodies, forests, cultivated land, settlements, and bare land increased by 100%, 37.84%, 13.93%, and 11.73%, respectively. In contrast, grassland, and shrub and bushland areas decreased by 82.08% and 18.84%, respectively. Moreover, the results show that about 62.92% of the landscape experienced at least one LULC transition, with 18.27% net change and 43.65% swap change. The findings show that most land covers have changed to cultivated land and settlements. The accuracy analysis of the Landsat images showed a satisfactory level, as the changes in LULC align with local perceptions. Local farmers ranked the factors influencing LULC change as follows: agricultural expansion, free grazing, overgrazing, firewood collection, topography, and poor land management, from most to least important. Similarly, local farmers ranked poverty, population growth, rainfall variability, lack of awareness about proper natural resource use, and government intervention, from most to least important, as the root causes of LULC change. Therefore, controlling the causes of LULC change and promoting sustainable resource use are crucial to prevent further LULC change and the loss of scarce natural resources, which are essential for maintaining landscape sustainability.