Monitoring Land Use and Land Cover Changes in the Lower Mara River Basin, using Google Earth Engine

Changes in land use and land cover (LULC) have significant impacts to the hydrological cycle, altering the precipitation patterns and water quality hence posing a threat to natural watersheds over time. This study assessed the temporal and spatial land use and land cover changes in the Lower Mara River Basin in Tanzania by using Google Earth Engine (GEE), focusing on years 2003, 2013, and 2023. It monitored classification accuracy, analyzed changes in LULC classes, tracked their persistence and trajectories, and examined the spatial distribution for specific periods (2003-2013 and 2013-2023). Similarly, it employed supervised image classification using the random forest algorithm on GEE using high-resolution imagery from Landsat 7 for year 2003 and Landsat 9 for years 2013 and 2023, respectively. Four land cover classes were identified, which are vegetation, water, settlement, and other LULC (especially mining, agriculture, bare land and range land). In 2003 the total study area was 439771.59ha whereby the LULC showed 21.8% vegetation, 19.4% water bodies, 7.3% settlement and 51.3% other LULC. In 2013 the total study area was 502832.71ha whereby 2013, the LULC proportions shifted to 28.96%, 7.66%, 28.48% and 34.8% respectively. In 2023 (with the same study area as that of 2013), the proportions further transformed to 23.1%, 3.69%, 42.32% and 30.87%, respectively. Classification achieved a high overall accuracy of more than 80% and Coefficient of Kappa index above 0.7 for each year indicating the acceptability of results. Main finding exhibits a decrease in areas covered with vegetation, waterbodies and Other LULC while settlement increased. The rate of this replacement differs temporally and spatially across Lower MRB. These results imply that there have been significant changes in LULC for the past 20 years. Changes have been attributed to the lack of good plans and the changes in livelihood systems in the area