CLOUD-BASED SPATIOTEMPORAL DATA MINING USING GOOGLE EARTH ENGINE AND MACHINE LEARNING APPROACH FOR VEGETATION COVER CHANGE ASSESSMENT IN NECH SAR NATIONAL PARK, ETHIOPIA

Abstract This study examines the application of cloud-based satellite imagery mining for the continuous detection of geospatial data dynamics and its implications for endemic wildlife in Nech Sar National Park, Ethiopia. Google Earth Engine addresses the limitations of traditional desktop computing methods that restrict effective monitoring of land cover changes critical for conservation efforts. The research aims to identify and quantify land cover transformations over the past two decades, improve classification accuracy through machine learning algorithms, and develop a comprehensive model utilizing satellite imagery and Google Earth Engine (GEE) for near-real-time monitoring of the park. Employing a mixed-methods approach, the study integrates quantitative analyses using advanced machine learning techniques—including Random Forest, Support Vector Machine, Classification and Regression Tree, and Gradient Tree Boosting—with qualitative assessments to evaluate model effectiveness. Results highlight the diverse land use/cover changes within the park, particularly in the eastern region, where hotspot maps indicate substantial encroachment of built-up areas and agricultural land. Over the two-decade period analyzed, the extent of evergreen broadleaf vegetation has increased due to planting in built-up areas, while grassland coverage has diminished. Notably, the Random Forest algorithm demonstrated superior performance, achieving an overall classification accuracy of 97.81% and a Kappa coefficient of 0.9715, underscoring its effectiveness in land cover classification. This study establishes a well-known model for enhancing land cover monitoring in Nech Sar National Park, facilitating sustainable management practices, and informing conservation strategies.