A Google Earth Engine and Machine Learning Model for Near-Real Time Spatiotemporal Change Detection: Enhancing Vegetation Cover Assessment in Nech Sar National Park, Ethiopia

Nech Sar National Park (NSNP), a vital biodiversity sanctuary in Ethiopia, is experiencing increasing pressure from human activities. This research employs Google Earth Engine (GEE) and machine learning (ML) techniques to examine vegetation patterns and land cover transformations from 2015 to 2024. The study utilized high-resolution Planet NICFI (4.77 m) imagery and evaluated four ML algorithms for land cover classification: Random Forest (RF), Support Vector Machine (SVM), Classification and Regression Tree (CART), and Gradient Tree Boost (GTB). RF emerged as the most accurate model, with an overall accuracy of 97.81% and a Kappa coefficient of 0.9715, surpassing SVM (79.56%), CART (96.35%), and GTB (94.16%). The findings revealed substantial ecological changes: dense vegetation decreased from 34.2% (2015) to 22.9% (2024), while savanna coverage expanded from 27.9% to 61.1%, primarily due to agricultural expansion and deforestation. Bushland saw a dramatic reduction (19.4% to 0.8%), and grassland fluctuated, reaching a peak of 21.5% in 2019. Bare land remained relatively constant (2–8%). Hotspot analysis identified the eastern part of NSNP as susceptible to land-use alterations. The study introduced an innovative integrated model, combining GEE's cloud computing capabilities with RF, enabling real-time monitoring of land cover dynamics. This framework allows for scalable, high-resolution analysis, providing policymakers with valuable insights for developing adaptive conservation strategies. The research highlights the transformative potential of ML and cloud-based platforms in addressing ecological degradation, offering a replicable approach for protected areas worldwide facing similar challenges.