Wetlands are crucial ecosystems providing vital services, but are vulnerable to degradation. Hulu Sungai Utara (HSU) Regency, South Kalimantan, holds extensive tropical peatland wetlands covering over 64,821 hectares, significant yet facing considerable shrinkage due to reduced water flow and land conversion. This study analyzed spatiotemporal wetland cover changes in HSU from 2000 to 2024 using Landsat 7 ETM+ and Landsat 8 OLI data on Google Earth Engine (GEE). We employed NDWI, NDVI, and MNDWI indices and Machine Learning algorithms Random Forest (RF) and Support Vector Machine (SVM) for classification into water, vegetation, built-up, and barren land. RF demonstrated superior accuracy (97.33% total accuracy, 96% average kappa), effectively mapping transitions. Results show a long-term wetland decline (71.4 km² loss) driven by development, climate change, and human activity. Critically, a significant regeneration trend emerged in the past 15 years, with gains surpassing losses in this recent period. This highlights the value of GEE-based geospatial technology for data-driven tropical wetland conservation and indicates positive impacts from recent initiatives.

Keywords: Wetland, Water Index, Machine Learning, Google Earth Engine, Land Cover.