Integrated Sentinel-2 LULC, SCS-CN, and HEC-HMS analysis of hydrological response changes in the Ake Kobe Watershed, Indonesia 2019–2024

This study quantifies the hydrological impacts of land use and land cover (LULC) change on the Ake Kobe Watershed (792.54 km 2 , Halmahera, North Maluku, Indonesia) between 2019 and 2024, coinciding with rapid industrial estate development in eastern Indonesia. Annual LULC maps were produced from Sentinel-2 imagery using a Random Forest classifier on Google Earth Engine, achieving an Overall Accuracy of 90.0% and a Kappa coefficient of 0.875, confirmed across 80/20, 70/30, and 60/40 training-testing splits. Reference samples were generated through systematic visual interpretation of Sentinel-2 composites cross-referenced with Google Earth high-resolution imagery and available regional land-use maps. Between 2019 and 2024, the watershed experienced substantial LULC transformation: forest cover declined by 4.85%, while industrial areas expanded by 638.8%, bare land by 127.9%, and built-up areas by 123.3%. GIS-based overlay of LULC and Hydrologic Soil Group maps produced composite Curve Numbers (CN) for nine sub-watersheds following USDA-NRCS TR-55 procedures; the area-weighted composite CN increased from 73.32 to 80.36 (+9.59%). SCS lag times were computed using the NRCS (2010) corrected formula Tlag = [L^0.8(S+1)^0.7]/[1900×Y^0.5] (L in feet, S in inches, Y in percent), yielding physically plausible lag values of 47.5–235.9 min. Design rainfall for a 25-year return period (166.92 mm, 24-h, SCS Type II) was determined via Gumbel frequency analysis, validated by Kolmogorov–Smirnov test. Under the 25-year design storm, scenario-based simulations—presented as uncalibrated comparative scenario estimates given the absence of an independent streamflow gauge—showed that 2024 LULC conditions produced a 49.13% increase in outlet peak discharge (from 3,176.49 to 4,737.15 m 3 /s), a 20.42% increase in direct runoff volume, and a 30-min acceleration in time-to-peak. These findings demonstrate that industrial expansion and deforestation have significantly increased flood hazard in the Ake Kobe Watershed. The integrated Sentinel-2, SCS-CN, and HEC-HMS framework provides a reproducible baseline approach for hydrological impact assessment in data-sparse, rapidly industrializing tropical watersheds, against which future studies employing machine learning-enhanced or uncertainty-aware modeling frameworks may be benchmarked.