Seasonal and temporal dynamics of optically active water quality parameters in Gefersa Reservoir, Ethiopia.

Reservoirs in subtropical highlands face intensifying pressures from pollution, unsustainable practices, and climate change; however, multi-decadal water quality dynamics under distinct seasonal regimes remain poorly understood. This study developed a seasonally optimized, cross-sensor harmonized remote sensing and ensemble machine learning framework integrating in situ historical measurements, meteorological records, and GEE platform to investigate the 25-year (2001-2025) dynamics of five optically active water quality parameters; chlorophyll-a (Chl-a), TSS, TURB, CDOM, and SDD, in the sediment-dominated Gefersa Reservoir, a critical drinking-water source for Addis Ababa. Findings revealed evolving seasonal patterns: CDOM, TSS, and TURB remained highest during the wet season; however, wet-dry differences diminished because of a 147-173% increase in dry season Chl-a and CDOM. Chl-a also shifted from dry season to wet season dominance after 2015, accompanied by significant decoupling from seasonal climate cycles. TSI analysis indicated persistent eutrophic to hypereutrophic conditions, with dry season TSI increasing by 7.2 units and exceeding wet season values after 2011. Ensemble models achieved strong predictive performance (R2 up to 0.878). Overall, seasonally adaptive monitoring is essential for targeted erosion control, water treatment, and sustainable reservoir management in data-sparse subtropical systems.