Abstract
<title>Abstract</title> <p>Lakes İznik and Uluabat, ~ 50 km apart in northwestern Turkey, share the same regional precipitation regime and evaporative demand, yet followed divergent hydrological drought trajectories between 1993 and 2024. Using this natural experiment, we present a transferable, satellite-only attribution framework for separating climatic from anthropogenic drivers of lake-level drought, requiring no in-situ discharge data. Meteorological drought was quantified using the Standardized Precipitation Index (SPI) and Standardized Precipitation-Evapotranspiration Index (SPEI), computed from satellite precipitation and evapotranspiration products; hydrological drought was tracked via a Standardized Lake Level Index (SLLI) built from bias-corrected altimetric and gauge records. Propagation from meteorological to hydrological drought was assessed via lagged correlation, event matching, regression-based attribution, and Granger causality, while land-cover change was mapped using a 30-m global dataset (1985–2022). Precipitation showed no significant trend at either lake, whereas potential evapotranspiration rose significantly at both sites, widening the climatic water deficit and doubling SPEI-based drought frequency after 2000, implicating evaporative demand rather than rainfall decline as the primary post-2000 driver. Basin morphology governed propagation timescale: the shallow, flow-through Uluabat responded within ~ 2 months of meteorological forcing, while the deep, closed-basin İznik integrated forcing over 9–10 months, with causality confirmed at longer lags. Despite opposing land-cover trajectories, agricultural contraction and afforestation at İznik versus stable cropland and river abstraction at Uluabat, both lakes showed statistically similar non-climatic hydrological decline. These findings offer a scalable diagnostic tool for water managers in data-scarce lake basins across Central Asia, East Africa, and the Tibetan Plateau.</p>