Answers are grounded in this paper and cite their sources.
Study region The Jiaoyi Basin (3345.12 km²) in Shandong Province, China, contains five small and medium-sized reservoirs, three of which lack continuous storage observations. The study covers 2018–2022. Study focus We developed an automated waterbody-extraction method combining iterative Otsu thresholding with Random Forest object recognition and a Sentinel-1/Sentinel-2 storage-fusion method. In total, 1016 satellite scenes (690 Sentinel-1 and 326 Sentinel-2) were used across the five reservoirs. The methods were developed and evaluated at the two data-rich reservoirs using 392 scenes (278 Sentinel-1 and 114 Sentinel-2) and then applied to the three data-scarce reservoirs. Reconstructed daily storage was analyzed using lagged Spearman correlation, the Reservoir Influence Index (RI), and reservoir-aware Grid-Xinanjiang (GXAJ) simulations. New hydrological insights for the region Waterbody extraction achieved a coefficient of determination (R²) up to 0.9103 and Kappa up to 0.9848, while storage fusion reduced normalized root mean square error (NRMSE) by 0.0609 and 0.0429 in the validation period at Tangcun and Xujiaya, respectively. The five-reservoir RI was 0.09, and peak storage-change–streamflow correlations were 0.29–0.53, indicating detectable but modest daily regulation. Incorporating reservoir dynamics increased Nash–Sutcliffe efficiency (NSE) by 0.01 and reduced peak-flow and absolute runoff-volume errors by 1.6 and 1.1 %age points, respectively, mainly improving peak and recession simulations.
No takes yet. Share an insight, caveat, or question.
Wu et al. (2026) studied this question.
Synapse has enriched 10 related papers. Consider them for comparative context: