Variability and patterns of surface water levels during dry season under environmental change: A case study at the Gan River-Poyang Lake-Yangtze River confluence area

Study Region Gan River-Poyang Lake-Yangtze River confluence area (GRPLYR), China. Study Focus Changing environmental and river–lake conditions complicate the interpretation of dry-season water-level records. This study assesses the velocity of water level descending (VWD), a window-based indicator of water-level recession, and tests whether recession information is more consistent than absolute water levels. Daily water levels from six stations during 2010–2021 were used to construct VWD series at 3-, 5-, 7-, 10-, 15-, and 30-day scales. Mann–Kendall trend and mutation tests were applied to water-level and VWD series. Four candidate theoretical probability distribution functions were fitted to VWD samples. Parameters were estimated using particle swarm optimization (PSO), maximum likelihood estimation, and L-moments. The optimal distribution-estimation framework was selected using goodness-of-fit metrics and information criteria. New Hydrological Insights for the Region: Absolute water-level series showed contrasting trends: Nanchang declined significantly, Xingzi, Hukou, and Pingfeng increased significantly, and Ganjiang Wucheng and Xiushui Wucheng showed no significant trend. In contrast, VWD series showed no significant trend or clear mutation point across the tested scales. PSO performed best, and the PSO-fitted Pearson type Ⅲ distribution was selected as the most suitable distribution. VWD provides a more consistent probabilistic description of dry-season water-level recession and supports low-water-level risk assessment and urban water-supply management.