Probability analysis of wet and dry encounters in the Yangtze and Yellow River basins under changing environmental conditions

Abstract

Assessing the probability of dry-wet runoff encounters under changing environmental conditions provides critical scientific support for sustainable water resource management and watershed security. Therefore, this study enhances the Generalized Additive Model for Location, Scale, and Shape (GAMLSS) through variable reconstruction and error correction, thus proposing an innovative methodology for assessing high-low runoff encounter probabilities in the Yangtze and Yellow River Basins under changing environmental conditions. Atmospheric circulation pattern analysis is further integrated to elucidate mechanisms underlying concurrent low-flow events. Key findings reveal that: (1) The Log-Normal distribution exhibits superior goodness-of-fit for runoff frequency in the Yangtze River’s headwater (source) and downstream regions, while Gamma and Normal distributions emerge as optimal for the upper and middle reaches, respectively. The Inverse Gaussian and Reverse Gumbel distributions demonstrate enhanced performance in the Yellow River Basin. (2) The optimized GAMLSS achieves remarkable accuracy, with empirical–theoretical value deviations constrained between − 0.1 and 0.1, and Nash-Sutcliffe Efficiency (NSE) values ranging from 0.9756 to 0.9966 across basins. (3) Analysis of 1963–2022 data identifies the highest dry-dry encounter probability (23.48 %) in the upper reaches of both basins, followed by headwater (20.89 %) and middle reaches (18.35 %), with the lowest probability (15.37 %) observed in lower reaches. (4) While the Zhimenda-Tangnaihai, Yichang-Toudaoguai, and Datong-Huayuankou combinations show decreasing dry encounter probabilities, the Dajin-Lanzhou combination exhibits a statistically significant upward trend (p < 0.05) in low-flow synchronicity. (5) Concurrent low-flow events in the Yangtze and Yellow Rivers are predominantly linked to two atmospheric circulation patterns: (a) the Lake Baikal high-pressure ridge, and (b) anomalous strengthening of the western Pacific subtropical high. This study advances hydrological extreme event prediction by integrating statistical modeling innovation with climatic mechanism analysis, providing critical insights for adaptive watershed management under global change scenarios.