Evolution characterization and attribution analysis of hydrological drought in Ganjiang River based on hydrological model and deep learning coupling

Studying evolution characteristics and attribution analysis of hydrological drought in the Ganjiang River Basin in recent years can better prevent and control hydrological drought in Ganjiang River basin. Using monthly runoff data from Waizhou station in Ganjiang spanning from 1961 to 2020, this article first employs two mutation testing methods to comprehensively identify the year of runoff mutation. Afterward, we utilize the ABCD hydrological model, coupled with seven deep learning algorithms to simulate the streamflow change of Waizhou station in Ganjiang River basin. Finally, the standardized runoff index is applied to describe the hydrological drought, and we analyze the evolution characteristics of hydrological drought and quantitatively assess the effects of human interventions and climate change on hydrological drought in the Ganjiang River Basin. The insights drawn from this research can be summarized as follows: (1) The results of the mutation analysis method indicate that there was a significant mutation in runoff in 1991. (2) The ABCD model can perform well in simulating and predicting runoff, with accuracies reaching 0.82 and 0.88. (3) Combining the ABCD hydrological model with deep learning algorithms can improve the accuracy of simulating runoff changes in the Ganjiang River. Among them, the ABCD-random forest method has the highest accuracy, reaching 0.89 and 0.94. (4) Climate change has a stronger impact on monthly hydrological drought compared to human activities. (5) Climatic factors are the primary determinants of seasonal hydrological drought changes. The findings of this study could provide a valuable reference for the optimal use of water resources and the proactive management of hydrological disasters in the Ganjiang area.