Response Analysis of Terrestrial Water Storage Components to Drought Based on Random Forests During 2011–2020 in Yunnan, China

Abstract

The interaction between terrestrial water storage (TWS) and drought is a crucial aspect of hydrological dynamics. We utilized data from the gravity recovery and climate experiment (GRACE) to derive the TWS of Yunnan from 2011 to 2020 and calculated the GRACE drought severity index (GRACE-DSI) to assess hydrological drought events based on the run theory. To understand the underlying causes of drought, we analyzed precipitation data alongside the spatial and temporal distribution of drought events. Each component of TWS was obtained through hydrological models, and we used random forest to determine the contribution of each component to drought. Our findings revealed that Yunnan experienced four drought events from 2011 to 2020, with the third and fourth drought events between 2019 and 2020 being particularly severe. The occurrence of these drought events was primarily attributed to a relative lack of precipitation. Among the components of TWS, water storage in groundwater, canopy, soil, and lake were identified as having the highest contribution rates to drought. We also applied random forest to simulate the contribution rates of each TWS component to drought in other regions of China. Our analysis confirmed that groundwater in the North China Plain, lake, and soil water in the Yangtze River middle-downstream Plain, snow, and groundwater in Tianshan Mountain were the components with the highest contribution rates to drought, respectively. Finally, we found that the contribution of TWS components to drought in different regions of Yunnan Province was significantly different. This novel method was valuable for hydrological drought management.

Keywords

• Drought index
• GRACE
• hydrological drought
• random forests (RFs)
• terrestrial water storage (TWS)