Meteorological drought predictability dynamics and possible driving mechanisms in a changing environment in the Loess Plateau, China

Abstract

Drought forecasting is important for water resources management and effective response to drought, and the predictability of drought may change under a changing environment. Most of the studies have focused on developing drought forecasting techniques, but limited attention has been made to the theory of drought predictability, such as dynamics of meteorological drought predictability and possible driving mechanism. Here, we characterized the predictability of meteorological drought, based on the Kling-Gupta efficiency (KGE") coefficient of support vector machine regression model. Then we measured the spatial distribution, agglomeration, and dynamic changes of drought predictability, and quantitatively analyzed the main driving forces and relationships of the spatial and temporal dynamics. The Loess Plateau (LP), which is a drought-prone region with frail ecological environment in China, was chosen as a case study. Results indicated that: (1) drought predictability in the western region was higher than that in the eastern region of the LP, with the hot spots concentrated in the western sandy land and agricultural irrigation; (2) meteorological drought predictability in the LP showed a downward trend from 1962 to 2019 under the changing environment, which the autumn drought predictability declined significantly; (3) meteorological, terrestrial factors and air-sea coupling elements dominated the spatial-temporal pattern of meteorological drought predictability via strongly affecting the coefficient of variation of drought index series, and related causal paths were explored. This study sheds new light on drought predictability dynamics under a changing environment, and has significance for improving the ability of drought forecasting, warning, and mitigation.