A new framework for the identification of flash drought: multivariable and probabilistic statistic perspectives

Flash drought is distinguished with seasonal drought by its characteristics of sudden onset and rapid intensification, and has attracted increasing concern due to the severe negative impacts on socioeconomic development and ecosystems. To accurate identify the onset time and propagation of flash drought is still a big challenge, as it is influenced by many factors (e.g., deficit in precipitation, increase in evapotranspiration and temperature, shortage in soil moisture, and etc.). This study aims to develop a probabilistic and multivariable flash drought identification method (PMFDI) from the perspective of considering the interaction effects from different driving factors on flash drought occurrence and propagation with a multivariable statistical tool. The PMFDI uses the weekly standardized precipitation evapotranspiration index and the standardized soil moisture index to describe the abnormally changes in multiple hydrometeorological variables over a short duration. The linearly multivariable index (LMI) is employed to describe the correlation relationship between multiple hydrometeorological variables and to identify flash droughts from the aspects of sudden onset and rapid intensification. We select the Loess Plateau (LP) in China to verify the efficiency of the developed PMFDI with the comparison to the other two recently proposed flash drought identification methods. The results indicate that PMFDI can identify high- and low- frequency regions similar to those identified by the other two methods and can generally recognize actual flash droughts. In addition, the PMFDI can capture the comprehensive impact of multiple hydrometeorological variables as well as their interaction on flash drought occurrence, which provides a useful perspective to understand the propagation processes of the actual flash droughts in the LP.