Detecting nonlinear dependence between drought and disaster using information entropy and identification of driving factors by the DPSEEA

Drought disasters have significant impacts on societies, economies and ecological environment. A comprehensive understanding of disaster-causing processes is essential for drought disaster early warning and prevention. This paper commenced with the drought disaster itself and categorized historical drought disasters from textual records. A framework that considered the linear and non-linear dependencies between droughts and disasters was developed. Nonlinear dependencies between droughts and disasters are recognized based on information entropy. The Driving-Pressure-State-Exposure-Effect-Action (DPSEEA) framework was applied for analyzing the factors causing disaster. A case study of China showed that the disaster was severe in the northern part of Northeast China, North China, the eastern part of Northwest China, and the southern part of Southwest China. The frequency of disaster in southern China exhibited an increasing annual trend. Central and South China had the shortest response time from drought to disaster, mainly within two months. In contrast, the western part of Southwest China and Northwest China had the longest time, concentrated in 9–12 months. Key factors of drought disaster were water consumption per 10,000 yuan of GDP, proportion of rural residents and GDP per capita. These findings suggest that disaster mitigation measures should focus on three areas: improving water-use efficiency, restructuring agricultural practices, and promoting economic upgrading. Fig. 1 Framework of drought disaster study. • Historical drought disasters are classified. • Nonlinear dependence between drought and disaster using information entropy is found. • Key factors of drought disaster by the DPSEEA are identified. • Remarkable differences of drought disaster in China are presented.