Spatio-temporal variation of meteorological, hydrological and agricultural drought vulnerability: Insights from statistical, machine learning and wavelet analysis

The study of how agricultural drought (AD) is responsible for meteorological drought (MD) and hydrological drought (HD) is crucial for drought prevention and the socio-economic development of a nation. This is due to AD constitutes a significant threat to the nation's food productivity and security. In depth comprehension and mitigation of drought incidents depend on understanding their frequency and propagation patterns. In this study, spatio-temporal variation of three types of droughts have been assessed in the sub-tropical environment of eastern India. In this perspective, seasonal i.e., pre-monsoon, monsoon, post-monsoon, and winter MD, HD, and AD were assessed considering Standardized Precipitation Index (SPI), Standardized Water Level Index (SWI), and Standardized Soil Moisture Index (SSMI) statistical tool respectively in sub-tropical agro-climatic zone of eastern India. In addition to this, spatial drought vulnerability of MD, HD and AD was assessed using Analytic Hierarchy Process (AHP) considering suitable factors for each drought type, and overall drought vulnerability was assessed using “Random Forest (RF)” and “Artificial Neural Network (ANN)” methods. Furthermore, drought periodicity has been measured using a wavelet power spectrum analysis. The result of seasonal drought revealed that pre-monsoon season has more frequent drought occurrences than other seasons among the applied three types of droughts. The outcomes of overall drought vulnerability revealed that RF gives the optimum result followed by ANN i.e., 0.841 and 0.828, respectively, for validation purposes. The periodicity of drought ranges from 0.25 to 4 as obtained from wavelet analysis. In general, this research on how AD spreads from MD and HD is crucial for drought resilience, drought management, and food security among the stakeholders and policymakers for achieving the SDGs.