Integration of Multiple Drought Indices for Agriculture Drought Categorization and Impact Assessment in Central India

Drought is a complex and multidimensional phenomenon affecting the global population. The widespread impacts of drought propagate through the climatic and hydrological cycle and affect the socio-economic security of the related stakeholders, especially farmers. Countries like India use several indices to determine the severity of the drought for governmental relief and mitigation measures, which is crucial for farmers facing agricultural stress and failures. However, the use of single or several separate drought indices cannot capture the combined effect of principal drivers responsible for the drought, where the effect of groundwater availability for agriculture is often neglected despite its heavy use in irrigation through groundwater extraction. In this study, we focus on the multidimensional response of drought in a single joint index to better capture the spatiotemporal variability in drought severity. The semi-arid region of Marathwada from central India, which frequently faces drought and is infamous for farmer suicides due to agriculture failures is taken as the study area. The response of hydroclimatic variables viz. precipitation, evapotranspiration, soil moisture, surface runoff, and groundwater storage were captured in their respective standardized indices (SPEI, SSI, SRI, and SGI respectively) which were then used to construct the Joint Drought Index (JDI) using two principal methods: 1) Principal Component Analysis (PCA) and 2) Gaussian copula. Both the methods were found to be capable of identifying the severity of the drought along with its onset, duration, and termination. Although individual indices such as SPI can sometimes acknowledge the meteorological response better, the JDI has the potential of capturing the response of multiple hydrological variables together at once for drought monitoring and assessment. During the period between 2003 to 2020, the drought of 2015 was identified as exceptionally severe in both the methods, where copula could better accommodate the severity of every integrated index whereas PCA averages the response of the variables to drought by allocating the weights to each index for each month.