GIS-Based Drought Assessment in Climate Change Context: A Case Study for Sone Command, Bihar

Abstract

Indian economy by far is primarily agro-based which in turn depends on the monsoon showers. Seasonal variations in monsoon lead to dry and wet spells. Inadequacy in rainfall for different needs leads to drought, and drought indicators are generally used to describe different drought conditions. In this paper, an attempt has been made to assess meteorological drought in the climate change context in Sone Command, Bihar. Three different statistical downscaling techniques, namely Wavelet decomposed neural network (WNN), feed-forward neural network and support vector machine, are used, and their performance is compared in this study. Rainfall is projected for eight raingauge stations for the period 2015–2045. The model performance is evaluated using different metrics, and the outperformance of WNN is found in projecting the rainfall variability. Standardized Precipitation Index (SPI) is calculated using the best performed model rainfall for future conditions. An attempt has also been made for the spatial drought analysis during the projected period of 2015–2045. It is found that stress on drought is prevalent during November–May for the projected period. The areas covered under different drought zones in the Sone Command ranges from near normal to severely dry. For rain-fed crops, such spatial distribution maps are useful for better crop yield with minimum chances of crop water stress. In addition, the rainfall data are projected for the period 2015–2045 from two selected GCMs: MPI-ESM-MR (for RCP 2.6 scenario) and CMCC-CMS (for RCP 4.5 and RCP 8.5 scenarios) using bilinear interpolation method of downscaling. The SPI is calculated for the projected rainfall data, and the results are compared with that of the best performed statistical downscaling model. The results reveal the occurrence of drought of higher severity in the projected period.