Multivariate Modeling of Projected Drought Frequency and Hazard over India

Abstract

India is one of the most drought-vulnerable countries in the world and faces at least one drought in one region or another every 3 years. To alleviate drought risk, an efficient policy framework is needed for water resources management, which in turn needs future projections of the spatiotemporal distribution of droughts. This study presents a spatiotemporal analysis of drought occurrence, frequency, and hazard. Meteorological data from a selected regional climate model for representative concentration pathways (RCP 8.5) was used to compute the Standardized Precipitation Evapotranspiration Index (SPEI) at a 12-month scale to characterize drought in different regions of India. An improved methodology was then developed to improve the current framework used for the development of severity–duration–frequency (SDF) curves. The improved methodology for developing SDF curves is copula based and uses Markov chain Monte Carlo (MCMC) simulation for parameter estimation under the Bayesian framework. Moreover, an improved fuzzy clustering-based drought hazard assessment measure, known as the Modified Drought Hazard Index (MDHI), is also presented. It was found that drought frequency is increasing with time for all the regions of India except for Region 2, i.e., Western Ghats. Further, a significant increase in potential evapotranspiration would cause a prolonged dryness in most of the regions during the twenty-first century. With the progression of time, drought severity associated with various durations is expected to significantly increase with the increase in duration for most of the regions. Drought hazard is expected to be the highest for the period 2071–2100 as compared to two other analyzed periods.