Assessing consistency in drought risks in India with multiple multivariate meteorological drought indices (MMDI) under climate change.

This study investigates the spatio-temporal consistency of different MMDI formulations and their role in meteorological drought characterization uncertainty under historic and future climates using ERA5 reanalysis, and outputs from eight Coupled Model Intercomparison Project Phase 6 models, respectively, across different climate zones and shared socioeconomic pathways (SSP) in the Indian subcontinent. Six MMDI formulations namely the Standardized Precipitation Evaporation Index (SPEI), Reconnaissance Drought Index (RDI), and self-calibrated Palmer Drought Severity Index (scPDSI), Standardized Palmer Drought Index (SPDI), Standardized Moisture Anomaly Index (SZI) and Supply Demand Drought Index (SDDI) are used. A suite of analysis including agreement mapping, category difference analysis and uncertainty contribution analysis using global sensitivity analysis (GSA) are employed to quantify the consistency of MMDIs and uncertainty in drought characterization due to the MMDI formulation. The variation in MMDI consistency due to different reference evapotranspiration (ETo) methods is also studied. Results demonstrate strong agreement among the MMDIs under historic climate. Under climate change scenarios our findings demonstrate broad agreement among majority of the MMDIs across the study domain, but in substantial areas where MMDI not agree, especially for higher emission scenarios and arid zones. Increased uncertainty under climate change is due to SDDI and SPEI projecting dryer conditions while scPDSI projecting wetter conditions in the far future period owing to varying degrees of sensitivity of MMDIs to its constituent variables (Precipitation and ETo). Results also show that the uncertainty due to MMDIs varied considerably based on ETo methods as well. Finally, based on GSA analysis, the most significant sources of uncertainty in drought projections under climate change are attributed to MMDI-GCM interactions and MMDIs for the Penman-Monteith method. Discrepancies in drought estimates caused by the MMDI selection highlight the need for careful evaluation of drought indices before adopting for climate change impact assessment.