Sensitivity of PET estimation methods towards drought characterization under climate change in the Indian subcontinent

In the Indian subcontinent, the uncertainty associated with potential evapotranspiration (PET) over drought characterization is inadequately studied. This study was conducted to understand the sensitivity of PET estimation methods towards drought characterization using multiple PET-based drought indices under future climate change. We used eleven PET estimation methods (Blaney-Criddle (BC), Hamon (HM), Hargreaves (HG), Kharrufa (KF), Thornwaithe (TW), Dalton (DN), Meyer (MR), Irmak-Rn (IRN), Irmak-Rs (IRS), Priestley-Taylor (PT), and Penman-Monteith (PM)) for the future period (from the Coupled Model Intercomparison Project 5). Further, for drought characterization six PET-based drought indices are utilized in this study: the Standardized Precipitation Evaporation Index (SPEI), the Supply-Demand Drought Index (SDDI), the Reconnaissance Drought Index (RDI), the self-calibrated Palmer Drought Severity Index (sc-PDSI), the Standardized Moisture Anomaly Index (SZI), and the Standardized Palmer Drought Index (SPDI). We also employed a variance-based global sensitivity analysis to determine the relative sensitivity of projected drought indices to the GCM and PET estimation methodologies under climate change scenarios. Results indicate that different PET-based drought indices show vastly different drought projections for the future, which is highly influenced by the PET methods. Overall, SPEI and SDDI produce comparable results, indicating an increase in future drought estimates compared to the rest (RDI, SPDI, SZI, and sc-PDSI). The TW method reported higher drought projections compared to other PET methods irrespective of the drought indices.  This is due to the fact that the TW method also showed the highest increase in PET compared to the rest of the methods. Results from the sensitivity analysis indicate that all the drought indices are more sensitive to the choice of PET methods compared to the GCM. However, analysis was done after excluding the TW approach significantly altered the sensitivity, and GCMs were found to be more sensitive compared to PET methods. The results from this study reveal that drought projections derived from multiple PET estimation methodologies indicate drier conditions in the future, albeit at variable levels. Thus, the selection of the PET estimation method and drought index will be crucial in the Indian subcontinent for future drought investigations.