Uncertainty separation of drought projection in the 21st century using SMILEs and CMIP6

Abstract

Reliable projection of future drought conditions is critical for informing adaptation in response to future climate change. However, the uncertainty associated with drought projection impedes the precise estimation of drought risk and consequently efficient adaptation strategy. To better understand the role of uncertainty in drought projection, this study distinguished the total uncertainty of drought projection into scenario uncertainty, model uncertainty, and internal variability uncertainty by variance analysis using realizations from the Single Model Initial-condition Large Ensembles (SMILEs) and the Coupled Model Intercomparison Project Phase 6 (CMIP6). The results showed that SMILEs differ from CMIP6 mainly in the magnitude of internal variability and model uncertainty. The contribution of internal variability to total drought projection uncertainty in SMILEs (31–56 %) is greater than that in CMIP6 (9–21 %), while model uncertainty of SMILEs (36–41 %) is nearly half that of CMIP6 (67–76 %). The estimation of total uncertainty in drought projection is comparable between SMILEs and CMIP6, indicating that total uncertainty attains a minimum in the mid-21st century (the 2060 s for SMILEs and 2050 s for CMIP6). In addition, SMILEs and CMIP6 consistently show that model uncertainty is dominant in tropical regions, and scenario uncertainty is the main uncertainty contributor in western North America, eastern South America, the Mediterranean, and southern Australia.