A two-stage Bayesian multi-model framework for improving multidimensional drought risk projections over China

Abstract. Understanding future drought risk is a prerequisite for developing climate change adaptation strategies and for enhancing disaster resilience. In this study, we develop multi-model probabilistic projections of multidimensional drought risks under two representative emission scenarios (RCP4.5 and RCP8.5) through a copula-based Bayesian framework. An ensemble of five regional climate simulations, including four from the CORDEX East Asia experiment and one from the Providing REgional Climate Impacts for Studies (PRECIS) simulation, is used to project future changes in hydroclimatic regimes over China. A new Bayesian copula approach is introduced to uncover underlying interactions of drought characteristics and associated uncertainties over 10 climate divisions of China. The proposed Bayesian framework explicitly addresses the cascade of uncertainty in high-resolution projections of multidimensional drought risks. Our findings reveal that precipitation and potential evapotranspiration (PET) are projected to increase for most areas of China, while increasing radiative forcing is expected to amplify the increase in PET but does not cause significant changes in the precipitation projection. In addition, the drought duration and severity are projected to substantially increase for most areas of China. The estimated drought risks in China are expected to become more than double under both emission scenarios. The extreme droughts are projected to intensify in terms of frequency and associated risks as the radiative forcing increases.