Evaluation of extreme events in global coupled climate models by satellite gravimetry

Under the assumption that a warming climate leads to an intensification of the global water cycle, it can be hypothesized that also the occurrence frequency and severity of extreme events such as droughts or floods will increase in the upcoming decades to centuries. Global coupled climate models, which project the future evolution of various variables of the Earth's climate system are important tools for the analysis of such expected changes. To assess the reliability of the models and to identify possible systematic discrepancies, it is essential to evaluate the model output against observations. In this study, present and future occurrences of extreme events are analysed in water storage time series simulated by coupled global climate models participating in the Coupled Model Intercomparison Project Phase 6 (CMIP6) and compared against spatio-temporal changes in water mass derived from GRACE and GRACE-FO. This comparison is based on Extreme Value Theory, as the exact timing of modelled extreme events cannot be assessed by observations due to the stochastic behavior of climate variability in unconstrained model experiments. From estimated extreme value distributions return levels are calculated, a quantity describing the magnitude or frequency of extreme values.  Challenges that have to be overcome in the analysis are the non-stationary data and the relatively short time span of the GRACE observations. The latter issue is addressed by additionally assessing GRACE-based water storage reconstructions available over many decades. This study provides insights into the ability of global climate models to model the occurrence of TWS extremes, namely unusual dry and wet phases. It also examines whether the climate model projections predict an increasing intensity of extreme events.