Storylines for future changes in regional extreme precipitation

Despite the high confidence in the overall intensification of extreme precipitation at global scale in response to warming, uncertainties in regional intensity and spatial distribution remain large. Changes due to thermodynamical processes are largely consistent across global climate models and account for the globally homogeneous increase in extreme precipitation, whereas changes in dynamical processes modify the regional responses of extreme precipitation and are also mainly responsible for the projected uncertainties. We here aim at developing and characterizing different storylines for regional changes in extreme precipitation.

The latest CMIP6 multi-model ensembles allow for disentangling the three potential sources of uncertainty across climate models. Specifically, we find that scenario uncertainty in annual maximum precipitation (Rx1day) per K global warming is relatively small except over tropical Pacific and subtropical Africa regions. In order to isolate the inter-model uncertainty, which primarily relates to different responses in atmospheric dynamics, we here average multiple members from initial condition ensembles within CMIP6. To further characterize the dynamical processes and their effects on extreme precipitation, clusters of different model responses to climate change are identified for Mediterranean and Asian monsoon regions. The clusters are defined based on metrics that characterize regional dynamics. Different atmospheric fields are evaluated to develop physical storylines of future changes in extreme precipitation with model clusters representing similar dynamical responses. Specifically, an overall anticyclonic change is found to be characteristic for projections indicating highest reduction in Rx1day over Mediterranean. Future extreme precipitation changes over South Asia are linked to the models' representation of western coastal precipitation and orographic rainfall over the Himalaya. Other potential factors characterizing dynamical responses will be further examined to help understand changes in regional projections of extreme precipitation.