Differences in representation of extreme precipitation events in two high resolution models

Abstract

High resolution regional climate models are needed to understand how climate change will impact extreme precipitation. Current state-of-the-art climate models are Convection Permitting Models (CPMs) at kilometre scale grid-spacing. CPMs are often used together with convective parameterised Regional Climate Models (RCMs) due to high computational costs of CPMs. This study compares the representation of extreme precipitation events between a 12 km resolution RCM and a 2.2 km resolution CPM. Precipitation events are tracked in both models, and extreme events, identified by peak intensity, are analysed in a Northern European case area. Extreme event tracks show large differences in both location and movement patterns between the CPM and RCM. This indicates that different event types are sampled in the two models, with differences extending to much larger scales. We visualise event-development using area-intensity evolution diagrams. This reveals that for the 100 most extreme events, the RCM data is likely dominated by physically implausible events, so called ‘grid-point storms’, with unrealistically high intensities. For the 1000 and 10,000 most extreme events, intensities are higher for CPM events, while areas are larger for RCM extreme events. Sampling extreme events by season shows that differences between RCM and CPM in intensity and area in the top 100 extreme events are largest in autumn and winter, while for the top 1000 and top 10,000 events differences are largest in summer. Overall this study indicates that extreme precipitation projections from traditional coarse resolution RCMs need to be used with caution, due to the possible influence of grid-point storms.