Compound Wet/Windy Extremes in Australian Observational and Reanalysis Datasets

Abstract The co-occurrence of extreme rain with extreme wind can result in enhanced impacts, and there is interest in understanding their occurrence, drivers, and potential changes. These compound events are frequently identified based on the co-occurrence of values above a locally defined extreme of the distribution, allowing them to be easily translated between datasets. In this study, we instead use consistent, impact-informed thresholds for daily maximum hourly rainfall and daily maximum wind gusts based on hourly station data from Australia, to better identify extreme events more likely to be associated with damaging impacts. Using this dataset, we evaluate the ability of three reanalyses with different spatial resolutions to identify wet/windy compound events and find that the global 0.25° ERA5 reanalysis underestimates the frequency of wet/windy compound extremes in tropical regions including those linked to thunderstorm activity when compared to observational data or a 0.04°-resolution regional reanalysis. These results highlight important uncertainties in global analyses and projections of wet/windy extremes when using datasets that do not resolve convective-scale processes.