Improving Projections of Future Sub-daily Precipitation Events in a Changing Climate Across the U.S. Gulf Coast

Projection of future precipitation events under climate change scenarios is crucial for protecting our local communities and infrastructure from potential flooding and revisiting the management of regulating water infrastructure (e.g., dams and gated spillways). Existing projections of extreme precipitation have been mainly developed using daily data, with very limited attention for using sub-daily datasets. In this study, we applied a statistical hybrid method to improve the bias correction and projection of future rainfall events across the U.S. Gulf Coast. We applied the statistical hybrid method to bias-correct daily precipitation data simulated by a Global Climate Model (GCM). We then projected future precipitation time series under multiple CMIP6 scenarios called shared socioeconomic pathways (SSPs). In-situ (NOAA) and reanalysis datasets (ERA5 and Daymet) were applied as observations for the bias correction to evaluate the suitability of. A paired stochastic approach was then employed to disaggregate the daily precipitation data to sub-daily. Our findings revealed that standard statistical bias correction techniques significantly underestimate low-frequent extreme events, with rainfall observations themselves contributing to this underestimation. The statistical hybrid method provided a more accurate representation of these events. Analysis of future climate conditions indicated that sub-daily extreme precipitation events are likely to become more intense than those of longer durations. Furthermore, the most extreme events (higher quantiles) are expected to face greater changes. Our analyses showed that the use of daily data and reanalysis products can result in substantial errors in bias correction; the errors dominated the ones by the choice of bias correction method. In addition to the changes in intensity, the frequency of extreme rainfall events was projected to increase more compared to minor and moderate events. This study has implications for flood management, design of infrastructure and operation of regulating infrastructures.