标题：The Upper Tail of Precipitation in Convection‐Permitting Regional Climate Models and Their Utility in Nonstationary Rainfall and Flood Frequency Analysis
摘要：Computational advances have made atmospheric modeling at convection‐permitting (≤4 km) grid spacings increasingly feasible. These simulations hold great promise in the projection of climate change impacts including rainfall and flood extremes. The relatively short model runs that are currently feasible, however, inhibit the assessment of the upper tail of rainfall and flood quantiles using conventional statistical methods. Stochastic storm transposition (SST) and process‐based flood frequency analysis are two approaches that together can help to mitigate this limitation. SST generates large numbers of extreme rainfall scenarios by temporal resampling and geospatial transposition of rainfall fields from relatively short data sets. Coupling SST with process‐based flood frequency analysis enables exploration of flood behavior at a range of spatial and temporal scales. We apply these approaches with outputs of 13‐year simulations of regional climate to examine changes in extreme rainfall and flood quantiles up to the 500‐year recurrence interval in a medium‐sized watershed in the Midwestern United States. Intensification of extreme precipitation across a range of spatial and temporal scales is identified in future climate; changes in flood magnitudes depend on watershed area, with small watersheds exhibiting the greatest increases due to their limited capacity to attenuate flood peaks. Flood seasonality and snowmelt are predicted to be earlier in the year under projected warming, while the most extreme floods continue to occur in early summer. Findings highlight both the potential and limitations of convection‐resolving climate models to help understand possible changes in rainfall and flood frequency across watershed scales. Plain Language Abstract High‐resolution “convection‐permitting” regional climate model simulations hold great promise in projection of climate change impacts including extreme rainfall and flooding. The relatively short (~10‐year) model runs that are currently feasible, however, are insufficient for examining very rare events like 100‐year storms and floods. Meanwhile, existing rainfall and flood data sets have a number of shortcomings that make it difficult to understand how floods have and will continue to change. In this study, we use several novel computer modeling methods to help mitigate these limitations. We apply these methods together with detailed simulations of flood hydrology and high‐resolution regional climate simulation results to examine current and future extreme rainfall and flooding in an agricultural watershed in northeastern Iowa, in the Midwestern United States. Floods there are projected to become more severe, driven by complex seasonal changes in rainfall, temperature, and snow. The magnitude of these changes depends on upstream watershed area. This work demonstrates how cutting‐edge climate and hydrology simulations and methods, together with flood theory and data, can help to predict future changes in flooding.
关键词：process‐based flood hydrology;stochastic storm transposition;rainfall and flood frequency analysis;convection‐permitting regional climate models
