Extreme precipitation is expected to intensify as the climate warms, but the magnitude of the increase will vary regionally. In many cases, global climate models (GCMs) are not well-suited to project the changes in extreme precipitation due to their coarse resolution, particularly over complex terrain. Here, we analyze an unprecedented suite of eight bias-corrected dynamically downscaled GCMs over the western U.S., which allow us to assess extreme precipitation changes at high resolution. We pool data across the downscaled ensemble to adequately sample extreme events and characterize 99.99th percentile precipitation in Los Angeles County, home to 10M people. This high-resolution data allows us to advise a county government agency on expected changes in local extreme precipitation so that they may consider the suitability of their urban design standards in the coming decades. We find that the 99.99th percentile precipitation event is expected to increase by about 6.5% per degree Celsius global warming on average over Los Angeles County. However, Los Angeles County contains numerous micro-climates associated with, e.g., high mountains, marine ecosystems, and urban centers, whose future changes the downscaled projections are uniquely suited to predict. The absolute increases in extreme precipitation are shown to be magnified in the mountains and minimized in the desert regions. The agency will use this data to become more resilient to climate change. This project underscores the importance of stakeholder engagement with scientists for translating climate data into actionable guidance.

Flood risk across the western United States (US) has generally shown decreasing trends in recent decades. This region’s extreme streamflow is highly influenced by natural variability, which could either mask or amplify anthropogenic streamflow trends. In this study, we utilize a technique known as dynamical adjustment to assess historical (1970-2020) annual maximum 1-day streamflow (Qx1d) from unregulated basins across the western US with and without the impact of natural variability. After removing natural variability, the fraction of basins with a positive (>5%) trend in Qx1d shifts from 27% to 61%. Basins with significantly increasing (decreasing) Qx1d trends after dynamical adjustment exhibit weak (strong) drying, and furthermore are associated with intensifying precipitation extremes and/or large decreases in snowpack. Increasing flood risk will likely emerge for such basins as the current phase of natural decadal variability shifts, and anthropogenic signals continue to intensify.