Assessing the Characteristics of Extreme Rainfall Events in Climate Models

Extreme precipitation events are expected to increase across the globe in response to rising global temperatures. These events can have devastating effects on human life, wildlife, and infrastructure. However, current climate models do a poor job of simulating high-level precipitation statistics compared to observations. As a result, there is lower confidence in understanding the extent and severity of projected precipitation changes. Several techniques, including increased horizontal resolution (e.g., ~25 km) and super-parameterization (i.e., replacing parameterizations of convection with embedded km-scale cloud resolving models), have been shown to improve the representation of extreme precipitation events in climate models. In this study, we build upon these previous results to further compare these two methods. We focus on the United States and use observational data from a variety of sources including rain gauges, satellite, and radar. We evaluate not only the frequency and intensity of precipitation, but also assess the models’ ability to represent consecutive heavy rainfall events, which can contribute significantly to flooding conditions. We find that the way convective processes are represented and resolved in the models influence their ability to realistically capture many precipitation statistics, including consecutive rain days.