Regional Climate Models Differ in Extreme Precipitation Event Forecasts

Extreme precipitation events are proposed to be risk factors likely to increase under global warming scenarios. Estimates of local extreme event likelihood require the use of Regional Climate Models (RCMs), which run at a higher spatial resolution than Global Climate Models (GCMs). In a study by Schleip et al. (2010), six RCMs were compared. They were forced with a common set of reanalysis data, which was created by running a climate model that was fed real-world data for a 20-year simulation period. The area analyzed was North America, where winter precipitation was the response variable, and the 100-year extremum of daily winter precipitation was the test statistic, extreme values of which were estimated by fitting a tailed distribution to the data, taking into account their spatial aspects.

The six RCMs showed similar general spatial patterns of extrema across North America, with the highest extremes in the Southeast and along the West Coast. However, when comparing absolute levels, which are most relevant to risk forecasts, the models exhibited strong disagreement. The lowest-predicting model was low almost everywhere in North America compared to the mean of the six models; and, similarly, the highest-predicting model was above the mean almost everywhere. In fact, the difference between the two models was almost 60mm of daily precipitation (for the 100-year extreme event) over much of the United States.

The other four models showed greatly differing spatial patterns of extremes from each other, which differences were found to be statistically significant by F test. And the authors speculate that when driven by multiple GCMs rather than reanalysis data, the range of extreme outcomes would only increase. Thus, the study of Schliep et al. leads to the conclusion that depending on regional models for making policy about extreme rainfall events would be very uncertain and, therefore, essentially an exercise in futility ... and a dangerous one at that!