Bias-Correction of Extreme Temperatures and Precipitation in NA-CORDEX Regional Climate Model Output

Climate model outputs typically contain significant biases that can hinder their use in impacts analysis. Frequently, these biases are non-linear, and have a greater effect in the tails of the distribution than they do near the mean. Extremes of temperature and precipitation also have a disproportionate effect on the impacts of climate, making it important to understand and correct biases across the entire distribution of values for use in climate change impacts analysis.Analysis of various bias correction methodologies by Teutschbein & Seibert (2012) showed that distribution mapping has the best overall performance. Further work by McGinnis, et al. (2015) has shown that kernel density distribution mapping (KDDM) outperforms other distribution mapping techniques, and that it provides effective correction of extremes that simple mean and variance adjustments cannot.We use KDDM to bias-correct daily minimum and maximum temperatures and precipitation from NA-CORDEX, the North American branch of CORDEX, the COordinated Regional Downscaling EXperiment, which nests high-resolution regional climate models (RCMs) within general circulation models (GCMs) over a limited domain. We bias-correct the model output against the University of Idaho's METDATA high-resolution gridded daily observational dataset (Abatzoglou, 2012). We then examine the resulting changes to the climate change signal in the bulk and in the tails of the distribution of values.The structure of the NA-CORDEX experiment allows us to compare the effects of bias correction on the climate change signal at two different spatial resolutions (25 km and 50 km), for different emissions scenarios (RCP 4.5 and RCP 8.5), and for multiple different combinations of regional climate model and driving global model.