Projecting Changes in Extreme Precipitation in the Midwestern United States Using North American Regional Climate Change Assessment Program (NARCCAP) Regional Climate Models

Based on the physics of global circulation, many expect an enhanced greenhouse effect to lead to a more active hydrological cycle with more precipitation on average (Hennessy et al. 1997). This expected increase has been found in observations (Zhang et al. 2007) and has also been suggested by climate models, although these models are not consistent with respect to the spatial and temporal variability about this change. An increase in mean precipitation depth, assuming no change in the shape of the frequency distribution, would imply an increased frequency of heavy-precipitation events. However, some studies (Hennessy et al. 1997, Allen and Soden, 2008) also suggest the increase in these extreme events could be disproportionate to the change in the mean, with a greater fraction of the total precipitation being delivered by such heavy precipitation events. Such a shift towards heavy events is a common conclusion of climate models (Cubasch et al. 2001, Meehl et al. 2007) as well as analyses of observed rainfall data at the continental scale (Easterling 2000, Kunkel 2003, Groisman 2005, Min et al. 2011). However, there is great spatial variation of this average pattern. This study aims to establish likely future projections for how extreme precipitation frequency and magnitude could change in the Midwestern region of the United States, and investigate the spatial variation of such changes within the area. Present global climate models (GCMs) typically produce results at the spatial resolution of 150-300 km. This level of spatial resolution of GCMs is insufficient for establishing localized future climate projections and examining their spatial variations at the scale of a state. For increased spatial resolution, we used a set of Regional Climate Models (RCMs) run by National Center for Atmospheric Research (NCAR) under the North American Regional Climate Change Assessment Program (NARCCAP). RCMs involve nesting a higher resolution climate models within a coarser resolution GCM. The GCM output is used to define boundary conditions around a limited domain, within which RCM further models the physical dynamics of the climate system. These RCMs are designed to produce high resolution climate change simulations in order to investigate uncertainties in regional scale projections of future climate and generate climate change scenarios for use in regional and local impacts research (Mearns et al, 2009). This study aims to achieve two main objectives. First, we evaluate the performance of NARCCAP models in terms of whether they capture the frequency distribution of daily precipitation data. This evaluation is based on a comparison of retrospective model runs with observed station-based daily precipitation data. Second, based on the evaluation, we correct the bias in mean precipitation and frequency distribution of precipitation output from RCMs. After the model biases have been corrected, we then project future changes of mean and extreme precipitation patterns in the Midwest Region