Downscaling Method Evaluation for Regional Precipitation and Temperature Over CONUS

Projecting the impacts of future climate change is crucially important to water managers planning to mitigate those impacts. Unfortunately, global climate model output is generally too coarse to provide the detailed, localized information necessary to prepare for those impacts, and the extreme computational cost to run more finely gridded models for a long enough time period can be prohibitive. Downscaling methods are a relatively inexpensive way to refine those climate model outputs into a more useful form. Here, we assess many downscaling methods, representing direct rescaling of climate model data, statistical approaches that incorporate meteorological information, a simplified atmospheric model, and full dynamic downscaling using regional climate models. These methods have been applied to both historical and future CMIP5 and CMIP6 precipitation and (in most cases) temperature, and the climate change signals can vary substantially between methods.

We develop a metric suite to assess these methods across nine different CONUS regions, evaluating them based on historical comparisons to a suite of observational datasets. These metrics include traditional metrics of precipitation and temperature, such as comparisons of seasonal means and spatial variability, measures of teleconnection patterns associated with El Nino – Southern Oscillation (ENSO), and a novel comparison of downscaling method performance given various dominant large scale meteorological conditions identified by a K-means weather typing algorithm. This metric suite allows evaluation across a wide array of conditions, as well as the quantification of uncertainty associated with climate model selection and observational dataset selected for both training and evaluation. The implications of these evaluations on future climate projections by these downscaling methods are then explored. By determining regions and large-scale meteorological conditions where some or all the downscaling methods evaluated here perform well or poorly, we provide actionable information to decision makers as to which downscaling dataset or method is most appropriate for a given application.