| Description |
Convection-permitting ensembles (CPEs) can capture the large spatial variability and quantify the inherent uncertainty of precipitation forecasts in areas of complex terrain; however, such systems remain largely untested over the western U.S. In this study, we assess the capabilities of deterministic and probabilistic cool-season (October-March) quantitative precipitation forecasts (QPFs) produced by the high-resolution (3-km horizontal grid spacing), 10-member NCAR Ensemble using observations collected by Snow Telemetry (SNOTEL) stations at mountain locations across the western U.S and precipitation analyses from the Parameter-elevation Relationships on Independent Slopes Model (PRISM). We also examine the performance of operational forecast systems run by the National Centers for Environmental Prediction (NCEP) including the HRRR, NAM 3-km CONUS nest, GFS, and SREF. Overall, we find that higher resolution models, such as the HRRR, NAM-3km CONUS nest, and an individual member of the NCAR Ensemble, are more skillful than coarser models, especially over the interior ranges of the western U.S. This is likely because the high-resolution models better resolve topography, especially the narrow interior ranges, and thus better simulate orographic precipitation. Although probabilistic forecasts from the SREF are often more skillful than those generated by the NCAR Ensemble, the NCAR Ensemble generally outperforms its individual dynamical cores. While the NCAR Ensemble is shown to suffer from a spread deficiency, the SREF's multidynamical core configuration allows it to generate ample spread. These results should help guide future short-range model development and inform forecasters about the capabilities and limitations of several widely used deterministic and probabilistic modeling systems over the western U.S. |
