The mesoscale precipitation distribution in mid-latitude continental regions: observational uncertainty and evaluation of 25-km global model simulations.

Mid-latitude precipitation over land exhibits a high degree of variability due to the complex interaction of governing atmospheric processes with coastlines, the heterogeneous land surface, and orography. General circulation models (GCMs) have traditionally shown limited ability in capturing variability in the mesoscale range (here ~50-500 km), due to their low resolution. Recent advances in resolution have provided ensembles of multidecadal climate simulations with GCMs using ~25 km grid spacing. Here, we assess this class of GCM simulations, from the UPSCALE (UK on PrACE - weather-resolving Simulations of Climate for globAL Environmental risk) campaign. Increased model resolution also poses new challenges to the observational datasets used to evaluate models. Global gridded data products (e.g. from the Global Precipitation Climatology Project, GPCP) are invaluable for assessing large-scale precipitation features, but may not sufficiently resolve mesoscale structures. In the absence of alternative estimates, the intercomparison of specialised, regional observational datasets may be the only way to gain insight into the uncertainties associated with these observations. We focus on three mid-latitude continental regions where gridded precipitation observations based on higher-density gauge networks are available, complementing the global data sets: Europe (with a particular emphasis on the Alps), South and East Asia, and the continental US. Additional motivation, and opportunity, arises from continuing efforts to quantify the components of the global radiation budget and water cycle. Recent estimates based on radiation measurements suggest that the global mean precipitation/evaporation may be up to 10 Wm-2 (about 0.35 mm day-1) larger than the estimate obtained from GPCP. While the main part of this discrepancy is thought to be due to the underestimation of remotely-sensed ocean precipitation, there is also considerable uncertainty about 'unobserved' precipitation over land, in particular in the form of snow in regions of high latitude/altitude. Combining the strength of these different observations, we present case studies of how area-averaged mountain precipitation is represented in different observational datasets and by a GCM at different resolutions.