Validation of CPM’s wind and precipitation field against observations and the highresolution reanalysis dataset SPHERA

Abstract

The correct representation of fine-scale atmospheric processes, like convection, is vital for predicting extreme weather events and CPMs have already shown to provide more reliable representation of extreme precipitation. However, in most cases their validation is limited to the precipitation field and based on sparse in-situ observations or coarser resolution observational gridded dataset. In this study, we first explore whether high-resolution (i.e., grid spacing 2.2km) reanalysis product SPHERA provides a realistic representation of the in-situ observations, thus offering  a comprehensive overview of the atmosphere at fine scale and functioning as a reliable reference dataset for CPMs evaluation. Then the sub-daily precipitation and wind fields of the CPMs ensemble from the CORDEX Flagship Pilot project on Convective Phenomena over Europe and the Mediterranean (FPS Convection) is validated against both in-situ observation and SPHERA. The validation focuses on extreme quantiles, spatial variability and event representation with a quantile based approach (i.e., the event starts when atmospheric variables are above a certain quantile, and ends when it goes below). Results show a general good agreement between in-situ observations and SPHERA, that is found to be a good reference dataset to evaluate the CPM models. When looking at the extreme quantiles, the CPMs well represent  both wind and precipitation fields, although they underestimate heavy precipitation in summer (i.e., June-July-August). Similarly, the spatial distribution of precipitation and wind is well represented for all the season, with a decrease in the spatial variability and spatial correlation for the heavy precipitation in the summer. Finally the CPMs underestimate the number of the events when precipitation and wind are treated singularly, while they substantially overestimate the number of compound events of rainfall and winds. The analysis shows the capability of CPMs to represent the precipitation and wind fields and highlights the possibility of using high-resolution reanalysis into the evaluation of convection-permitting models. Moving from point-based measurements to high-resolution gridded observational datasets opens the path to the use of SPHERA for advanced bias correction methods that could take into account the full 3D dimension of the atmosphere and the processes within it.