Verification of high-resolution precipitation forecasts over Germany

The verification of precipitation forecasts from numerical models is essential for the future improvement of quantitative precipitation forecasts (QPF). In this study the "Local-Model" (LM) from the German Weather Service is verified over Germany and therefore a very dense network of about 4000 rain gauge stations with daily rain sum measurements and hourly composits from 16 radars are used. The LM, operational since December 1999, is a non-hydrostatic grid point model with a resolution of 7 km and 40 vertical layers based on the fully compressible dynamic equations. Each day two simulations are started at 00 and 12 UTC with a forecast duration of 48 hours. The advent of such high-resolution mesoscale model predictions (time scale 1 hour) prompts to design novel verification techniques which require high-resolution observational data sets on the time scale of one hour. This study introduces such a data set for Germany, based upon 24-hour accumulated observations from the rain gauge network, a 30-year climatology of precipitation from the German Weather Service, and hourly Radar composites. The Radar data and a disaggregation technique, developed earlier within the Mesoscale Alpine Project, are used to introduce the high temporal resolution into the gridded rain-gauge analysis. The temporal variability of the resulting data set is consistent with the Radar data, whereas the daily total of the hourly fields corresponds to the rain gauge measurements. This first hourly data set for precipitation over Germany was compared with the few existing hourly measurements. Standard verification scores were used to verify LM precipitation over Germany for accumulation time periods of 1, 6 and 24 hours. The results show seasonal differences, geographical differences and a strong sensitivity to the accumulation time period. Also shown are preliminary results with novel verification scores. That are grid point independent and focus on the structural characteristics of the precipitation field within river catchments.