Assimilation of the radar-derived water vapour mixing ratio into the LM COSMO model with a high horizontal resolution

Abstract

We use a WVC (Water Vapour Correction) method to assimilate radar reflectivity into the NWP (Numerical Weather Prediction) Local Model COSMO (LM; version 3.18) with a horizontal resolution of 2.8 km. The WVC method takes into account differences between the model and radar-derived precipitation by modifying vertical profiles of the water vapour mixing ratio at each model time step using the nudging approach. We describe the application of the WVC method and apply it to five severe convective events. The LM contains an explicit formulation of cloud and rain processes, in which microphysics parameterization includes rain water, snow, ice, and graupels. We evaluate the WVC method performance and compare it with the latent heat nudging method (LHN), which is a part of the LM code. The evaluation is focused on a very short range forecast of precipitation caused by particular storms. Results show that in most studied cases, both methods can forecast precipitation development three hours ahead, which is approximately the life cycle length of studied storms, and that the assimilation significantly improves precipitation forecasts obtained by the NWP model. Three approaches are used to evaluate and compare the methods. The first evaluates the forecast “by eye”, the second is based on similarity measure (SRMSE) between precipitation cores, which takes into account area distribution of precipitation, and the third compares values of forecasted maximum precipitation in given areas. The results show that WVC yields better results than LHN in some cases and could be a good alternative to LHN. The relatively good agreement between the verification “by eye” and SRMSE confirms that SRMSE can be employed in similar studies.