Application and evaluation of the WRF model for high-resolution forecasting of rainfall - a case study of SW Poland

The Weather Research and Forecasting (WRF) model is applied to provide quantitative precipitation forecasts at 10 km · 10 km and 2 km · 2 km spatial and one hour temporal resolution for the area of SW Poland. The forecasts are evaluated by comparing the WRF model precipitation with measurements gathered at a meteorological station operated by the University of Wrocław and 17 SYNOP (surface synoptic observations) sites for the period 03.03.2012–18.06.2012. The 2 km · 2 km domain is run with the KainFritsch parameterization convection, and, as a separate simulation, with deep convection explicitly resolved. The results show that the model is capable of reproducing the number of observed precipitation episodes, but the performance decreases with the forecast range and rainfall intensity. The Kain-Fritsch model runs show a significantly higher area covered with rainfall when compared to the simulations with deep convection explicitly resolved, and are biased high for both 2 km and 10 km domains. The model runs with convection explicitly resolved show higher values of Success Ratio, while the Kain-Fritsch based runs, both for 10 km and 2 km, have higher Probability of Detection. None of the tested model configurations was able to resolve a highly local episode of intensive rainfall observed in the vicinity of Wrocław on 03.05.2012.