Sensitivity of WRF precipitation intensity forecasts to microphysical parametrizations

The aim of this work was to set up an optimal configuration of moist physics in the Weather Research and Forecasting (WRF) model by assessing the sensitivity of precipitation intensity forecasts over most of the Balkans, with particular reference to Bosnia and Herzegovina (BiH). This system is supposed to be an operational forecasting tool at the BiH entity, Republic Hydrometeorological Institute of Republic of Srpska. The study focuses on cases when precipitation resulted from the activity of a large cyclone in the Mediterranean centered over Sardinia and Sicily during the winter period. Simulations were conducted for five cases of precipitation that affected BiH, with precipitation also recorded at stations in Montenegro and Greece. The WRF model was nested into the Global Forecast System (GFS). Initial and boundary conditions were available at approximately 13 km horizontal resolution. Forecasts were produced on a 10 km horizontal resolution and a vertical grid with 45 levels, with the top at a 20 hPa level. The model start time for all cases was 00 UTC. Sensitivity of the model results to the microphysical schemes (Ferrier, Thompson, and Thompson scheme with aerosols) was tested. Forecasts were classified according to actual precipitation intensity, and skill scores were calculated for different precipitation thresholds. Quantitative evaluation of model results was performed based on recorded daily accumulations. The statistical metrics included root mean square error and bias as standard measures, and sensitivity, specificity, and true skill statistics, as binary measures. Forecasted synoptic fields of pressure, temperature, and geopotential were compared with fields from GFS analyses.