Uncertainty in the area-related QPF for heavy convective precipitation

We studied five convective events producing heavy local rainfall with the help of the numerical weather prediction model LM COSMO. The model was run with a horizontal resolution of 2.8 km. We created an ensemble of 13 forecasts by modifying initial and boundary conditions. Uncertainties in the area-related quantitative precipitation forecast (QPF) were evaluated with the help of a fraction skill score that can quantify the ensemble spread and skill. The spread represents the differences between the control forecast and the forecasts provided by each ensemble member, and the skill evaluates the difference between the precipitation forecast and radar-based rainfall. Analyses show how the forecast lead time and spatial scale influence the spread and skill values. Despite the different areal structures of precipitation fields, the relationships between spread and skill appear to be similar.