Sensitivity of quantitative precipitation forecast to soil moisture initialization and microphysics parametrization

In contrast to many other aspects of numerical weather prediction, quantitative precipitation forecasts did not gain much from a continuously improved physics over the past decades. A gap of knowledge remains as to how different processes contributing to the uncertainty in precipitation forecasts compare to each other. This research aims at gaining further insight into how two aspects influencing moist processes interrelate in atmospheric models. For this, high-resolution (3000 m) integrations with the non-hydrostatic Advanced Regional Prediction System were performed for two cases of extreme convection in Belgium, one driven by strong buoyancy and no vertical wind shear and the other driven by strong vertical wind shear and moderate buoyancy. Sensitivity experiments consisted of three integrations with different soil moisture initialization and three experiments with different size distribution assumptions in the microphysical parametrization scheme. The gain of spatially distributed soil moisture information was small in our simulations, but it was found important to have at least the mean soil moisture content right for the realistic simulation of cold-pool intensity. A mechanism, based on cold-pool and buoyancy interaction was proposed to explain the inverse relation between surface precipitation and soil moisture content. Changing the size distribution assumptions of the precipitating hydrometeors in the microphysics parametrization had a larger impact on the hydrometeor distribution than on surface precipitation. Weighting the size distribution assumptions of large hail to small graupel led to a poor representation of the general storm structure but did not affect the surface precipitation as much as was found in previous idealized studies.