Relative impact of aerosol, soil moisture, and orography perturbations on deep convection

Abstract

The predictability of deep moist convection depends on many factors, such as the synoptic-scale flow, the geographical region (i.e., the presence of mountains), and land surface–atmosphere as well as aerosol–cloud interactions. This study addresses all these factors by investigating the relative impact of orography, soil moisture, and aerosols on precipitation over Germany in different weather regimes. To this end, we conduct numerical sensitivity studies with the COnsortium for Small-sale MOdelling (COSMO) model at high spatial resolution (500 m grid spacing) for 6 days with weak and strong synoptic forcing. The numerical experiments consist of (i) successive smoothing of topographical features, (ii) systematic changes in the initial soil moisture fields (spatially homogeneous increase/decrease, horizontal uniform soil moisture, different realizations of dry/wet patches), and (iii) different assumptions about the ambient aerosol concentration (spatially homogeneous and heterogeneous fields). Our results show that the impact of these perturbations on precipitation is on average higher for weak than for strong synoptic forcing. Soil moisture and aerosols are each responsible for the maximum precipitation response for three of the cases, while the sensitivity to terrain forcing always shows the smallest spread. For the majority of the analyzed cases, the model produces a positive soil moisture–precipitation feedback when averaged over the entire model domain. Furthermore, the amount of soil moisture affects precipitation more strongly than its spatial distribution. The precipitation response to changes in the CCN concentration is more complex and case dependent. The smoothing of terrain shows weaker impacts on days with strong synoptic forcing because surface fluxes are less important and orographic ascent is still simulated reasonably well, despite missing fine-scale orographic features. We apply an object-based characterization to identify whether and how the perturbations affect the structure, location, timing, and intensity of precipitation. These diagnostics reveal that the structure component, comparing the size and shape of precipitating objects to the reference simulation, is on average highest in the soil moisture and aerosol simulations, often due to changes in the maximum precipitation amounts. This indicates that the dominant mechanisms for convection initiation remain but that precipitation amounts depend on the strength of the trigger mechanisms. Location and amplitude parameters both vary over a much smaller range. Still, the temporal evolution of the amplitude component correlates well with the rain rate. Our results suggest that for quantitative precipitation forecasting, both aerosols and soil moisture are of similar importance and that their inclusion in convective-scale ensemble forecasting containing classical sources of uncertainty should be assessed in the future.