B3 - Sources of uncertainty for convective-scale predictability
Other researchers: Amirmahdi Zarboo (PostDoc)
Forecasting convective precipitation remains one of the key challenges in numerical weather prediction (NWP). The predictability of convective precipitation depends on many factors including the synoptic-scale flow, the geographical region (e. g. the presence of mountains), the underlying land surface, and microphysical uncertainties. The overall goal of this project is to identify sources of uncertainty for convective-scale predictability by investigating the mutual link between land surface and microphysical uncertainties, to better understand the mechanisms for convection initiation, cloud formation, and subsequent precipitation. Phase 1 addressed the different uncertainties in a univariate approach by perturbing these components independently in perturbed-parameter experiments using the COSMO model to study their relative influence on precipitation in different weather regimes. We found a decisive weather regime dependence of the perturbations, with greater impact during weak synoptic-scale forcing. The underlying terrain mainly influences convective precipitation by its effects on low-level convergence zones. Heterogeneous soil moisture perturbations lead to a spatial locking of convection at soil moisture gradients at scales of 40 to 80 km resulting in an earlier initiation of convection and a decreased day-to-day variability of area-averaged precipitation during weak synoptic forcing. In general, soil moisture and aerosols were found to have the largest impact on the precipitation amounts, whereas the location of precipitation is mainly controlled by terrain features and soil moisture heterogeneities through the triggering effects of convergence zones.
In Phase 2 we will extend the research of Phase 1 by addressing the more challenging question of how the collective impact of different uncertainties affects convective precipitation on different spatiotemporal scales for different weather regimes using a full ensemble forecasting system. The combination of different sensitivities will help answering the question if, and how, different processes compensate or enhance each other and if aerosol effects on clouds are modulated by soil moisture uncertainties (e. g. drier or wetter soils). We will additionally consider uncertainties in the parameters of the drop size distribution influencing a number of microphysical processes (e. g. evaporation, water vapor deposition, radiation, heating rates). Building on results of Phase 1 we carefully design numerical experiments using the new Icosahedral Nonhydrostatic (ICON) forecasting system at convective-scale. The use of the full ensemble forecasting system including data assimilation allows to examine how the structural model error of cloud microphysical uncertainties affects predictability and how it compares to the uncertainty introduced by the physically-based stochastic boundary layer perturbation scheme PSP. Finally, the role of orography in controlling land surface–atmosphere and aerosol–cloud interactions is addressed and the most important scales at which the terrain impacts precipitation predictability will be determined.