Comparação entre abordagem multi-física em mesoescala e determinística em alta resolução na simulação numérica de sistemas convectivos de mesoescala na Bacia do Prata

Abstract

This study has objectively-assessed the performance of numerical simulations of Mesoscale Convective Systems (MCSs) in the La Plata Basin (LPB) comparing two distinct approaches: deterministic simulations in the convective-allowing scale and a multi-physics ensemble in the mesoscale (ENS). Two episodes of MCSs that occurred in the LPB under distinct synoptic forcings (strong and weak) were selected. For each episode, numerical simulations utilizing the Weather Forecast and Research model (ARW-WRF) with two nested grids with horizontal resolutions of the 12 (G12) and 4km (G04), in "one-way" mode. For each SCM episode, eighteen simulations were performed in the G12 domain with each simulation utilizing a distinct combination of physical parameterization schemes, for which were selected three different convective schemes (EPC), three schemes for cloud microphysics (EMN), and two planetary boundary layer schemes (ECL). The eighteen G12 simulations formed the members of ENS, with each member also providing a respective G04 simulation that used the same corresponding options of EMN e ECL. To assess the skill of the simulations, 3-hourly surface data in METAR format and estimated precipitation from the Merged Microwave product generated by the Climate Prediciton Center (CPC/NOAA) were used. Traditional methods were employed to verify the simulation of surface variables, which were summarized with the Taylor diagram. For precipitation, a non-traditional objectoriented technique was utilized, as well as a hierarchical cluster analysis that allowed the ranking of the simulations as a function of the similarity of the rainfall patterns produced. Among the surface variables, the best model performance was obtained for the sea-level pressure, for both the G12 ENS and G04 deterministic runs. The latter ones displayed superior performance compared to G12 ENS for surface stations that were affected by convectively-induced mesoscale circulations. Larger spread was found for the simulations of the MCS under weaker synoptic forcing, with the surface wind being the variable exhibiting the highest spread. As for precipitation, the cluster analysis showed that the cloud microphysics and convective schemes diplayed the strongest influence upon the spatial and temporal distribution of the simualted rainfall. Overall, simulations with the Thompson EMN produced rainfall patterns and evolution with closest agreement with the counterpart from the estimated rainfall. The object-oriented forecast verification highlighted a clear distinction in the model performance as a function of the synoptic forcing, with better results for the strong synoptic forcing case. Results from this study showed that an approach combining the multi-physics ensemble in the mesoscale with deterministic simulations in the convective-allowing scale can be an interesting option for numerical weather prediction.