Comparison of cloud-resolving ensemble simulations using LM and MC2 simulations

Accurate quantitative precipitation forecasting is a difficult challenge due to the chaotic nature of the atmospheric dynamics and due to the use of imperfect NWP models. In this study, cloud-resolving ensemble simulations of MAP events (IOP 2b and IOP 3) are performed with two different limited-area models, the Lokal-Modell (LM) and the Mesoscale Compressible Community Model (MC2). Both models are used in a similar experimental setup with an explicit treatment of convection and an EPS strategy based on ensemble members that are generated by shifting the initialization time.

For the analysis, the focus is set not only on the validation and comparison of the simulated precipitation patterns, but also on the evolution of ensemble spread and forecasting skill throughout the forecasting period. Results demonstrate that the two ensembles yield a comparable assessment of the predictability, despite considerable differences in the simulated small-scale features. In particular, both models pinpoint IOP 2b (19-20.09.1999) as the case with a substantially higher predictability than IOP 3 (24-25.09.1999). Some hypotheses regarding the surprisingly different levels of predictability are also provided.