Large impact of tiny model domain shifts for the Pentecost 2014 mesoscale convective system over Germany

Abstract

The mesoscale convective system (MCS) that affected Germany at Pentecost 2014 (9 June 2014) was one of the most severe for decades. However, the predictability of this system was very low as the operational deterministic and ensemble prediction systems completely failed to predict the event with more than a 12 h lead time. We present hindcasts of the event using the COnsortium for Small-scale MOdeling (COSMO) model at a convection-permitting (2.8 km) resolution on a large (1668 km×1807 km) domain. Using this large domain allowed us to successfully simulate the whole life cycle of the system originating from the French Atlantic coast. However, even with the large domain, the predictability of the MCS is low. Tiny changes to the model domain produced large changes in the MCS, removing it completely from some simulations. To demonstrate this we systematically shifted the model domain by just one grid point in eight different directions, from which three did not simulate any convection over Germany. Our analysis shows that there were no important differences in domain-averaged initial conditions or in the preconvective environment ahead of the convective system. The main reason that one-third of these seemingly identical initial conditions fail to produce any convection over Germany seems to be the proximity of the track of the initial convective system to the coast and colder sea surface. The COSMO model simulates small horizontal displacements of the precursors of the MCS which then determine if the cells dissipate close to the sea or reach a favorable area for convective development over land and further evolve into an MCS. This study demonstrates the potentially huge impact of tiny model domain shifts on forecasting convective processes in this case, which suggests that the sensitivity to similarly small initial-condition perturbations could be a helpful indicator of days with low predictability and should be evaluated across other cases, models, and weather regimes.