Synoptic Control of Heavy-Rain-Producing Convective Training Episodes

This study examines the degree to which the downscale cascade of information from synoptic scale motions constrains error growth in simulations of a particular organization of heavy-rain producing mesoscale convective systems (MCSs) known as training lines. Training is defined as the repetitive motion of storm cells with > 40 dBz composite radar reflectivity over a fixed geographical location for greater than 6 hours. Twenty-four cases of training convection over a 7-year period from 2000-2006 that produced extreme rainfall are simulated using a high-resolution convection-permitting configuration of the Weather Research and Forecasting model (WRF), with initial and lateral boundary conditions provided from three different reanalysis data sets, each with different spatial resolutions. In most cases the model simulations were able to reproduce qualitative aspects of observed storm structure, including subjectively classified MCS archetype and training convection, despite the absence of mesoscale features in initial and lateral boundary conditions (IBCs and LBCs) in at least two of the three reanalysis data sets. Furthermore, models were capable of predicting that a heavy precipitation event would occur in nearly every case. Increasing the horizontal resolution of the reanalysis data set used for ICs and LBCs did not result in measurable improvement in simulated precipitation placement skill relative to observations. A connection was found between the magnitude of synoptic scale uncertainty and the magnitude of drift between model simulations, suggesting that upscale error growth between our simulations is constrained by the quality of information cascading downscale from synoptic scales. No such connection was found between the performances of model forecasts against observations, suggesting a practical limitation of their operational predictive capacity with respect to these events. Additionally, there is evidence that the robustness of synoptic scale forcing influences the performance of model forecasts relative to one another and to observations.