Convection-Allowing Ensemble Forecasts Using the Model for Prediction across Scales

Convection-allowing ensemble (CAE) forecast skill has almost exclusively been assessed with limited-area models, requiring boundary conditions originating from external, coarse-resolution models. These boundaries can impact the interior domain solution, decrease forecast skill, and impact spread growth in ensembles.

To investigate the effect of removing boundary conditions on CAE skill, ensemble forecasts were produced using the Model for Prediction Across Scales (MPAS). MPAS uses an unstructured, variable resolution mesh to reduce issues with sharp boundary transitions associated with traditional limited-area models. Here, we investigate forecast skill from a set of 35 warm-season, 10-member, 132-hour MPAS ensemble forecasts using 3-km horizontal grid spacing over the CONUS, increasing to 15-km elsewhere on the globe.

This presentation will focus on precipitation and severe weather skill, as well as ensemble spread, between Days 1 – 5 to quantify the ability of MPAS to produce accurate and reliable extended-range severe weather predictions. Additionally, the MPAS ensemble forecasts will be compared to several limited-area convection-allowing ensembles to reveal the impact of the lateral boundary conditions.