P38     Extended range severe weather forecasts using WRF and MPAS.

Sobash, Ryan, and Craig Schwartz, NCAR/MMM

Extended range forecasts of severe weather events were generated with WRF and MPAS convection-allowing ensembles. The 10-member, 132-hr, ensemble forecasts were initialized daily at 00 UTC between 23 April 2017 and 27 May 2017 with GFS ensemble (GEFS) members. Both WRF and MPAS had 3-km grid spacing over the CONUS, with MPAS having a 15-km mesh elsewhere on the globe, while the WRF forecasts used GEFS boundary conditions. Severe weather forecasts were produced using updraft helicity (UH), with evaluation focusing on daily (12 UTC – 12 UTC) and hourly probabilistic guidance for Days 1 – 5, verified using NWS storm reports.

Differences in UH computations in the MPAS and WRF ensembles resulted in differences in the UH model climatology, necessitating a bias-correction approach to produce equivalent guidance. The bias-corrected guidance from both WRF and MPAS exhibited positive skill over climatology for days 1 – 5, with skill decreasing with lead-time. The MPAS forecasts were more skillful than the WRF forecasts for Day 1, while the WRF forecasts were more skillful for days 2 – 5, with the skill gap increasing with lead-time. The diurnal cycle of severe weather reports was better captured in the WRF forecasts, especially beyond Day 2 and during the overnight periods (00 UTC - 12 UTC). Preliminary results suggest this may be due to deficiencies in the representation of large-scale overnight convective systems, such as MCSs, in the MPAS forecasts.