Evaluation of FV3 Convection-Allowing Forecasts over CONUS using Different Physics Schemes during the 2018 Hazardous Weather Testbed Spring Experiment

With the selection of the Finite Volume Cubed-Sphere (FV3) dynamical core to serve as the basis for the Next-Generation Global Prediction System (NGGPS), work has begun to enhance and evaluate FV3 for potential use as a unified model for forecasting across all scales by the National Weather Service. As the essential step of implementing and evaluating optimized physics suites that are more suitable for convective-scale forecasting, the Center for Analysis and Prediction of Storms (CAPS) at the University of Oklahoma implemented a number of leading Planetary Boundary Layer (PBL) and microphysics (MP) schemes into FV3 and will run the model with different physics options during the 2018 Hazardous Weather Testbed (HWT) Spring Experimental Forecast Program (EFP), with a convection-permitting 3-km grid covering the continental US (CONUS) nested within a global FV3 grid. The implemented PBL schemes include YSU, Shin and Hong scale-aware YSU, MYJ and MYNN, and the implemented MP schemes include the Thompson, NSSL, Milbrandt and Yau, and Morrison partially or fully two-moment schemes. With combinations of these PBL and MP schemes plus the hybrid Eddy Diffusivity Mass Flux (EDMF) PBL scheme from the GFS system, 10 forecasts of up to 120 hours starting from the same radar-enhanced GFS analyses will be run daily during the HWT EFP, and forecast products will be made available in realtime to the HWT EFP participants for subject evaluations. The 3-km CONUS grid will be two-way nested within a 13 km global grid; no lateral boundary condition will be necessary.

The new Tiedtke cumulus scheme implemented by CAPS within FV3 will be used on the global grid, and its performance for global precipitation forecasts will be evaluated relative to comparable FV3 global forecasts using the Simplified Arakawa-Schubert scheme of GFS. The FV3 3-km forecasts will be quantitatively evaluated in a quasi-realtime setting, against Multi-Radar Multi-Sensor (MRMS) gridded precipitation and reflectivity data and other observations. Verification metrics to examine will include neighborhood-based and object-based forecast evaluation metrics available within the MET (Meteorological Evaluation Tools) and Unified Post-Processing Package (UPP), for both precipitation and severe weather forecasting. The performance of individual FV3 forecasts will also be compared with operational HRRR and HREF (High-Resolution Ensemble Forecast) members, as well as certain WRF-based members of CAPS’s Storm Scale Ensemble Forecasts, all of which have grid spacings of about 3 km. The findings will help guide us to choose the optimal suite of physics for FV3, and to determine FV3’s performance relative to current operational systems, and to guide further improvement to the FV3 model for convection-allowing model (CAM) applications.