Test and evaluation of CCPP physics suites and data assimilation algorithms to improve the Rapid Refresh Forecast System for convection forecasts

The Unified Forecast System application for regional and convective scales, or Rapid Refresh Forecast System (RRFS), is under development and aims to replace the operational suite of regional models in the next upgrade. In order to achieve skillful convection forecasts comparable to operational models, each component needs to be exhaustively tested and the best configuration determined. RRFS currently includes a FV3 Limited Area Model with a Common Community Physics Package (CCPP), Unified Post-Processing system, and data assimilation capability using the Gridpoint Statistical Interpolation (GSI) analysis system, providing a suitable research framework to assess its ability to represent convection.  In this study, various physics suites and data assimilation algorithms were assessed to improve the RRFS forecasts of a squall line over Oklahoma on May 4th, 2020. Numerical experiments were conducted running hourly cycles from May 4th 00z to May 5th 06z with 18-h forecasts launched at each cycle. Forecast verification was performed using the Model Evaluation Tools. Four CCPP physics suites were tested: two Global Forecast System (GFS)-based physics, a suite developed at NOAA’s Global Systems Laboratory, and a suite based on RAP/HRRR physics. Various analysis algorithms in GSI were evaluated, such as the three dimensional (3D) variational versus 3D Ensemble Variational hybrid data assimilation, different analysis grid ratios, supersaturation removal, and various ensemble background error covariance weights in the hybrid analysis. Observation impact experiments were conducted and the HRRR and GFS as cold start initial conditions were also evaluated. Results obtained in this research may inform RRFS developers on the performance of the system and will be presented during the seminar.