Developmental Testbed Center's Evaluation of the FV3 Ensemble Member Performance During the 2018 Hazardous Weather Testbed—Spring Forecasting Experiment and Update on Future Support for the Stand-Alone Regional FV3

During the 2018 Hazardous Weather Testbed Spring Forecasting Experiment (HWT-SFE) several member configurations were contributed to the Community Leveraged Unified Ensemble (CLUE) dataset. Within this super ensemble, which has over 80 members in total, were several configurations employing the Next-Generation Global Prediction System (NGGPS) selection for the dynamic core – the Finite-Volume Cubed-Sphere (FV3). Variations in the microphysics, planetary boundary layer (PBL), and cumulus parameterizations (outside of the high-resolution nest) were configured to make up the 11 FV3 members. The careful coordination and construction of CLUE allows for meaningful comparisons among a variety of members to be performed. To compliment the subjective assessment performed daily during the experiment, extensive objective verification after the experiment allows for thorough investigation of the contributed model configuration or ensemble construction strengths and weaknesses.

Given the future unified forecast system will be based on the FV3 dynamic core, it is important to examine FV3 member performance within the HWT CLUE super-ensemble. In addition to assessing the probabilistic forecast performance of the FV3 ensemble, individual deterministic forecasts from the FV3 members will also be assessed to understand their contribution to ensemble spread. The objective evaluation will be conducted using the Model Evaluation Tools (MET) software system. The metrics used for probabilistic and deterministic evaluation will range from traditional metrics widely used in the community (spread, skill, error, reliability, etc.) to newer methods that provide additional diagnostic information such as the Method for Object-based Diagnostic Evaluation (MODE) and neighborhood methods applied to the output (e.g., Fractions Skill Score).

Additionally, an update on the DTC progress related to identifying the best approach for setting up and effectively supporting a community modeling infrastructure for the stand-alone regional FV3 currently under development will be provided.