Impact of Stochastic Physics in FV3-GEFS—Preliminary Results

The next version of the NCEP global ensemble (GEFS) will include several major changes to the underlying model and uncertainty estimation, two of which we will focus on here. First, the old global spectral model dynamical core is being replaced by the finite-volume dynamical core (FV3), as determined by the Next-Generation Global Prediction System selection process. The FV3 core will also move into operations for the deterministic global model (GFS) earlier the same year. Second, the current model of model uncertainty, stochastic total tendency perturbation (STTP) will be replaced by a suite of ‘stochastic physics’ schemes: stochastically perturbed physics tendencies (SPPT), stochastic kinetic energy backscatter (SKEB), and stochastic humidity perturbation (SHUM). These schemes have been extensively tested in the GSM, but thorough testing of the stochastic physics package at full length GEFS (16 days) and extended GEFS (35 days) in the FV3 dycore remains to be completed. Here we will compare the performance of FV3-GEFS with the stochastic physics suite over a short experimental period against an FV3 ensemble without modeled model uncertainty, as well as the GSM-GEFS using stochastic physics.