GFS 3km Cloud Resolving Simulations for Hurricane Forecasts

Many research and operational weather centers in the world (such as ECMWF, CSU-NCAR, German-ICON, Japan NICAM, etc.) are developing km-scale convection-permitting global Numerical Weather Prediction (NWP) models. In the near future, the NOAA Unified Forecast System global weather prediction model will reach the km-scale resolution, which will be EMC's model development goal in coming years. This abstract will highlight the current development for NCEP GFS 3km cloud resolving simulations for hurricane forecasts.

A fully coupled global model is developed with ocean-model (MOM6) and ice-model (CICE6). The atmospheric model is the FV3 Dycore (C3072/3km) with 127 vertical layers up to the mesopause. The CCPP Physics includes RAS deep convection, sa-SAS shallow convection, GFDL Microphysics, NOAH-MP land. Two forecasts are conducted for hurricane Florence (2018) with the same initial condition at 00Z of 20180901. One forecast is with the deep/shallow convections turned on, the other one is with the convections turned off. The experiment with the deep/shallow convections turned on can predict a stronger/deeper hurricane in the 10-days forecast period. The comparison of two forecasts indicates that the deep/shallow convections help to moisturize the environment of the hurricane, which strength the hurricane Florence (2018). The results from the two experiments show that the deep/shallow convections are needed even in the global 3km fully coupled global in order to predict a better hurricane intensity.