Subseasonal Skill Sensitivity in the Global FIM-HYCOM Coupled Model

A coupled atmosphere (FIM) and ocean (HYCOM) modeling system FIM-HYCOM is being developed for subseasonal to seasonal prediction. Both component models are 3-dimensional grid point models, operating on a common icosahedral horizontal grid, and using an adaptive near-isentropic vertical coordinates. FIM uses the 2015 Global Forecast System (GFS) physics package, with an option of switching its default Simplified Arakawa-Schubert convective cloud parameterization to a variant of the Grell and Freitas (2014) scheme (SAS and GF hereafter). Once-per-week subseasonal hindcasts with 4 time-lagged ensemble members over a 16-year period have been carried out with this coupled model using SAS and GF, respectively, at 60km horizontal resolution. Related experiments have also been carried out with the uncoupled atmospheric model using prescribed sea surface temperature, as well as with the coupled model at 30km horizontal resolution. Preliminary results from these multi-year hindcasts indicate that forecast skill of week-3-4 from FIM-HYCOM is comparable to that of the operational model CFSv2 used by NCEP. The fact that the coupled model has higher skill than the uncoupled model shows the importance of air-sea interaction in week-3-4 prediction. We will also discuss in detail the model's sensitivity to grid resolution (60km vs. 30km) and to physics parameterizations.