An examination of the bias in the NCEP GFS, CFS simulations associated with the marine stratus clouds

A preliminary investigation has been conducted to examine the seasonal variations of precipitation and associated oceanic condition and atmospheric circulations over the Atlantic and eastern Pacific in the NCEP CFS AMIP and CMIP simulations. In the CFS AMIP simulation, the atmospheric component of the CFS model (GFS03) is forced by observed oceanic condition, while the same atmospheric model is fully coupled with an OGCM (MOM3) in the CFS CMIP simulation. Comparisons between the AMIP and CMIP simulations are able to provide us with insights into how imperfect definition of oceanic condition may influence the CFS model forecasts. Our initial results showed the following:

1) While large-scale precipitation patterns are reproduced reasonably well over the target regions in both the CFS AMIP and CMIP runs, differences exist in the magnitude of precipitation and in the latitudinal position of the ITCZ over both the Atlantic and the eastern Pacific sectors;

2) The latitudinal displacement of the ITCZ in the CFS CMIP run is closely related to the warm SST bias in the SE Atlantic and SE Pacific stratus deck regions;

3) The warm SST bias is primarily a result of the insufficient amount of stratus clouds simulated by the CFS model (and most other climate models as well); and

4) The stratus clouds over the regions have very low cloud top and present strong diurnal cycle generated by regional circulation caused by land-sea contrasts between the oceanic regions and their adjacent continents.

Further work is underway to examine the relative importance of the convection over the land areas and the SST in the ocean underneath the marine stratus clouds. Results will be reported at the workshop.