Improving boundary layer cloudiness in the NCEP-GFS

Abstract

Substantial biases exist in shortwave cloud radiative forcing in the National Centers for Environmental Prediction Global Forecasting System (GFS), an important global weather forecasting model. An offshoot of this model, the Coupled Forecast System, also shows major cloud biases. Much of the shortwave bias is attributable to problems with cloud fraction in subtropical boundary layer clouds. Simulations of shallow cumulus, stratocumulus, and stratocumulus-cumulus transition regions are performed with a single column version of the GFS. Numerical experiments with changes to parameterizations of shallow convection, cloud top-driven turbulence, and stratiform microphysics are tested. It is found that fairly simple changes considerably reduce biases in simulated precipitation, cloud fraction, and liquid water path. These changes shift the mechanism of moisture removal from the boundary layer away from precipitation and toward enhanced mixing with the free troposphere. Preliminary tests of these changes in the 3D GFS are promising, but more global model tests are needed to determine how general the improvements seen in the single column model experiments are.