Reducing Forecasting Errors of Near Surface Fields in the NCEP Global Forecast System

Accurate forecast of near-surface fields in numerical models is regarded as a key ingredient for improving numerical weather and climate prediction, and also a challenging task owing to the multiplicity of the related physical processes and their complex interactions. This study addresses recently identified systematic errors in forecasts for near-surface-fields in the current operational NCEP Finite Volume Cubed Sphere (FV3) dynamic core based GFS (GFSv15). Different types of biases and systematic errors are reflected in near-surface forecasts. They can be attributed to various factors such as the land surface model, planetary boundary scheme, and other physics processes as well as the coupling between the surface and atmosphere. In this study, several different practical approaches have been tested in attempt to reduce the errors. A comprehensive set of daily 7-day forecast experiments spanning more than one month in various seasons demonstrate that using the proposed approaches can effectively reduce systematic deficiencies and substantial errors in the near-surface forecasts, along with a notable reduction of temperature errors throughout the lower atmosphere and improvement of forecast skill scores. The methods used, results obtained, and broader implications will be discussed in the presentation.