Radiance bias correction in the National Centers for Environmental Prediction's Gridpoint Statistical Interpolation

Radiance bias correction is an important and necessary step in the proper use of satellite observations in a data assimilation system. The current radiance bias correction scheme used in the Gridpoint Statistical Interpolation (GSI) (Derber and Wu 1998) consists of two components: a variational air-mass dependent component and a scan angle component. The air-mass component is updated within the GSI while the scan angle component is updated outside of the GSI. Despite the effectiveness of the scheme, the split treatment of the two components of the correction, along with other details, leaves room for investigation and improvement.

This study examined and enhanced several aspects of the radiance bias correction problem. First, a new pre-conditioning was applied to the bias correction coefficients in order to allow the removal of pre-specified predictor parameters. Second, the background error variances for the bias correction coefficients were automatically adjusted using an approximation of the analysis error variances from previous cycles. Third, the capability to perform bias correction for passive channels within the GSI was developed.

The two step bias correction procedure currently used was replaced with a single variational bias correction scheme within the GSI. Numerous tests of the new scheme have been run. Due to the interaction between the radiance bias correction scheme and radiance quality control, different approaches of initializing the correction scheme have been tested and their analysis and forecast impacts assessed. Results from these tests will be presented at the conference.