Enhanced Radiance Bias Correction in the NCEP ' s Gridpoint Statistical Interpolation Data Assimilation System

The enhancements to the operational radiance bias correction in the NCEP's Gridpoint Statistical Interpolation (GSI) Data Assimilation System have been developed and tested on the development versions of the parallel Global Forecast System (GFS). Enhancement efforts have focused on several aspects of radiance bias correction. First, a modified preconditioning is applied to the bias correction coefficients and the analysis variables to speed up the convergence of the minimization process. Second, the capabilities of detecting any new/missing/recovery radiance data and initializing the bias correction for new radiance data are implemented. The background error variances for the bias correction coefficients are automatically adjusted using an approximation of the analysis error variances from the previous cycle, and the pre-specified predictor parameters are removed. Third, the capability to perform bias correction for passive channels within the GSI is developed. Finally, another major endeavor is that the two-step bias correction procedure currently used is replaced with a one-step variational bias correction scheme within the GSI. The scan angle and the air-mass bias corrections are now obtained consistently along with other analysis variables inside the GSI in the one-step procedure, and it is expected that they could respond quickly to any changes in data quality or algorithm accordingly.