Improving Radiance Data Assimilation over Land through Surface Emissivity Study

Over-land data assimilation, particularly in the microwave regime, has been hampered by poor representation of the dynamic variability of land-surface emissivity within operational data assimilation frameworks, often resulting in the rejection of many satellite observations due to the large differences between simulations and observations. This research seeks to extend and improve efforts to assimilate satellite radiance data over non-ocean surfaces through the study of the surface emissivity simulation in the NCEP operational data assimilation system, the Gridpoint Statistical Interpolation (GSI) system.

In this work, the default GSI emissivity, calculated using the Community Radiative Transfer Model (CRTM), was compared with alternative emissivity products, e.g., a Tool to Estimate Land Surface Emissivities (TELSEM), TELSEM-2, and tested for the Advanced Technology Microwave Sounder (ATMS) radiance assimilation. In addition, an emissivity control variable was added to GSI to improve the emissivity background, and, therefore, the simulated radiance. It was found during that in clear sky runs over land, the updated system has issues with bias correction. Still, positive improvement is seen in all sky simulations. Currently, the bias correction for radiance data over land is being investigated and tested for this emissivity study. This paper will present the impacts of these updates on all surface radiance data assimilation, which includes using an alternative emissivity background as well as the control variable approach. The assessment will include statistics on observation counts, analysis statistics (O-B, O-A), and preliminary forecast impacts. This paper will also discuss other improvements related to all surface assimilation including updates to the bias correction scheme.