The degree to which these potentials are being realized is being evaluated for the RAP model and the subsequent High-Resolution Rapid Refresh (HRRR) model (initialized from the RAP), through a series of retrospective experiments. We are examining data usage improvements and impacts on the bias correction and overall forecast skill from the direct readout feeds of ATMS and CrIS datasets obtained from Univ. of Wisc. / SSEC. Our initial experiments over a 7-day period with ATMS data showed only a small forecast impact, despite encouraging bias correction results. We are further examining the details of this experiments as well as documenting the specific degree of data usage improvement and examining other aspects such as channel selection associated with the relatively low RAP model top (10 hPa). We are also conducting further ATMS experiments and examining impacts for CrIS data. In this presentation, we will summarize the research results, examining data coverage, bias correction, channel selection, and forecast improvement.