AbstractThe present investigation evaluates the assimilation of synthetic data which has properties similar to actual Global‐scale Observations of the Limb and Disk (GOLD) level‐2 (L2) temperatures and other conventional lower atmospheric observations. The lower atmospheric and GOLD L2 temperature (Tdisk) are assimilated in the Whole Atmosphere Community Climate Model with thermosphere‐ionosphere eXtension using Data Assimilation Research Testbed. It is found that inclusion of the GOLD Tdisk improves the forecast root mean square error (RMSE) and bias by 5% and 71%. When compared to lower atmosphere only assimilation, the improvements in RMSE and bias are 20% and 94%. An investigation of the global diurnal westward‐propagating wavenumber 1 (DW1) and local diurnal tidal characteristics shows that inclusion of the GOLD temperatures improves the DW1 by about 8% and diurnal tide by more than 17%. Larger percentage improvements in the tides are seen in the lower thermosphere. Considerable improvements in the model state are also seen at times and locations where there are no GOLD observations available. These results and the background data assimilation procedure are presented here, which demonstrate that GOLD thermospheric temperature is an excellent data set that can be used for thermospheric assimilation studies and operational purposes.