Assimilation of Land Surface Skin Temperature Observations Derived from GOES Radiances

The assimilation of surface sensitive radiances (SSR) over land is identified as a promising avenue to better exploit the rich information content of satellite observations. In recent work, Dutta et al. (JAMC, 2016) evaluated the impact of global assimilation of SSR observations from infrared sounders.  Even if surface skin temperature (T_s) is part of the model state, T_s increments were ignored because the atmospheric and surface analyses are done separately.  To reduce that inconsistency, we are now evaluating the impact of assimilating T_s observations in the surface analysis.  The Canadian Land Surface Analysis System (CALDAS) is used for that purpose.  We focus on the use of GOES T_s 1D-var retrievals (using 3 channels, more with upcoming GOES-R) in the regional analyses covering Canada, US and adjacent oceans. Consistency between the surface emissivity used in the forecast model (broadband) and that used in retrievals (spectrally dependent) was improved by using the same source for both (MODIS database).  The advantage of using geostationary observations is that these are available at high spatio-temporal resolutions along with improved cloud detection (in comparison to the same information obtained from infrared sounders).  First results from assimilation cycles will be presented at the conference.