Toward All‐Surface Assimilation of Microwave Sounding Data in the CMA‐GFS 4D‐Var System

The CMA's NWP system, CMA‐GFS, has achieved assimilation of surface‐sensitive channels on AMSU‐A over land, but the assimilation of overland data from low‐layer humidity‐sounding channels, overland Fengyun satellite data, and microwave data over other complex underlying surfaces are yet to be fully realized. In view of this, an all‐surface assimilation framework based on dynamic emissivity retrieval is established for radiance data over four main complex underlying surfaces (snow‐free land, snow, sea ice, and coast) from all microwave lower‐atmosphere‐sounding channels (at 50–60 GHz, 118 GHz, and 183 GHz) on all operational microwave instruments in CMA‐GFS, on the basis of summarizing previous research. A new emissivity correction scheme is proposed and the regression coefficients are determined for the assimilation of humidity‐sounding channels on ATMS over sea ice. Quality control procedures are designed for each instrument over each surface type. The performances of these modules are evaluated according to first‐guess departures, and the results are given for channel selection over each surface type. Five series of batch experiments, which add the observations over one surface type on top of the former experiment one by one, are carried out. Results indicate that all‐surface assimilation has significant positive impacts on the analysis of all variables across the entire atmospheric layer globally, and the 1–6 days forecasting in the Northern Hemisphere, 6–8 days forecasting in the Southern Hemisphere, and short‐to‐medium‐term forecasting in the tropics are all well improved. This work marks an important step toward the all‐surface assimilation of all the microwave atmospheric sounding channels in CMA‐GFS.