GSI/WRF Regional Data Assimilation System and Its Application in theWeather Forecasts over Southwest Asia - Data Assimilation for Atmospheric, Oceanic and Hydrologic Applications - page 619

Fig. 25.12 Scan coverage of ATOVS (AMSU-A, AMSU-B) radiance being used in current data
assimilation system at 00Z during May 2006 (Updated from Xu et al. ( 2009 ))
the AMSU-A was under 60,000 pixels. On average for the 30 days, the evidence
shows through this two-step checking procedure, the amount of radiance data going
into the model is reduced substantially. The percent usage of AMSU-A radiance
data was over 40 %, but for AMSU-B it was only 16 %.
It is obvious that bias correction and quality control toss out non-useful data. This
is less taxing on the minimization procedure within variational data assimilation
systems However, because of the imperfections inherent in bias correction and
quality control schemes, a lot of valuable observations are tossed out. Future studies
should continue to refine good bias correction and quality control schemes.
For the control experiments described in Sect 25.2 (referred to as CTRL),
the initial conditions generated for the GFS forecasts were assimilated using
several different satellites, such as AMSU-A/B, High Resolution Infrared Radiation
Sounder (HIRS), Microwave Sounding Unit (MSU) and so on. For the purpose
of eliminating the effect of the radiance assimilation in the first guess field from
GFS global analysis data, we first generated a spin-up run for 6 h from 18Z on
previous day to 00Z on the forecast day in the data assimilation (referred to as
DA) experiments; then the AMSU-A, AMSU-B radiance data are assimilated in the