NESDIS Progress in the OSSE

Thomas J. Kleespies 9 January 1998

The initial stage of the satellite data OSSE is to set up the infrastructure and simulate an existing observing system. The TIROS Operational Vertical Sounder (TOVS) system was selected for this purpose. The TOVS has flown on the NOAA polar orbiting spacecraft series since 1978. It consists of a 19 channel High Resolution InfraRed Sounder (HIRS), a three channel Stratospheric Sounding Unit (SSU), and a four channel Microwave Sounding Unit (MSU). The TOVS product file includes the cloud cleared, limb adjusted radiances as well as the retrieved temperature and moisture profiles. The product file for NOAA 11 and 12 from February and March 1993 was acquired from the National Climate Data Center. Unfortunately, it was discovered that the data from the first seven days of February was missing. Nevertheless, a subsampled data set for the missing period was acquired from a different source. This subset has approximately one quarter the data density of the full set. A substantial software infrastructure was developed to unpack the and organize the data, and considerable effort was made in characterizing the data.

The radiative transfer code used in this study is OPTRAN/RTATOV. OPTRAN (Optical Path Transmittance) is a gaseous transmittance model described in McMillin et al (1995). RTATOV (Radiative Transfer Advanced TOVS) is a radiative transfer code described by Eyre (1991). OPTRAN/RTATOV is RTATOV radiative transfer shell with OPTRAN transmittance core. Transmittance coefficients were generated for NOAA 11 AND 12 from line-by line data generated by Hal Woolf of the Cooperative Institute for Meteorological Satellite Studies at the University of Wisconsin using FASCD3P (Smith et al 1978).

The radiances computed from the 6 hour forecast nature run were compared with the TOVS observed clear radiances within +- 6 hours of the forecast time. The bias and the standard deviation of the difference are given in figure 1. Also given are the corresponding figures for the operational NCEP GDAS system, which are from some preliminary tests performed by John Derber. The nature run biases are in general similar to the NCEP biases. Notable exceptions are the high peaking channels 2, 3, 23 and 24, which differ because the nature run top level is about 10 mb, while the GDAS top is much higher. Also different are the shortwave channels 18 and 19, which for the nature run comparison include scattered sunlight, but include only nightime data in the NCEP results. The standard deviations of the differences have very similar patterns between the two system. The differences can be attributed to the different analysis systems, the different time periods, and the different satellite instruments.

Figure 1: Bias and Standard deviation of computed minus observed TOVS radiances. Nature run is a 6 hr ECMWF forecast valid 06 UTC 5 Feb 93 compared with NOAA 11. NCEP run is anumber of 6 hour forecasts versus NOAA 14 observations. Both sets have essentially no quality control.

Ref:
Eyre, J. R. 1991: A fast radiative transfer model for satellite sounding systems. ECMWF TM 176.

McMillin, L. M., L. J. Crone, 1995: Atmospheric transmittance of an absorbing gas. 5. Improvements to the OPTRAN approach. Appl. Opt. 34(35), pp8396-8399.

Smith,H.J.P. et al, 1978. FASCODE - Fast Atmospheric Signature Code. AFGL-TR-0081. Air Force Geophysics Laboratory, Hanscom AFB MA 01731.

Any question Mail to: