Extended Abstract (148K)P3.14
Validation of global simulations of microwave and infrared brightness temperatures using AMSR and AIRS data
Thomas Greenwald, Univ. of Wisconsin, Madison, WI; and R. Bennartz, C. O'Dell, and A. Heidinger
Satellites are the leading source of data assimilated into numerical weather prediction models. At NCEP, for example, 99% of these data come from satellites. One assimilation approach, which has been gaining acceptance, is the direct use of radiances. The principal advantage of this approach is that it allows for better control of errors in the assimilation process.
With support from the Joint Center for Satellite Data Assimilation, this study evaluates forward computed 2D fields of microwave and infrared brightness temperatures from a radiative transfer modeling system in conjunction with forecasted global model fields (an essential step in a direct radiance assimilation system) by comparing them against satellite measurements. The comparisons are made in both clear and cloudy conditions, with an emphasis on the cloudy case, but limited to the oceans. NCEP’s Global Forecast Model (GFS) fields of temperature, humidity, cloud water, and precipitation are used as input to the radiative transfer modeling system. Retrospective 12-hour forecast fields (not model analyses) are utilized. A new radiative transfer model will be used to compute brightness temperatures at thermal wavelengths, called the successive order of interaction (SOI) model. It explicitly accounts for multiple scattering and in two/four-stream modes is faster than and as accurate as the well-known delta-Eddington model. Gas absorption is computed from the Optical Path TRANsmittance (OPTRAN) band model. A lookup-table approach based on pre-computed values from Lorenz-Mie Theory is used for microwave precipitation particle scattering properties, while anomalous diffraction theory is used to compute particle scattering properties at infrared wavelengths. Satellite validation includes window frequency data from the AMSU and AMSR and primarily window channel data from AIRS. Gridded AVHRR-derived and MODIS-derived cloud products will aid in identifying cloud conditions for these sensors.
Early results for AMSU frequencies indicate good agreement between the simulations and measurements in cloudy regions, however, large differences (up to 5 K) occur in clear regions most likely due to systematic errors in the microwave surface emissivity model. More complete validation using AMSU data and comparisons to AMSR and AIRS data will be reported at the conference.
Poster Session 3, Data Assimilation
Tuesday, 21 September 2004, 9:30 AM-11:00 AM
Previous paper Next paper