HR: 1340h
AN: A13D-1498    [Abstracts]
TI: Experimental processing of spectral information from the Atmospheric Infrared Sounder: non-linear estimation of carbon dioxide spatial distribution in three dimensions
AU: * Plokhenko, Y
EM: yourip@ssec.wisc.edu
AF: Cooperative Institute for Meteorological Satellite Studies (CIMSS) University of Wisconsin-Madison, 1225 W. Dayton St., Madison, WI 53706, United States
AU: Menzel, P
EM: paulm@ssec.wisc.edu
AF: Cooperative Institute for Meteorological Satellite Studies (CIMSS) University of Wisconsin-Madison, 1225 W. Dayton St., Madison, WI 53706, United States
AU: Knuteson, R
EM: robert.knuteson@ssec.wisc.edu
AF: Cooperative Institute for Meteorological Satellite Studies (CIMSS) University of Wisconsin-Madison, 1225 W. Dayton St., Madison, WI 53706, United States
AU: Revercomb, H
EM: hankr@ssec.wisc.edu
AF: Cooperative Institute for Meteorological Satellite Studies (CIMSS) University of Wisconsin-Madison, 1225 W. Dayton St., Madison, WI 53706, United States
AB: CO2 horizontal field at different atmospheric levels has been derived from AIRS hyperspectral measurements. Results for granule 001 (0 GMT) over South Europe – North Africa region of Nov 16, 2002 are presented. Results were obtained using non-linear spatial-spectral analysis performed in four steps. At step 1, spectral data filtering, ‘bad' channels are identified (statistics of second spatial differential of spectral fields are used as predictors). At step 2, cloud identification, effective cloud amount is estimated. At step 3, spatial filtering, measurement noise over cloud free pixels are reduced using spatial smoothing. At step 4, physical interpretation, geophysical parameters at cloud free pixels are retrieved with a non-linear radiative transfer formulation. The radiative transfer model for a cloud free atmosphere includes spectral reflection at the lower boundary. A modified UMBC SARTA code is used for atmospheric spectral transmittance calculations. The physical parameters, included in the model, are: (1) surface emissivity spectrum (13 spectral parameters), (2) surface temperature, (3) atmospheric temperature vertical profile (35 vertical parameters), (4) atmospheric moisture vertical profile (22 vertical parameters), (5) atmospheric ozone vertical profile (17 vertical parameters) and (6) atmospheric CO2 vertical profile (16 vertical parameters). In all 104 variables are estimated with each spatial pixel. A solution is derived from minimization of the spatial integral of a weighted absolute difference (measurement – model) plus absolute values of spatial derivates of the atmospheric parameter estimates plus absolute values of the atmospheric parameter estimates. The spatial derivates of the atmospheric parameter estimates introduce a spatial filter to remove un-physical short wave spatial oscillations and to separate spectral effects from different atmospheric constituents using a priori information about their spatial variability scale. Spatial variability scale of CO2 vertical profile of ~250km. was used. The ECMWF forecast was used for solution initialization. First guess of vertically uniform CO2 profile of 365ppmv was assumed (value of ~370 ppmv is predicted by direct CO2 measurements in 2002). Measurements at cloud free areas (~7000 pixels) were interpreted. The average absolute measurement residual is within the range of 0.1-0.35K in used channels(~1900 channels) and in CO2 channels the statistic value of residual is 0.1-0.2K ( 184 channels: 101 channels in SW and 83 channels in LW). CO2 fields were derived for 3 combinations of measurement spectral bands: SW, LW and SW+LW (for the rest of the problem physical parameters measurement spectral content was fixed SW+LW). Results indicate a CO2 increment of +1-5ppmv. SW channels provided a CO2 signal for atmospheric layer of 250 -500mb, LW channels for 100 – 300mb. CO2 results of SW, LW and SW+LW experiments exhibit mutual spatial consistency. They show presence of spatially local maxima adjacent to cloud areas. Horizontal consistency, location and shape of these CO2 maxima between SW and LW bands results indicate on rather cloud particle spectral effects than on the CO2 spectral absorption. Cloud spectral effects cause a positive bias in CO2 estimates based upon IR measurements. That bias can not be removed by spatial-temporal averaging and will have spatial-temporal distribution pattern similar to the clouds pattern. Local CO2 maxima can be used to identify and remove clouds from the assimilation.
DE: 0365 Troposphere: composition and chemistry
SC: Atmospheric Sciences [A]
MN: 2007 Fall Meeting