A Study On The Retrieval Of Atmospheric Parameters By Using The Satellite-based Infrared Hyperspectral Measurements

AIRS is a primary instrument on the Aqua polar orbiting satellite, which was launched on May 4,2002 as one of the Earth Observing System (EOS) platform. The spectral range is from 650 to 2700 cm^-1 with 2378 spectral IR channels, spectral response functions with full widths at half maximum of～v/1200. The ability of global coverage makes it to be able to observe the atmosphere and its variation on a global basis, which has truly realized the high spectrum high accuracy infrared detection.The dataset for global clear-sky atmospheric profiles from CIMSS (Cooperative Institute for Meteorological Satellite Studies, University of Wisconsin-Madison) is used here, half part as training samples, the other for retrieving and testing. Furthermore, AIRS simulating data derived from SARTA (Stand-Alone Radiative Transfer Algorithm) are used as regression factors to retrieve the atmosphere profile of temperature , ozone and other variables on a clear sky basis by means of eigenvector regression algorithms. The results are compared with observations to evaluate the retrieve.In order to improve the retrievals of the atmospheric temperature profiles, the number of eigenvectors is increased firstly, then, surface temperature and IR surface reflection index are introduced as restraint factors for retrieving equation. Analysis shows that:1. Eigenvector regression algorithm is a useful way to retrieve the atmospheric profile of temperature with speediness. The error of temperature profile is 2～3 K in the lower troposphere and 1～2 K in middle and upper levels. The maximum of the error of ozone profile is 0.85 ppmv at 10hPa.2. The retrieval precision of temperature is proved when eigenvectors is increased. The purpose is best as the number of eigenvectors is 80. But overabundant eigenvectors couldn't prove the precision much better. 3. Temperature profile retrieval is sensitive to characteristics of the earth's surface. The retrieval precision of temperature in low level is evidently proved when the surface temperature is added, while that in 50～200 hPa is ameliorated when IR surface reflection is added. If they are added at the same time, the precisions in both low and upper levels are improved.