HR: 1330h
AN: H32B-0569    [PDF]
TI: Spatial-Spectral EOF analysis of AIRS data: an exploratory study
AU: * Huang, X
EM: hxl@gps.caltech.edu
AF: Division of Geological and Planetary Sciences, California Institute of Technology, Mail Stop 150-21, Pasadena, CA 91125 United States
AU: Yung, Y L
EM: yly@gps.caltech.edu
AF: Division of Geological and Planetary Sciences, California Institute of Technology, Mail Stop 150-21, Pasadena, CA 91125 United States
AU: Lambrigtsen, B H
EM: bhl@airs.jpl.nasa.gov
AF: Earth and Space Sciences Division, Jet Propulsion Laboratory, California Institute of Technology, 4800 Oak Grove Drive, Mail Stop 169-237, Pasadena, CA 91109 United States
AB: We apply spatial-spectral EOF analysis to 14 days of AIRS (Atmospheric Infrared Sounder) calibrated radiance data collected over the tropics and subtropics (32S to 32N) from July 1 to July 14, 2003. We limit our analysis to the nadir-view (scan angle less than 5$^{o}$) spectra only. After the quality control procedure, we have an average of 1400 spectra over this period for each 4$^{o}$ by 5$^{o}$ grid box. We obtain a 14-day averaged spectrum for each grid box and apply EOF analysis to these averaged spectra to obtain principal components in spectrally resolved radiance and associated spatial patterns. The first principal component (PC1) can explain more than 90% of the total variance. With the second principal component (PC2), these two leading principal components can explain more than 99% of the total variance. The PC1 spectral features are consistent with spectral features due to the change of surface (or cloud deck) emission temperature. A couple of features can be clearly seen in the PC1 spatial map: ITCZ due to the low emission temperature of optically-thick high cloud, Sahara due to the high surface emission temperature and the clear sky. The spatial map of PC1 closely resembles that of NCEP/NCAR reanalysis of outgoing longwave radiation over the same period. It is also highly correlated with the map of high cloud amount. Both the spectral features and spatial map indicate that the PC1 is mainly due to the spatial variation of cloud emission temperature (for grid boxes with the optically thick clouds) and surface temperature. The PC2 shows spectral features similar to those due to the change of the optical depth of low clouds. Moreover, the PC2 spatial map shows maxima near the coasts of Peru, Namibia and California, as well as over the southern ocean west of Australia. All these regions are known for high frequency of marine stratus. Unlike the traditional approach of observing low clouds from the visible reflectance, these results indicate that we can actually see variations related to low clouds from AIRS infrared data. The PC2 spectral shape also shows contributions from the stratosphere. This demonstrates that AIRS data, with its good quality and dense sampling patterns, can be a useful dataset for future validation and development of general circulation models.
DE: 0320 Cloud physics and chemistry
DE: 3309 Climatology (1620)
DE: 3359 Radiative processes
DE: 3360 Remote sensing
DE: 3364 Synoptic-scale meteorology
SC: Hydrology [H]
MN: 2003 Fall Meeting