Principal Component Analysis of IASI Spectra with a Focus on Non-Uniform Scene Effects on the ILS

Exploiting the redundancy of spectrally correlated information in high spectral resolution infrared radiance observations, Principal Component Analysis (PCA) is useful for extracting signals from the observations, noise filtering, and for characterization of variable artifacts within the data. Here, PCA is used to investigate the spectral nature of observations from the Infrared Atmospheric Sounding Interferometer (IASI) with a focus on the effects of non-uniform scenes on the Instrument Line Shape (ILS). For IASI and other similar FTS systems, non-uniform scenes produce a non-uniform weighting of angles through the interferometer that can produce subtle distortions in the ILS from that of uniform scenes. For the PCA approach used here, it is found that the spectral signature due to the non-uniform scene effects are contained primarily within a single principal component, and that reconstruction of the radiance spectra with this component excluded provides an accurate and robust correction algorithm. Further investigation shows that precomputed synthetic principal components representing the signatures of non-uniform scenes can be used in the correction, greatly reducing the computational expense of the algorithm.