Analyses of the relationship between atmospheric temperature structure and EOFs

The EOF (empirical orthogonal function) technique is usually used in thermal remote sensing in order to reduce the number of the parameters to be inversed. Few researchers put emphases on the physical meaning of the eigenvectors. Our study shows that different eigenvector corresponds with different atmospheric layer. It is necessary to study the physical meaning of the eigenvectors in remote sensing inversion, especially when we put emphases on one or some original unknown parameter(s), which is a common need in remote sensing inversion. First, one EOF choice scheme is proposed in our paper. It not only satisfies the need of whole explained variance but also considers the partial variance related with the parameter we are interested in. The inversion result shows that the scheme is useable to the special unknown parameters. At last, the physical meanings of EOFs have been discussed based on the large sample.