Channel selection method for hyperspectral 1 atmospheric infrared sounder using AIRS data 2 based on layering 3

Because a satellite channel’s ability to resolve 16 hyperspectral data varies with height, an improved channel selection 17 method is proposed based on information content. An effective 18 channel selection scheme for a hyperspectral atmospheric infrared 19 sounder using AIRS data based on layering is proposed. The results 20 are as follows: (1) Using the improved method, the atmospheric 21 retrievable index is more stable, the value reaching 0.54. The 22

methods, the use of which allows sensitive channels to be selected. 149 The above-mentioned studies also take into account the sensitivity of 150 each channel to atmospheric parameters during channel selection, 151 while ignoring factors that impact retrieval results. The accuracy of 152 retrieval results depends not only on the channel weight function but 153 also on the channel noise, background field, and the retrieval 154 algorithm. 155 Currently, information content is often employed in channel 156 selection. During retrieval, this method delivers the largest amount 157 of information for the selected channel combination (Rodgers, 1996; ) .

245
Define R =  sense. In addition to its high computational efficiency by using this 276 method, another advantage is that all channels can be recorded in the 277 order in which they are selected. In the actual application, if n′ 278 channels are needed, and n ′ < n, we will not need to select the 279 channel again, but record the selected channel only. one of the highest precision methods (Chedin et al., 1985). Therefore, 299 the statistical inversion method will be used for our channel 300 selection experiment and a regression equation will be established.

301
According to an empirical orthogonal function, the atmospheric where * and * are the eigenvectors of the covariance matrix of 310 temperature (or humidity) and brightness temperature, respectively.

311
A and B stand for the corresponding expansion coefficient vectors of 312 temperature (humidity) and brightness temperature.

313
Using the least squares method and the orthogonal property, the 314 coefficient conversion matrix, V, is introduced: Using the orthogonality, we get:

335
Assuming there are k sets of observations, a sample anomaly 336 matrix with k vectors can be constructed: The retrieval error covariance matrix is:  not been used because their errors exceed 1 K. If data from these 419 channels were to be used for retrieval, the accuracy of the retrieval 420 could be reduced. Therefore, it is necessary to select a group of 421 channels to improve the calculation efficiency and retrieval quality.

422
In this paper we study channel selection for temperature profile 423 retrieval by AIRS.  v12 can accept input profiles on any defined set of pressure levels.

439
The majority of RTTOV v12 coefficient files are based on the 54 440 levels (see Table A1 in Appendix A), ranking from 1050 hPa to 0.01 441 hPa, though coefficients for some hyperspectral sounders are also 442 available on 101 levels.

443
The weight function matrix, K (Jacobian matrix), in this paper is   The ARI tends to be 0.38 and is not convergent, so the PCS 563 method needs to be improved. In this paper, the atmosphere is between the number of iterations and the ARI is shown in Fig. 6.

568
When the number of iterations approaches 100, the ARI of ICS tends 569 to be stable, and reach to 0.54. Thus, in terms of the ARI and 570 convergence, the ICS method is superior to that of PCS.  Table A2 (see Table A2 in Appendix A).

614
The locations of selected profiles of temperature, specific

672
For the statistical inversion comparison experiments, the standard 673 deviation of temperature retrieval is shown in Fig. 10      In order to further compare the regional differences of inversion   this paper is feasible and shows great promise for applications.