A priori estimation for spectral shift of atmospheric carbon dioxide satellite measurement

Atmospheric carbon dioxide (CO2) satellite measurements need high spectral resolution and high signal-to-noise (SNR) to achieve high precision of 1% or better. However, measurements with high spectral resolution are suffered to spectral shift and channel mismatch is inevitable, which will causes CO2 retrieval error. For spectral shift correction, a priori information of wavenumber offset is important. We build the spectral shift model with correction factors composed of wavenumber squeeze and offset, designed the method of estimating the a priori information of wavenumber offset through convolution kernel, and developed the retrieval method of correction factors based on a priori information. The ability of estimating wavenumber offset a priori is most stable for 0.27 cm−1 convolution kernel, which can accurately find the reference line from GOSAT 1 September 2013 global measurements with SNR larger than 100 and without cloud contamination. Based on a priori information, correction factors retrieval precision is better, and RMS is reduced nearly by 85% for twelve days of global GOSAT measurements, implying better agreement between the simulated spectra and GOSAT. This technique can be applied to other high spectral resolution measurements, such as OCO-2, TanSat, GMI and so on.