Evaluating degradations of TANSO-FTS/GOSAT using principal component analysis

Abstract

It is reported that the sensitivities of the short wavelength infrared bands of Thermal And Near-infrared Sensor for carbon Observation (TANSO) - Fourier Transform Spectrometer (FTS) on Greenhouse gases Observing SATellite (GOSAT) have been degraded with the wavenumber dependencies1. These degradations affect to the XCO2 and XCH4 retrievals, so they have to be correctly considered in the retrieval algorithm. In this work, we developed a new algorithm to evaluate these degradations from on-orbit solar calibration spectra using principal component analysis. The effectiveness of this algorithm is to be able to distinguish the time-dependent components of the spectral variability from the independent ones. This possibly enables us to evaluate the temporal change of sensitivity more precisely. The degradation models were constructed by decomposing spectra, fitting principal components scores using the appropriate functions, and reconstructing those functions. The first components of the decomposed eigenvectors have less spectral dependencies for each band, because they are due to the angular dependency of reflectance of the diffuser plate. On the other hand, the other components have significant spectral dependencies and their temporal variabilities do not correspond to that of the first component. This fact indicates that the components have to be separately considered to construct the degradation models of TANSO-FTS.