Relative Spectral Response Function Retrieval of Hyperspectral Instruments in Atmospheric Spectrometry

Abstract

Nowadays, tens of satellites carry hyperspectral spectrometers. Such instruments allow decomposing the light that exits the atmosphere from its top into hundreds to thousands of contiguous spectral channels. By analysis of the light spectral distribution, and in particular the depths of selected absorption lines, researchers and meteorological agencies can retrieve the atmosphere composition and thermodynamic state. To get a global view of the Earth, several instruments are generally operated synergistically, therefore, a harmonized calibration must be achieved between them. To cross-calibrate two spectrometers, a common practice is to analyze an ensemble of collocated measurements, meaning acquisitions performed at the same time and under the same geometry. Nonetheless, such analysis always faces the issue of setting appropriate temporal and geometric thresholds in defining the collocations, trading off between statistics and quality. Consequently, some collocation mismatches may have a substantial impact on the cross-calibration results. Thus, the following manuscript describes in detail the inclusion of collocation errors into the mathematical description and presents an application which is designed on purpose to be robust to such errors. Then, the knowledge of the spectral sensitivities of each channel to the incoming light, called the spectral response functions (SRF), are key to the exploitation of the acquisitions. In that context, the authors have studied and designed a novel methodology to retrieve relative SRF between two or more spectrometers, within a single instrument or between instruments embarked on different platforms. The objective of the methodology is to characterize discrepancies of responses between flying spectrometers, track long-term evolutions and harmonize their responses with post-processing when necessary.