Scene-based spectral characterization of spaceborne imaging spectrometers in different spectral windows

Accurate knowledge of the spectral response of spaceborne imaging spectrometers, including center wavelength (CW) and full width at half maximum (FWHM), is essential for reliable retrievals of atmospheric and surface parameters from at-sensor radiance data. Pre-flight characterizations often fail to capture changes in spectral response arising from launch, orbital conditions, and instrument aging, necessitating in-flight characterization. In this contribution, we develop a scene-based spectral calibration algorithm that operates on at-sensor radiance fitting, incorporating rigorous atmospheric radiative transfer modeling to account for coupling between gaseous absorption and atmospheric scattering effects. The algorithm models surface reflectance using polynomials and simultaneously retrieve CW and FWHM shifts across the instrument swath. Sensitivity analysis investigates the potential impacts of various factors on the calibration algorithm, revealing that water vapor uncertainty significantly affects calibration accuracy, with 5 mm uncertainty causing bandwidth errors up to 0.75 nm in a specific window. Surface reflectance characteristics also influence performance, with spectrally non-linear surfaces introducing systematic biases. We applied the method to four spaceborne imaging spectrometers: EnMAP, PRISMA, GF-5A AHSI, and EMIT, revealing distinct performance characteristics and temporal evolution patterns. EnMAP demonstrates stable spectral performance with systematic spectral shifts below 0.4 nm and peak-to-peak (P2P) differences under 1 nm in both VNIR and SWIR regions. GF-5A AHSI exhibits excellent across-track uniformity in VNIR (P2P difference in CW <0.1 nm) and shows segmented variations in SWIR due to its spectial design. PRISMA displays significant temporal degradation with P2P differences reaching 3.8 nm and 6.15 nm for CW and FWHM, respectively. EMIT shows characteristic m-shaped patterns with moderate across-track variability. Quantitative assessment reveals that spectral miscalibration can cause up to 37\% systematic underestimation in methane emission quantification. The proposed algorithm provides a cost-effective complement to on-board calibration systems, enabling continuous monitoring of spectral performance and reducing potential biases in subsequent quantitative retrievals.