Using SCIAMACHY to improve corrections for spectral band differences when transferring calibration between visible sensors

Abstract

The advent of well-calibrated and well-characterized visible sensors, such as the MODerate resolution Imaging Spectroradiometer (MODIS), has allowed the opportunity to cross-calibrate other contemporary, un-calibrated visible sensors. Most of the operational geostationary-Earth-orbit satellite (GEOsat) visible sensors do not have a direct means of on-orbit calibration, and the cross-calibration of MODIS with GEOsats is plagued by the differences in the sensor spectral response functions (SRFs). Spectral band adjustment factors (SBAFs) are needed to correct for the solar flux and inter-band gaseous absorption discrepancies that are caused by SRF differences, which are sometimes significant. In addressing this problem, this manuscript describes a spectral band correction technique that employs Envisat SCanning Imaging Absorption spectroMeter for Atmospheric CartograpHY (SCIAMACHY) hyper-spectral radiances to derive pseudo-radiance, or equivalent-radiance, values for the MODIS and GEOsat sensors over the calibration targets, which include a desert, deep convective clouds, and a MODIS-with-GEOsat ray-matching ocean domain. The regressions from these equivalent-radiance comparisons constitute the necessary adjustment factor. The regressions of MODIS and GEOsat pseudo-radiance values are well-behaved, with small standard error and offsets, for spectral bands that are similar. When comparing narrowband to broadband, however, the correction difference between deep convective and maritime stratus clouds can be as large as 6%. New scene-selection criteria are investigated to derive spectral band adjustment factors that are dependent on the calibration-target domains, which reduces this uncertainty. Application of these SBAFs, which are validated for accuracy using ray-matched SCIAMACHY and GEOstat radiances, is shown to bring independently derived absolute calibrations to within 1% agreement, or better, with one-another. These spectral band adjustment factors are critical for obtaining accurate and consistent absolute calibration among multiple independent and scene-dependent inter-calibration techniques given that the variation of the SBAFs as a function of scene type can be close to 8% for a narrowband-to-broadband correction.