Angal, A; Geng, X; Xiong, XX; Twedt, KA; Wu, AS; Link, DO; Aldoretta, E (2020). On-Orbit Calibration of Terra MODIS VIS Bands Using Polarization-Corrected Desert Observations. IEEE TRANSACTIONS ON GEOSCIENCE AND REMOTE SENSING, 58(8), 5428-5439.

The Moderate Resolution Imaging Spectroradiometer (MODIS) instrument on the Terra spacecraft is completing two decades of successful Earth observations, providing the scientific community with numerous products and supporting applications, such as land surface cover, sea surface temperature, aerosol properties, and vegetation. The 20 reflective solar bands (RSBs) cover a wavelength range from 0.41 to 2.1 mu m and are calibrated primarily using a solar diffuser (SD), with lunar measurements and Earth-view (EV) response trends from desert sites used for the response versus scan-angle (RVS) characterization. Prelaunch analysis showed that a few short-wavelength RSBs of Terra MODIS are particularly sensitive to the polarization of the incident light, with on-orbit results further indicating that the polarization sensitivity has experienced changes that are wavelength, mirror side (MS), and scan-angle dependent. Although the primary calibrator, the SD, should provide an unpolarized light, supplemental inputs from the EV response trends used in the RVS characterization are impacted by the Earth scene polarization. The EV trends, uncorrected for polarization, pose a significant challenge in the RVS characterization for the short wavelength bands (3, 8, 9, and 10), therefore impacting the long-term trends and uncertainty characterization. Previous studies from the MODIS Characterization Support Team (MCST) and the National Aeronautics and Space Administration (NASA) Ocean Biology Processing Group (OBPG) have independently estimated the polarization correction to be applied on the calibrated products to mitigate the impacts to some extent. In this article, the RVS is characterized using the EV response trends after correcting for the polarization effects. Results indicate a significant improvement in the long-term trending, reduced uncertainties in the forward prediction, and a more accurate per-pixel uncertainty provided by the uncertainty index (UI) in the MODIS L1B. This enhanced approach will, therefore, be considered for implementation in the future L1B reprocessing efforts. The enhanced approach also vastly simplifies the process of adding long-term trend corrections on top of the MODIS L1B, which should streamline the science data production in the future.