Sensitivity of Ocean Color EDRs to Calibration Errors: Implications for the NPOESS Preparatory Project (NPP) and the Visible and Infrared Imager / Radiometer Suite (VIIRS).

This study quantifies the effect of radiometric errors on the retrieval of normalized water-leaving irradiance, chlorophyll concentration, and inherent optical properties (IOPs) from satellite remote sensing. These quantities are planned to be the Ocean Color Earth Data Records (EDRs) that will be operationally produced by the National Polar-orbiting Operational Environmental Satellite System (NPOESS). These EDRs require highly stable and accurate radiometric measurements of top-of-the-atmosphere (TOA) reflectance using the visible and NIR reflective bands. NASA is interested in factors that influence accuracy and consistency in these EDRs to ascertain whether they can be used for climate research. Three key areas of EDR sensitivity to radiometric error must be examined, including: 1) amplification of relative errors in the TOA reflectance by the atmospheric correction algorithm when computing the normalized water-leaving irradiances (nLw) for the visible bands (M1-M5), 2) the additional sensitivity of the aerosol correction part of this computation to band-to-band errors in the NIR bands, and 3) the sensitivity of the Chlorophyll and in-water optical properties (IOPs) to the band-to-band and calibration error response in the nLws. This study applies the one-at-a-time (OAT) approach to sensitivity analysis to quantify error response, however, work with band groups and pairs are explored. The results reveals a more complex response to perturbations in the input radiometry than previously thought. The also study introduces metrics for assessing data quality based on fraction of systematic data loss and subsequent biases, which is more useful than the accuracy and precision metrics used for NPP product specifications. The results show the importance of instrument stability and characterization, and the need for post-launch calibration procedures, such as those using in situ or lunar measurements.