An empirical approach to ocean color data: Reducing bias and the need for post-launch radiometric re-calibration

A new empirical approach is developed for ocean color remote sensing. Called the Empirical Satellite Radiance-In situ Data (ESRID) algorithm, the approach uses relationships between satellite water-leaving radiances and in situ data after full processing, i.e., at Level-3, to improve estimates of surface variables while relaxing requirements on post-launch radiometric re-calibration. The approach is evaluated using SeaWiFS chlorophyll, which is the longest time series of the most widely used ocean color geophysical product. The results suggest that ESRID 1) drastically reduces the bias of ocean chlorophyll estimates, most impressively in coastal regions, 2) modestly improves the uncertainty, and 3) reduces the sensitivity of global annual median chlorophyll to changes in radiometric re-calibration. Simulated calibration errors of 1% or less produce small changes in global median chlorophyll (< 2.7%). In contrast, the standard NASA algorithm set is highly sensitive to radiometric calibration: similar 1% calibration errors produce changes in global median chlorophyll up to nearly 25%. We show that 0.1% radiometric calibration error (about 1% in water-leaving radiance) is needed to prevent radiometric calibration errors from changing global annual median chlorophyll estimates more than the maximum interannual variability observed in the SeaWiFS 9-year record (± 3%), using the NASA standard method. This is much more stringent than the goal for SeaWiFS of 5% uncertainty for water leaving radiance. The results suggest that ESRID, combined with a field sampling program, can improve the quality and reliability of ocean color data, while promoting a unified description of ocean biology from satellite and in situ platforms. Although the results here are focused on chlorophyll, in principle the approach described by ESRID can be applied to any surface variable potentially observable by visible remote sensing.