Evaluating Atmospheric Correction Algorithms for Sentinel 3 OLCI in Chesapeake Bay

Chesapeake Bay is a large and optically complex estuary as suspended sediment, phytoplankton biomass, and dissolved organic matter are highly variable through space and time. This optical complexity, common to many coastal environments, poses two major challenges in the remotely sensed retrieval of water quality data. First, non-negligible water leaving radiance in the near infrared can confound the accuracy of atmospheric correction (AC) algorithms (Wang & Son, 2012). The varying contributions of inherent optical properties can also reduce the accuracy of remotely sensed algorithms that estimate water quality parameters. These challenges currently preclude the adoption of routine satellite monitoring into Chesapeake Bay management and monitoring efforts. In this study, we evaluate three different AC algorithms using data from the Ocean and Land Color Instrument (OLCI) onboard Sentinel 3A and 3B: the baseline AC, the Case 2 Regional Coast Color (C2RCC) neural network, and the Polymer AC. This study will provide an overview of each AC algorithm and contrast their ultimate impact on remote sensing reflectance (Rrs) in Chesapeake Bay. A statistical analysis will compare atmospherically corrected Rrs to matching in situ hyperspectral Rrs measurements to identify the best performing AC algorithm. Results from this study will highlight existing satellite oceanography limitations with the aim of improving operational application, which can ultimately advance water quality monitoring and management in coastal environments.