Advantages and limitations of ocean color remote sensing in CDOM-dominated, mineral-rich coastal and estuarine waters

Current operational approaches to ocean color remote sensing tend to be least accurate in coastal regions with waters rich in terrigenous material. Semi-analytical models can be used to retrieve the absorption and backscattering properties of dissolved and suspended materials unique to these environments, and subsequently biogeochemical parameters such as total suspended material (TSM) and chlorophyll (Chl). In this study, optical data collected over several years are used to select and optimize a semi-analytical ocean color algorithm for the dynamic and optically complex Long Island Sound estuary. The most successful algorithm requires a red reflectance channel between 600 and 650 nm, which is not included in many current ocean color sensors, but is essential in highly scattering waters. Regional optimization including the use of a dynamic, spectrally variable f/Q, a value related to the bidirectional reflectance distribution function (BRDF), results in an approximately five-fold decrease in retrieval bias in highly backscattering, sediment-laden waters near river outflows. Retrievals of dissolved and particulate spectral absorption, backscattering, dissolved and detrital absorption coefficients and total suspended matter obtained from the optimized algorithm agree well with field observations (r2 ≥ 0.90). These parameters are useful for assessing riverine discharge, mixing and residence times of surface waters, as well as assessing the turbidity and light penetration in this estuary. Estimation of Chl remains challenging (r2 = 0.59) due to the stepwise nature of the algorithm and the relatively high proportion of dissolved and non-algal constituents masking phytoplankton absorption (generally < 20% of total absorption at 440 nm). Moreover, diverse phytoplankton assemblages throughout the region create variability between spectral absorption and chlorophyll and highlight the benefits of increased spectral resolution of ocean color satellites going forward.

► Optimized algorithm significantly improves IOP and Chl retrievals.
► Reference band between 600 and 650 nm critical to algorithm retrievals in turbid waters.
► Spectrally varying f/Q significantly improves algorithm retrievals.
► Chl retrievals hindered by step-wise nature of algorithm and plankton diversity.