Remote sensing of bio-optical water types, phytoplankton seasonality, and algal pigments in ocean margin waters

A frontal edge detection algorithm was applied to remotely sensed ocean color satellite data to identify incorrect retrievals of phytoplankton chl a concentrations, and refine estimates of primary producer abundance in bio-optically complex ocean margin waters. Improvement of the remotely sensed biological signal will facilitate establishment of more accurate daily to decadal phytoplankton spatial patterns in these waters, and enable prediction of phytoplankton blooms or features from space. Spatial patterns of chlorophyll a (chl a) and water-leaving radiance (L_wn) from 1998 SeaWiFS (Sea-Viewing Wide Field-of-View Sensor) images were examined from ocean margin waters off the southeastern continental United States (SEC). Ocean margin waters are bio-optically complex due to riverine input, terrestrial runoff, and associated dissolved and particulate materials. Dissolved and particulate materials affect water-leaving radiance values in regions of the electromagnetic spectrum (412, 443, 555 nm) where their absorption and scattering properties are strongest. The radiative signal of non chlorophyll-containing fractions is misinterpreted as chl a. Waters are bio-optically classified as dominated by phytoplankton and derivative products (Morel Case I), or non chlorophyll-containing in-water constituents (Morel Case II). An edge detection algorithm delineated bio-optical water masses. Spatial congruence of L_wn(555) and chl a fronts defined Case II waters, and residual chl a fronts identified Case I waters. Monthly phytoplankton spatial variability was examined during January, March, May, August, and November, representing major seasonal periods. Phytoplankton were associated with a shelf region based on their response to local physical forcings. River flow and wind stress affect inner shelf chl a distributions, while offshore chl a distributions are controlled by Gulf Stream meanders. Carolina Capes' oceanography influenced chl a frontal variability. Radiance data at 443nm are affected by detrital, phytoplankton chl a, and carotenoid absorption, while blue green algal phycobilipigment absorption affects 490nm radiances. Waters at the Gulf Stream boundary showed larger, carotenoid containing phytoplankton cells and phycobilipigment presence. It is assumed that SeaWiFS has an accurate atmospheric correction, and major spectral absorption peaks for phytoplankton accessory pigments are coincident with literature values. Radiance fronts at 443nm better reproduced chl a fronts than those at 490nm (used in the SeaWiFS bio-optical algorithm).