Particulate organic carbon (POC) plays an important role in coastal carbon cycling and the formation of hypoxia. Yet, coastal POC dynamics are often poorly understood due a lack of POC observations and the complexity of coastal hydrodynamic and biogeochemical processes that influence POC sources and sinks. Using a dataset of field observations and satellite ocean color products, we developed a new multiple regression algorithm to derive POC from satellite observations in two river-dominated estuaries in the northern Gulf of Mexico: the Louisiana Continental Shelf (LCS) and Mobile Bay. The algorithm had reliable performance with mean relative error (MRE) of ~40%, and root mean square error (RMSE) of ~50% for MODIS and SeaWiFS images in the two systems. Substantial spatio-temporal variability was observed from satellite on the LCS, with higher POC on the inner shelf (< 10 m depth) and lower POC on the middle (10-50 m depth) and outer shelves (50-200 m depth), and with higher POC in winter (January to March), and lower POC in summer to fall (August to October). Correlation analysis between long-term POC time series and several potential influencing factors indicated that river discharge dominants POC dynamics on the LCS. Wind and surface currents also affect POC spatial patterns on short time scales. This study demonstrates that algorithms that can determine coastal POC from satellites greatly increase the spatial and temporal extent of observations available for characterizing POC dynamics and their relations to various dominant physical forcings to the continental shelf and estuaries.
Keywords: Louisiana continental shelf; MODIS; Particulate organic carbon; SeaWiFS; ocean color; remote sensing.