A satellite ocean color observation operator system for eutrophication assessment in coastal waters

International audienceDuring the INSEA project the potential positive role that remote sensing products can play in coastal eutrophication assessment systems using assimilation into coupled hydrodynamic–biogeochemical models has been shown. However, products derived from satellite ocean color data continue to suffer from high levels of inaccuracy when compared with in situ measurements of the surface layer of the ocean. This has been particularly pronounced for coastal waters and waters optically classified as Case-II. The early success of using empirical relationships between chlorophyll and simple band ratios to derive estimates of surface layer chlorophyll from the first ocean color satellite sensors' data (i.e. CZCS), has led mainstream ocean color remote sensing and standard ocean color products towards following this approach for subsequent sensors (e.g. SeaWiFS and MODIS). Chlorophyll has continued to be the main focus product but is only related to one of the optical properties of sea water, namely the absorption of light by phytoplankton, whereas empirical band ratio approaches use wavelength banded water leaving radiance resultant from all absorption and scattering of light by all the optically active components of the ocean surface layer. We suggest that using approaches that do not fully exploit remote sensing optical data through a parameterization of the optical properties of sea water, is the main reason for the poor performance of many ocean color products when compared with in situ data. This is in concordance with the International Ocean Color Coordinating Group (IOCCG) and following their recent guidelines, novel inherent optical properties approaches (e.g. for MERIS) and the lines of research that are being used in atmospheric remote sensing, we present a demonstration ‘observation operator’ system that is based on biogeochemical model output, optical properties (apparent and inherent), and radiative transfer modeling. In the forward mode we demonstrate the system by producing MODIS and SeaWiFS synthetic images of water leaving radiance for the coastal test sites of INSEA. We show that the observation operator approach has the potential to allow the consistent mapping of model variables into observed quantities which simplifies the transport of measurement errors and reduces the need for approximations inherent in previous approaches. In conclusion we discuss the future development and potential of inversion of the system in order to obtain more accurate ocean color biogeochemical products (including chlorophyll) from satellite radiance data for eutrophication assessment. We also highlight the additional advantages there may be for ecological models from having stronger links to bio-optics