Assimilation of SeaWiFS chlorophyll data into a 3D-coupled physical–biogeochemical model applied to a freshwater-influenced coastal zone

Abstract

In order to predict eutrophication events in coastal areas we tested an assimilation scheme based on sequential data assimilation of SeaWiFS chlorophyll data into a coupled 3D physical–biogeochemical model. The area investigated is a semi-enclosed estuarine system (Gulf of Fos–North-western Mediterranean Sea) closely linked to the Rhone River delta. This system is subjected to episodic eutrophication caused by certain hydrodynamic conditions and intermittent nutrient inputs. The 3D hydrodynamic model Symphonie was coupled to the biogeochemical modelling platform Eco3M. Surface chlorophyll concentrations were derived from SeaWiFS data using the OC5 algorithm and were sequentially assimilated using a singular evolutive extended Kalman filter. Assimilation efficiency was evaluated through an independent in situ data set collected during a field survey that took place in May 2001 (ModelFos cruise). An original approach was used in constructing the state vector and the observation vector. By assimilating pseudo-salinity extracted from the model biogeochemical dynamics in both open sea and plume region were respected. We proved that substantial improvements were made in short-term forecasts by integrating such satellite-estimated chlorophyll maps. We showed that missing freshwater inputs could be corrected to a certain extent by the assimilation process. Simulated concentrations of surface chlorophyll and other basic components of the pelagic ecosystem such as nitrates were improved by assimilating surface chlorophyll maps. Finally we showed the coherent spatial behaviour of the filter over the whole modelled domain.