Détection satellitaire et modélisation opérationnelle de la production végétale non-fixée dans la bande côtière bretonne

The negative impact of the man on the environment unquestionably increases. The coastal environment as a whole is the maritime environment which is the most exposed to man-made eutrophication. For instance, nitrate loads to coastal waters off Brittany (France) have increased 10 times (from 3 to s 30mg L-1NO3) during the four last decades. Eutrophication of this coastal zone is now a recurrent problem, more and more worrying. This eutrophication exhibits two main forms : " green tides ", which are caused by local proliferations and stranding of green macro-algae (particularly the Ulva species) in nitrogen-enriched areas ; phytoplanktonic blooms, which are responsible for coloured water events, toxic species can be present (Pseudo-Nitzschia, Alexandrium, Dinophysis), and induce banning of shellfish consumption. The objectives of this study are to assess the respective role of various rivers (especially the Loire and the Vilaine) in the eutrophication of the Brittany shelf ecosystem and test the impact of nitrogen enrichment in the triggering of Pseudo-Nitzchia toxicity. For these purposes, a coupled hydrodynamical/ biogeochemical 3D model has been developed and validated against past measured time-series and satellite images. After a brief analysis of primary production over the Brittany shelf under different conditions, we show a description of the real time adaptation of this model. Daily results are currently presented in real time on www.previmer.org web site. The inability of the model to correctly simulate near shore phytoplankton is worsened in this model by the fact that the turbidity field used is a climatological one, made by satellite images averaged on a decadal basis. First, near shore pixels of satellite images are scarcely reliable, secondly a climatological mean cannot represent the sudden bursts or drops of the real turbidity, highly dependent on the actual sea-state : this can explain also why the model at this moment fails to reproduce a sudden bloom occurring in a short period of sunny and calm weather. The simulated Pseudo-Nitzschia biomass and induced risk of toxicity in the water column cannot really be validated offshore, because monitoring measurements are available only near the coast, and deal only with toxicity in benthic filter feeders. To get temporarily rid of that problem, an interesting improvement would be to add to the operational model the bio-accumulation of toxin in the shellfishes : simulated concentrations in shellfish fresh would become directly comparable with usual measurements of the monitoring networks. Our model points out the important impact of Loire river loads on the primary production on the western and southern Brittany. This result is crucial for administrations in charge of reducing the terrestrial loadings responsible for eutrophication, because it can delay reduction programs on small tributaries until the main input from the Loire has been significantly reduced. A question arises, about the reliability of the haline stratification in the model, because the horizontal extent of the Loire plume will be enhanced by a stronger stratification.