Validation of the 3D biogeochemical model MIRO&CO with field nutrient and phytoplankton data and MERIS-derived surface chlorophyll a images

This paper presents results obtained with MIRO&CO-3D, a biogeochemical model dedicated to the study of eutrophication and applied to the Channel and Southern Bight of the North Sea (48.5 degrees N-52.5 degrees N). The model results from coupling of the COHERENS-3D hydrodynamic model and the biogeochemical model MIRO, which was previously calibrated in a multi-box implementation. MIRO&CO-3D is run to simulate the annual cycle of inorganic and organic carbon and nutrients (nitrogen, phosphorus and silica), phytoplankton (diatoms, nanoflagellates and Phaeocystis), bacteria and zooplankton (microzooplankton and copepods) with realistic forcing (meteorological conditions and river loads) for the period 1991-2003. Model validation is first shown by comparing time series of model concentrations of nutrients, chlorophyll a, diatom and Phaeocystis with in situ data from station 330 (51 degrees 26.00'N, 2 degrees 48.50'E) located in the centre of the Belgian coastal zone. This comparison shows the model's ability to represent the seasonal dynamics of nutrients and phytoplankton in Belgian waters. However the model fails to simulate correctly the dissolved silica cycle, especially during the beginning of spring, due to the late onset (in the model) of the early spring diatom bloom. As a general trend the chlorophyll a spring maximum is underestimated in simulations. A comparison between the seasonal average of surface winter nutrients and spring chlorophyll a concentrations simulated with in situ data for different stations is used to assess the accuracy of the simulated spatial distribution. At a seasonal scale, the spatial distribution of surface winter nutrients is in general well reproduced by the model with nevertheless a small overestimation for a few stations close to the Rhine/Meuse mouth and a tendency to underestimation in the coastal zone from Belgium to France. PO4 was simulated best; silica was simulated with less success. Spring chlorophyll a concentration is in general underestimated by the model. The accuracy of the simulated phytoplankton spatial distribution is further evaluated by comparing simulated surface chlorophyll a with that derived from the satellite sensor MERIS for the year 2003. Reasonable agreement is found between simulated and satellite-derived regions of high chlorophyll a with nevertheless discrepancies close to the boundaries. (c) 2006 Elsevier B.V. All rights reserved.