| dc.description.abstract | Over the last century, the North Sea experiences eutrophication in its coastal zones due to elevated nutrient loads from rivers (Neal et al. 2000). In several studies, observed changes in phytoplankton biomass in the North Sea have been coupled to an increase in anthropogenic nutrient loadings (Vermaat et al. 2008; Los et al. 2008; Blauw et al. 2010). Elevated phytoplankton primary production rates can locally lead to harmful situations such as a rapid increase of a phytoplankton population, called a ‘bloom’ (Peperzak and Poelman, 2008). The link between nutrients and phytoplankton primary production rates is not always straightforward. Especially in coastal waters, modelling of ecosystems provide a challenge in terms of physics, biogeochemistry and ecology as large fluctuations and strong spatial gradients exist in the salinity, suspended matter concentrations, nutrient concentrations and phytoplankton biomass. Computer models prove a valuable tool to further examine the driving forces of phytoplankton growth in the North Sea. In this study, validation results of a Generic Ecological Model (GEM) with regard to the prediction of phytoplankton growth and nitrate, phosphate and silicate concentrations in the North Sea are presented. Up till present, mainly measurements from the Dutch water authority Rijkswaterstaat were used for the validation and calibration of GEM, focussing on the prediction of nutrient and chlorophyll-a concentrations in the Dutch Exclusive Economic Zone (EEZ). It is not known how GEM performs when tested using a larger dataset, covering the entire North Sea. Possible causes of errors in modelled nutrient concentrations are identified and suggestions for the improvement of model fit in the Dutch EEZ are given. The GEM model shows a negative bias of modelled chlorophyll-a concentrations with respect to observations in most regions of the North Sea. In the Dutch EEZ, this is because the initialisation of the spring bloom is predicted 2-4 weeks too late. This can be due the fact that the residual flow through the Strait of Dover is being underestimated, leading to high concentrations of DOC in the Dutch coastal zone. Nutrient concentrations are generally being overestimated, which is likely to be partly due to an underestimation of the residual flow through the Strait of Dover. Modelled nutrient concentrations can be improved by recalibration several parameters related to biochemical processes. The sensitivity of the model to the parameters is strongly region dependent. Therefore, they would need to be adjusted regionally. Generally, the model is most sensitive to sediment related parameters in coastal regions and to water column related parameters in offshore regions. | |
