Error quantification of a high-resolution coupled hydrodynamic–ecosystem coastal–ocean model: Part 1 model overview and assessment of the hydrodynamics

Abstract

We present an assessment of the uncertainties present in a high-resolution (∼7 km horizontal resolution and 20 s-levels in the vertical) coupled hydrodynamic–ecosystem model of the northwest European continental shelf. The models in question are a three-dimensional baroclinic circulation model developed for massively parallel computers (the Proudman Oceanographic Laboratory Coastal–Ocean Modelling System) and a sophisticated ecosystem model, representing the functioning of the ecosystem with 52 state variables (The European Regional Seas Ecosystem Model). In this paper and part 2, which focuses on the ecosystem model, we simulate the period of the North Sea Project (August 1988 to October 1989) and attempt as comprehensive an examination of model uncertainties as the available data permits. We are able to make some degree of assessment for all aspects of the physics model, apart from the turbulence closure scheme, and use a simple cost function to systematically compare the accuracy of different variables. We find some aspects (notably tidal currents and elevation, and sea surface temperature) are well modelled, with RMS errors less than 0.4 standard deviations of the data, whereas some aspects (e.g. residual current speed and salinity) have RMS errors similar to the standard deviation of the data. Improving the residual currents and salinity comparison requires improving several aspects of both the model (forcing and formulation) and observational data (quantity and quality). In addition to directing improvements in the model, this exercise also suggests where data assimilation effort might prove most fruitful. We also demonstrate that this resolution is sufficient to model complex biophysical interaction in both the horizontal (e.g. enhanced production at fronts) and vertical (e.g. mid-water production modulated by the spring–neap cycle).