Evaluation of numerical models by FerryBox and Fixed Platform in-situ data in the southern North Sea

Abstract

Abstract. FerryBoxes installed on ships of opportunity (SoO) provide high-frequency surface biogeochemical measurements along selected tracks on a regular basis. Within the European FerryBox Community, several FerryBoxes are operated by different institutions. Here we present a comparison of model simulations applied to the North Sea with FerryBox temperature and salinity data from a transect along the southern North Sea and a more detailed analysis at three different positions located off the English East coast, at the Oyster Ground and in the German Bight. In addition to the FerryBox data, data from a Fixed Platform of the MARNET network are applied. Two operational hydrodynamic models have been evaluated for different time periods: results of BSHcmod v4 are analysed for 2009–2012, while simulations of FOAM AMM7 NEMO have been available from MyOcean data base for 2011 and 2012. The simulation of water temperatures is satisfying; however, limitations of the models exist, especially near the coast in the southern North Sea, where both models are underestimating salinity. Statistical errors differ between the models and the measured parameters, as the root mean square error (rmse) accounts for BSHcmod v4 to 0.92 K, for AMM7 only to 0.44 K. For salinity, BSHcmod is slightly better than AMM7 (0.98 and 1.1 psu, respectively). The study results reveal weaknesses of both models, in terms of variability, absolute levels and limited spatial resolution. In coastal areas, where the simulation of the transition zone between the coasts and the open ocean is still a demanding task for operational modelling, FerryBox data, combined with other observations with differing temporal and spatial scales serve as an invaluable tool for model evaluation and optimization. The optimization of hydrodynamical models with high frequency regional datasets, like the FerryBox data, is beneficial for their subsequent integration in ecosystem modelling.