AbstractA forecast experiment in the Balearic Sea is presented which is based on the Harvard Ocean Prediction System (HOPS). HOPS is modular, containing a high-resolution primitive equations model, packages for objective analysis and data assimilation (Optimum Interpolation), an interface to implement atmospheric forcing and another interface for one-way nesting of HOPS into any other larger-scale circulation model. Here, to prevent false advection from open boundaries, HOPS is nested into the basin-scale DieCAST model [Dietrich, D.E., Haney, R.L., Fernández, V., Josey, S.A., Tintoré, J., 2004. Air–sea fluxes based on observed annual cycle surface climatology and ocean model internal dynamics: a non-damping zero-phase-lag approach applied to the Mediterranean Sea. J. Mar. Syst., 52, 145–165] and atmospheric forcing fields were provided in terms of HIRLAM fields by the Spanish National Institute of Meteorology.
The forecast capability of HOPS is demonstrated in terms of a hindcast experiment, utilising two observational data sets of a subregion of the Balearic Sea which were acquired in mid September and early October 2002. While the data of the first survey is used for model initialisation, that of the second survey serves for validation of the forecast products. The forecast skill of the system is evaluated quantitatively by three different objective methods, comparing the rms difference of vertical profiles and horizontal fields, and pattern correlations, both for temperature and salinity. In five out of six cases, the forecasted fields are closer to the validation data set than the fields used for initialisation, i.e. the forecast beats persistence and the forecast is successful.
Taking into account further available options of HOPS (implementation of additional tracers, tracking of Lagrangian particles, biological modules, two-way nesting), the system is operational for a wide field of possible applications.