The tasks below have been formulated in order to meet the primary goals of the present project, namely to identify regions where we should have high and low confidence in the model results, and to describe details in the ocean circulation variability that are related to flow instabilities due to errors and uncertainties in the initial conditions.

Task 1. Design of the experiment

Care must be taken to avoid significant impacts on the simulation results from artifacts related to the design of the experiment. Issues that will be addressed under Task 1 include:

• size of model domain: Open boundaries must be sufficiently far away from the region of interest, so that non-perturbed boundary conditions don't significantly influence the analysis in Task 3 and Task 4. An alternative is to extend the domain in the spin-up phase, so that the ensemble simulations will be nested using perturbed open boundary conditions. Another alternative is to actually perform a set of perturbations of the boundary conditions for the ensemble simulations.
• resolution: Obviously, the resolution must be sufficiently high for generation of eddies. Results from deep ocean simulations of the Pacific Ocean reveal that eddies emerge in abundance north of 55°N with a resolution of 14 km (e.g. Melsom et al. [1999]). However, everyday experience from operational ocean circulation models for the shelf and shelf slope off Norway reveals that the resolution required for simulation of ocean eddies is about 5 km [Rĝed, pers. comm.], see also Figure 2. (Note that, in the initial years of the spin-up phase the resolution may well be coarser.)
• forcing: The initial vertical structure of model layers, and the boundary values, will be derived from the hydrography data listed in Section 3. For the spin-up phase, we will use the mean monthly winds from the 10 year set of winds listed in Section 3.

Hence, some simplified resolution tests with a duration of one or a few years will be performed in order to assess the various options listed above. Also, the derived fields for hydrography and winds will be computed.

Task 2. The experiment

The model will be spun up until the ocean circulation becomes quasi-stationary. This procedure will be performed first for a very coarse resolution, then for a resolution that is finer by a factor of 2 etc., until quasi-stationary circulation is obtained with the full horizontal resolution (as determined in Task 1). Then, the model will run for at least an additional five years with the climatological wind forcing, in order to produce a set of initial values for the ensemble simulations. Circulation features that will be checked during spin-up include kinetic energy and mean levels of selected isotherms and isohaline surfaces. Minor modifications of the model code may be advantageous in order to compute these quantities online. Finally, the set of ensemble simulations will be performed, see the depiction in Figure 3.

Task 3. Analysis of determinism and SSH as an integrated measure of the ocean circulation

Analysis of results for the sea surface height (SSH) will be performed as a separate task in the present project. There are a number of reasons for the special attention that we intend to pay to the SSH variability. First, SSH is an integrated measure of the ocean's stratification, and the circulation patterns of cyclonic and anti-cyclonic eddies are associated with SSH lows and highs, respectively. Moreover, remote sensing data for SSH anomalies exist with a temporal and spatial resolution that are sufficient for resolving eddies and meanders such as those that will result from the experiment in Task 2. The data are produced by active satellite-born sensors, so the coverage is independent of weather conditions such as the cloud cover.

In this project, the data for anomalous SSH will be used to check various aspects of the model SSH, e.g. coastal sea levels, amplitudes of SSH variability, and preferential sites for mesoscale activity. In fact, the existence of reliable SSH data is the main reasons for choosing the 1990s as the simulation period.

Task 4. Analysis of determinism and the three-dimensional structure of ocean circulation

Although we expect to gain considerable insight into the deterministic/non-deterministic nature of the ocean circulation along the Norwegian coast from the analysis in Task 3, several questions will be left unanswered. These questions are related to topics such as: Are shallow eddies less deterministic than deep eddies? Is the level of non-determinism related to the vertical extent of the upper ocean currents? In order to answer such questions, the analysis will be extended to include the depth of isopycnic surfaces.