Harvard/SACLANTCEN -- Nowcasts/Forecasts

Harvard Ocean Prediction System Model Results

Currently, two distinct models are being used. The first is a subtidal Primitive Equation model (PE). The PE model predicts the temperature, salinity and velocity fields in the domain. The second is a barotropic tidal model, which is used to predict the tidal contributions to the barotopic component of velocity. The two models are complementary, as the actual velocity is the sum of the velocities produced by the two models.

PE

Data from RR98 is assimilated into the Harvard Primitive Equation model in a fashion which mirrors the manner in which it was collected. The results presented are of the nowcast days and of the final forecast days. In this context, we define nowcast and forecast as:

Nowcast

Fields taken shortly after full assimilation of a set of data. Generally fields shown have 1/4 - 1/2 day of dynamical (model) adjustment to the data.

Forecast

Fields taken after all data has been assimilated.

Data available

At the time of model re-initialization (11 March), the following data was available to be incorporated into the model:

Model run description

The model run has an initialization based on RR98 data. The model is initialized on 16 February. The hydrographic data is assimilated on 18, 24 February and 01, 04 & 08 March.

The model run is continuously forced by FNMOC analyses and FNMOC forecast products.

Error Forecast (ESSE)

To forecast the uncertainty in the temperature and salinity fields, an ensemble of perturbed forecasts were carried out. The initial perturbations were computed based on the observational errors and data-dynamics estimated ocean variability. The technique uses the concepts of Error Subspace Statistical Estimation (ESSE).

Tidal Prediction

The tidal currents have been calculated using a linearized, barotropic tidal model designed by Lozano. The model was forced in the oceanic (western and southern boundaries) regions by the M2 and S2 tidal components from Egbert et al. (1994) oceanic tidal model, while the Straits of Gibraltar were forced by Candela et al. (1990) observations.

The model reproduces the Candela et al. (1990) co-tidal charts and velocities. The model is also in good agreement with the tidal elevation prediction from The Admiralty Tide tables, but it appears to underestimate the tidal range at neap tides (probably due to the lack of diurnal component).

Surface Maps

Four types of plots are distributed at this time.

1. Surface temperature with model (sub-tidal) velocity vectors superimposed.
2. Tidal velocities, presented at three hour intervals.
3. Forecast Error fields of Temperature & Salinity at 5 & 100 meters.
4. Mixed layer depth estimated from model output fields.