Forecast verification of a 3D model of the Mediterranean Sea. Analysis of model results and observations using wavelets and Empirical Orthogonal Functions.

The quality assessment of the three-dimensional GHER (GeoHydrodynamics
and Environmental Research) model of the Mediterranean Sea is presented in
this work. The verification of the model results is done in a spatio-temporal
approach. Traditional error measures (i.e. correlation, mean error, etc) are
very useful to assess the quality of a model, but they do not take into account
the high complexity of three-dimensional models. The verification process is
thus done in three main parts: first, the model is compared to observations
and climatology in a qualitative approach, in order to make a preliminar
study about the model behaviour. Then, the error assessment is done, using
traditional statistic measures. In order to take into account the complexity of
the model and observations, the last step in the verification process consists in
a spatio-temporal analysis using wavelets and empirical orthogonal functions.
This last analysis will allow us to have an insight about the model quality in
a more detailed way.
This verification process has been applied to the GHER model. This model
is implemented in a two-way nesting approach in the Mediterranean Sea,
Liguro-Provençal basin and Ligurian Sea, where the highest resolution is
reached. Assimilation of sea surface temperature and sea level anomaly is
made during a nine-week experiment. Another test is carried out, to assess the
quality of sea surface temperature from the SOFT predictor of the Ligurian
Sea. The predicted sea surface temperature is assimilated in the model and
the quality of the forecast is compared to the first assimilation experiment.
The assimilation of the SOFT statistical predictors can be very useful to
force models in a real forecast experiment, where no observations are available.