Forecast verification in the Ligurian Sea. Multiresolution analysis and study of the thermocline

The results of the GHER 3D model are analysed, in order to establish the benefits of a Sea Surface Temperature (SST) assimilation experiment. The influence of the assimilation into the results of the model is examined first in the studied domain, the Ligurian Sea. Then, the benefit of the assimilation outside this domain (in the nearby zones at the surface, and in the Ligurian Sea at depth) is also studied. Finally, the effect of the SST assimilation on the other variables is examined.
The procedure for the skill assessment of the model is as follows. First, the classical verification tools are applied: Root Mean Square Error (RMSE), Anomaly Correlation Coefficient (ACC), and Mean Square Error Skill Score (MSESS). Climatology and a free run of the model are used as reference systems. After this, a multiresolution analysis is carried out, to decompose the model results into different spatial scales. At each scale, the error measures mentioned above are applied. This allows to establish which scales are mainly contributing to the error. For this multiresolution analysis, a Discrete Wavelet Transform is used.
The study of the assimilation benefits at depth is made by comparison with CTDs. The aim is to study the position and strength of the thermocline, as this zone presents high variability and it has an important impact into the system. A good representation of the thermocline is thus interesting.