Benchmarks: evaluation of the results
In WP4, the benchmarks and metrics described in this webpage will be applied to compare and assess the impact of data assimilation methods of various kinds. This includes: (i) an evaluation of the common approach as defined in WP2, (ii) intercomparison of the existing assimilation tools, (iii) assessment of the stochastic assimilation methods developed in WP3. The small and medium case benchmarks will be especially used to assess the statistical and numerical effectiveness of the methods and tools gathered in the project. On the other hand, the large case benchmark will be used to evaluate the potential impact of advanced stochastic approaches (including non-Gaussian assumptions) in the real MyOcean systems. The final step of this assessment will include the assimilation of real ocean observations (satellite altimetry, ARGO float data, ocean colour) in the coupled circulation/ecosystem model.
Our expectation is that this can provide a reliable information about the current status of stochastic data assimilation methods in the perspective of improving the quality of the MyOcean products. With the metrics that have been defined, our focus is clearly the improvement of the information that is provided by MyOcean users about the expected accuracy of the data that they receive. We indeed believe that associating reliable information about uncertainties (in terms of probability distributiion, histograms, error bars,...) is crucial to any practical application of the data, whether it is for technical or political decision making (navigation, fisheries management,...) or for scientific research (model validation, forcing,...).