The demonstrator “Marine Environmental Indicators” has a specific focus on data related to the marine environment, and it is led by the CMCC Foundation, in collaboration with IFREMER, Mercator Ocean International, the Royal Netherlands Meteorological Institute (KNMI), and the University of Bergen.

The objectives of this demonstrator are to:
  • Calculate and distribute online information and indicators on the environmental quality of the marine area.
  • Obtain new added-value data applying big data analysis and machine learning methods on multi-source data sets.
  • Enable users to perform online and on-the-fly operations, such as selecting a portion of a dataset, to perform statistical analysis or display the data.

The target audience of the service includes intermediate users such as environmental protection agencies, and international stakeholders in the Marine Strategy Framework Directive (MSFD), in the UN Sustainable Development Goal 14 "Life below water", and in the Blue Economy.

The service offers to them a flexible capacity to perform statistical analyses of the quality and characteristics of the marine environment for the Mediterranean Sea region, with the possibility to scale in the next version up to the Global Ocean.

Attention is also dedicated to scientific users, providing to them a tool to facilitate the discovery of new climatic indicators based on machine learning, and a simplified way to analyse oceanographic data.


The prototype MEI Generator service is a web graphical interface that allows the user to generate and display value-added environmental data from generic marine data. These value-added environmental data can be an average over time, ocean patterns or regimes … depending on the method chosen by the user; the output is proposed as a time series or a map. The current prototype uses Copernicus Marine products (physical modelling products from the Mediterranean Sea) as input data. The method implemented allows various averages.

The Ocean Patterns Indicator is based on a machine learning approach. It consists in applying a clustering, or classification method called GMM (Gaussian Mixture Model), a probabilistic model, to a set of profiles, from a structured (model output, reanalysis) or an unstructured (set of observations) dataset. Any type of variable can be used: temperature, salinity … The ocean profiles are automatically gathered into several clusters, or classes, depending on their vertical structure. When analysing the different clusters, spatial and temporal coherences can be revealed, that is what we define as the Ocean Patterns Indicator. The service offers users a flexible and innovative approach to perform statistical analyses on ocean datasets.

The Ocean Regimes indicator is based on the same clustering method as for the Ocean Pattern indicator, a machine learning approach based on a Gaussian probabilistic method, but applied to a dataset of ocean time series (Chlorophyll-a, SST…). The time series are gathered into clusters depending on their seasonal variability. For this indicator, spatial coherences can be revealed when plotting the different classes in a map. The service offers users a flexible and innovative approach to perform statistical analyses on ocean dataset.

The Storm Severity Index (SSI) service calculates maps and time series of exceptional atmospheric wind or storm circumstances that can impact seas such as the Mediterranean Sea. The SSI service can be used to study individual storms or storm/SSI distributions for a given area (in the Mediterranean Sea) and period of time (e.g. a winter season or 30 years of storm climatology). In addition, series of SSI distributions can be calculated using a time step (e.g. every year/month over the entire chosen period). The level of wind speed above which impact is expected can be indicated using a wind speed threshold value. For this wind speed threshold value, percentiles (e.g. P98 with minimum value) can be selected. These percentiles use specific threshold values for each location (grid-cell). Alternatively, a fixed wind speed threshold value can be given for the entire area. Each calculated map or time series can be plotted and saved to a file.

The service provides information on how to search for and retrieve subsets of inorganic carbon data without having to download the full file(s) by using a data server largely used by the oceanographic community. The data server ERDDAP, developed by the NOAA, allows users to explore, select and download subsets of data, in their preferred format, regardless of the format of origin. It standardizes the variable names and units of position (latitude, longitude, altitude/depth) and time. It can be accessed via script and removes the need of downloading ("copy" or "replicate" in BlueCloud) multiple and/or large files that may or may not be of interest to the user. Any kind of data can be used either gridded (as model data) or discrete (as in situ data). Many marine research data holders currently have ERDDAP servers; some examples in Europe are EMODnet, IFREMER, ICOS, EMSO, Irish Marine Institute, BODC…