Eutrophication, i.e. the enrichment of nutrients in estuaries and coastal areas by industry, agriculture and other human activities represents a severe problem for ecosystems and coastal aquatic communities. Due to an increase in algal growth, triggered by nutrient injection, the oxygen concentration can decrease below deadly thresholds, carbon dioxide is emitted and lowers the pH of the water, known as ocean acidification, a danger for shellfish and other organisms. Additionally, harmful algal blooms  (HAB) can occur, which emit toxic substances that can cause animal mortality and also poisoning of humans due to seafood and many more.
Thus the development of early warning systems for severe eutrophication states is of great importance. The data needed for such a system are provided by EMODnet, which has the mission to make marine data for the European sea regions, easily accessible, interoperable and free to use. 
In this project, we will explore the feasibility of novel machine learning algorithms, trained and evaluated on historical EMODnet data, as an early warning system for severe eutrophication conditions in European coastal waters. 

The idea is to include as many measured parameters as available, e.g. oxygen, pH, chl-a, nitrate, phosphate, salinity, temperature, etc. to extract patterns, which give early warning signals for upcoming severe conditions. Fully open questions are how good the prediction skill of such an approach might be and on which time scales skilful predictions are feasible. For the data pre-processing and visualization we will use the popular Ocean Data View (ODV) software.
 

 

 


The Challenge

PREDICTING ENVIRONMENTAL RISKS - Challenge: 3A

EutroWarn

  • Team Lead
    Germany
  • Other
    Italy
  • Other
    Italy
  • Other
    Germany
  • Other
  • Other
  • Other
    Nigeria
  • Other
    Myanmar
  • Other
    India