Projecting critical coastal oxygen deficits by the example of the Eckernförde Bight.

• Applicant: Professor Dr. Matthias Renz, since 10/2022
• Subject Area: Oceanography ; Data Management, Data-Intensive Systems, Computer Science Methods in Business Informatics
• Term: since 2022
• Project identifier: Deutsche Forschungsgemeinschaft (DFG) - Project number 491008639

The impact of men on the Earth System has reached a level of magnitude comparable to natural influences. Among the changes apparently accompanying the gradual warming of the Earth, are decreasing oxygen concentrations in the global oceans. This decrease in oxygen is most pronounced in coastal regions: in the 1960s only 42 of the so-called ”dead zones”, that no longer permit the survival of higher animals, have been reported. In 2008 this number has already increased to 400 (IPCC 2013). The implications for the respective ecosystem can be substantial and model-based predictions for (long- and short term) adaptation measures are highly desirable. Thus, today coupled biogeochemical ocean models are frequently used to support political decision making. A prominent example is the Baltic Sea in central Northern Europe, where ongoing periods of declining oxygen levels have reached a level that triggered the discussion of costly geo-engineering options (designed to counteract the enlargement of dead zones) and also the so called ”Baltic Sea Action Plan” was developed with the aid of numerical models to reach a good environmental status of the Baltic Sea (e.g., by limiting the nutrient supply). Despite the wide use and the great need, model inter-comparison studies still reveal considerable un- certainties and different models show a very large spread when projected into the future, even under identical atmospheric forcing and the same anticipated nutrient supply. The major aim of the proposed study is the development of techniques to make optimal use of such model ensembles and, with it, to limit uncertainties. We suggest to analyze ensemble simulations with proven methods from atmospheric sciences, in combination with novel data-mining techniques, to identify specific and generic model deficiencies and to explore the potential benefits of using ensemble predictions in the field of biogeochemical ocean modelling. With this the proposed study serves simultaneously as a test-bed for novel Data Mining and Machine Learning techniques that might be applicable for related problems. For our study, we choose the exemplary site in the Baltic Sea, the Eckernförde Bight, where commercial fishing faces socio-economic problems associated to declining catches, very similar to those experienced in other coastal upwelling systems. The site is chosen (1) since it has well documented intermittent suboxic events, resulting in increased mortality and/or emigration of several native species, and (2) since it is extraordinary easily accessible, due to its close proximity, and (3) since it is frequently sampled by our external collaboration partners. Particularly valuable is the wealth of historic data al- ready available including those at ”Boknis Eck” where one of the longest on-going monitoring time series world-wide is maintained.

• DFG Programme: Research Grants
• Ehemalige Antragstellerin: Dr. Ulrike Löptien, Ph.D., until 10/2022