unizeit titel

Big Data offensive in marine sciences

Kiel University and GEOMAR, together with partners in Bremen, are founding the Helmholtz graduate school MarDATA. Its goal is to train technical specialists in the intelligent analysis of big data sets, who are sought-after around the world. The focus is on applications for marine sciences.

a big wave pixelated
© based on "Die Große Welle vor Kanagawa", Hokusai; modified by pur.pur

Many questions in marine and ocean research are inconceivable without computer science or mathematical competence. The aim of the MarDATA Graduate School is to train experts in this field.

Google, Amazon and co. have shown how data can be put to profitable use with modern analysis methods. This field is known as Data Science. Technologies and theories from mathematics, statistics and computer science are used to gain new insights from large amounts of data. Terms that are used in this context are neural networks, machine learning, pattern recognition and artificial intelligence. "Data science is a hot topic in the sciences as well as the business world. In every sector, data is generated that needs to be usefully evaluated," emphasizes Kiel mathematician Professor Malte Braack. "People who are familiar with data science methods currently have their pick of jobs."

One field where large amounts of data are generated is the marine sciences. "This includes data captured from measurements taken around the clock and stored by autonomous ocean observatories, for instance. But it also includes data generated by numerical simulations, such as computer simulations of climate change," explains the Professor of Applied Mathematics at Kiel University, who is involved in the marine science priority research area Kiel Marine Science (KMS). Developing mechanisms for this purpose that can automatically extract important information or detect patterns in measurement series requires knowledgeable personnel. These technical specialists are exactly the people that the new Helmholtz School for Marine Data Science (MarDATA) would like to train.

"The marine science institutes provide the research environment and exciting and urgent questions related to the ocean and climate, while the universities provide the necessary methodological expertise," adds the spokesperson for MarDATA, Professor Arne Biastoch from the GEOMAR Helmholtz Centre for Ocean Research Kiel. "The concept has benefits for both sides. We get approaches to solutions and our colleagues at the university get concrete applications."

People who are familiar with data science methods currently have their pick of jobs.

Malte Braack

One of these specific questions has to do with real-time control of undersea robots. Up to now, the robots have travelled along a specified route over the ocean floor, taking photos and video sequences. It would be helpful for research if the robot could decide for itself, on the basis of the pictures it has taken, which areas would be worthwhile exploring more closely. "This could be a black smoker, for example, a particular animal, or an interesting geological formation. It would be convenient if the robot could recognize these features and make more detailed recordings of them," explains computer scientist and KMS member Professor Willi Hasselbring.

One doctoral project that Hasselbring is supervising deals with the subject of machine learning for petrophysical predictions. Using techniques for machine learning, the results of existing bores will be analysed. The goal is to discover previously unknown relationships between the physical characteristics of drill cores and the various geological conditions, such as distance from shore, sediment depth or age of the crust.

These are two of the nine projects by Kiel doctoral researchers that are expected to produce important insights for marine science research. At least as important is the advancement of research with respect to new methods for data analysis. "This is critical," says MarDATA spokesperson Biastoch. "We want to make use of the development of current methods, which means that we need to join forces with the methodological experts." This multidisciplinary education is new: doctoral researchers will publish not only in typical marine science journals, but also in publications from their own actual fields of computer science or mathematics. With this methodological expertise and challenging technical background, they will have excellent prospects in the job market after graduation, even beyond science and research.

Author: Kerstin Nees

Expertise for data-driven science

The Helmholtz School for Marine Data Science (MarDATA) is a graduate school financed by the Helmholtz Association. Partners include the GEOMAR Helmholtz Centre for Ocean Research Kiel, the Alfred Wegener Institute (AWI) Helmholtz Centre for Polar and Marine Research in Bremerhaven, Kiel University, the University of Bremen, and the private Jacobs University in Bremen. During the six-year funding period, a total of 40 doctoral researchers will be trained in two cohorts. They are working on marine science questions for which new methods of data analysis are needed. They are advised by tandems of experts from the marine sciences, mathematics, computer science and bioinformatics. Participants receive a broad education in joint interdisciplinary block courses and summer schools. Regular colloquia allow for continuous exchange among all participants. The first group is already full. A total of eight women and six men from Germany, Russia, India, France, Iran, Pakistan and Kenya are starting their doctoral projects in Kiel and Bremen.


Further information on MarDATA

PDF download




unizeit Archives