The killer shrimp (Dikerogammarus villosus).
Photo: S. Giesen, CC Creative commons

Ocean data and AI to tackle climate change and invasive species

The Ocean Data Factory (ODF) project wants to enable Sweden to be one of the leaders in using ocean relevant data and integrating AI into the understanding and analyzing of data. Yixin Zhang from the Department of Applied IT leads the continuous evaluation and innovation package within the project.

– It is not only about the ocean, we consider this also as an innovation platform where we bring, for example, data suppliers, data analysis and ocean data experts as well as the academia, the industry and the public together. So, people within the project have many different expertise, but the overall goal is to strengthen Sweden in terms in ocean data, says Yixin Zhang.

Another way of explaining the work within ODF is that the project is not about the data per se, but ocean challenges and how data can address those challenges.

– More concrete, one of the challenges is about the prediction of invasive species. We have finished the first cycle (out of four) within the project which had a focus on the so-called killer shrimp (Dikerogammarus villosus). It's really a killer and changing the oceans local ecosystems, says Yixin Zhang.

yixin zhang
IT-researcher Yixin Zhang.
Photo: Peter Larsson

How can machine learning methods help?

According to ODF initial problem formulation, the killer shrimp's presence has been recorded in rivers in Western Europe. It presumably ended up there by travelling through inland waterways from the Black Sea, and assumed to be carried by cargo ships. The proposed research question within the project was therefore; Can machine learning methods help to predict the areas of the Baltic sea, set on Sweden’s east coast, which would be suitable for the killer shrimp?

By using the open data collected about the killer shrimp and adding visualization the team have set up a platform where possible scenarios regarding the shrimp’s movement is presented through a map and other elements. The data on the killer shrimp is then open and accessible for everyone. A killer shrimp invasion challenge is also open on Kaggle, and everyone is welcome to join the challenge.

– We want to estimate how the shrimp will spread and it's not only for marine experts to look at and to work upon. But another very important thing is to help the public to build awareness and inform them, both on invasive species and that large amounts of ocean relevant data is open. So we publish news on GitHub and Kaggle which are open platforms. This is not only information for a closed to group of experts to look at. Rather, we want to get people engaged and in order to do that we have to lower the entry barrier for them, keep it simple and open, says Yixin Zhang.

Two year project funded by Vinnova

ODF is a two-year project funded by Vinnova and have four cycles (out of which the killer shrimp was number one). The second cycle will look into identifying marine species such as corals recorded in underwater videos and visual image analysis). But the relevance of the project has many different levels, according to Yixin Zhang.

– There are so many levels of importance, the policy level for example. Many governments are concerned about climate change, and more specifically the impacts on the ocean. And one thing in finding a solution is to have better policy, because it means we need to be better informed and how to get us better informed? Well, then we need knowledge which we get from the data within a project like this, she says.

It's about how we use the data and how it may help us to innovate.

– And the other thing, is the innovation perspective. It's about how we use the data and how it may help us to innovate. I think we have more than 20 partners from for example University of Gothenburg, Chalmers, RISE, SMHI and SLU as well as other governmental institutions. It's all about sharing information, resources and ideas since there is a knowledge gap in terms of the ocean and data scientists. Data scientists may not know that much about the ocean, and for people mainly work with ocean data, the AI and the machine learning, can be quite new to many of them, she says.

PhD from Hong Kong University of Science and Technology

Yixin Zhang is originally from China and earned her PhD in Information Systems from the Hong Kong University of Science and Technology and Bachelors in Business Administration and Software Engineering from the University of Hong Kong. After working four years in Finland she moved to Sweden and has been at the University of Gothenburg for 1,5 years.

She got involved in the project after colleague Fredrik Svahn asked her whether she was interested in working with open ocean data. She knew little about ocean data before, but more about open data and how information technology can be used to change peoples understanding and perceptions.

– I learn more and more about the ocean along the way, partly because of the project, also partially because of my interest already in nature. I quite love nature.


ODF Sweden is coordinated by University of Gothenburg, Department of Marine Sciences
ODF Sweden's director is Robin Teigland from Chalmers University of Technology
ODF Sweden has published Killer Shrimp Invasion Challenge on Kaggle, welcome to join the challenge!
ODF Sweden has published codes for data extraction on Github
Visualization of the invasive species model is provided online
Explore Sweden’s marine biodiversity by watching deepwater video recordings and identify the marine species