A multi-disciplinary approach to teaching climate change and data science

The UN sustainable development goals, of which climate action forms a part, are increasingly being taught within higher education. In France, for example, all institutions receiving students at the undergraduate level are required to provide core education relating to the ecological transition for sustainable development, regardless of the student’s chosen discipline. There is thus a need to communicate basic climate science to students who are not directly studying this topic.

In the geosciences, masters level programmes typically include one, or several, courses on data analysis. However, both the quantity of data available, and the tools that are available to analyse this data have changed dramatically over recent decades. These questions of how to practically manage the analysis of very large quantities of data and how to apply data science techniques are relatively new, and as such have not historically formed part of the typical geoscience data analysis curriculum. Further, these questions may require a knowledge of both computer science and applied mathematics that is substantially beyond that required for simple data analysis tasks.

Motivated by these problems, a course entitled “Big data and cloud computing for climate” has been developed by a multidisciplinary group of educators from different institutions, comprising an IT engineer, a statistician and two physical oceanographers. For the past 10 years, this course has been delivered to a multidisciplinary group of students, composed of engineers and physical oceanography masters students. The aim of the course is for students to learn to use data science tools on appropriate clusters of machines to treat questions related to climate change.

In the initial phase of the course, the students follow around 20 hours of theoretical and practical classes, which cover topics such as cloud computing, the map-reduce concept, some widely-used libraries (xarray, Dask, zarr), as well as descriptive and predictive statistics, applied to ocean data. The students then spend approximately 15 hours working on group projects, mentored by one of the members of the teaching staff, in which they manipulate large data sets to investigate a specific climate question. Because the students are enrolled in different programmes, they have complementary skills for this phase of the course: the engineering students have better knowledge of data science techniques, and the physical oceanography students have better knowledge of climate science. They are thus able to collaborate to address the scientific question more efficiently.

The learning outcomes for the course depend on the students’ backgrounds: for the engineering students, they improve their understanding of the physics of climate change, and gain insight into the potential applications of the data science techniques that they have studied previously. For the oceanography students, they learn to efficiently manipulate large quantities of data and apply modern statistical analysis techniques. Student feedback about the course has been consistently positive, and the teaching collaboration has also led to research collaboration amongst the teaching staff.