You are here: Home / Communication / News archive / Tackling Climate Change with Machine Learning: [Email protection active, please enable JavaScript.] in HafenCity

Tackling Climate Change with Machine Learning: [Email protection active, please enable JavaScript.] in HafenCity

Mar 17, 2020

The group Artificial Intelligence (AI) of the Meetup international platform in Hamburg “meetup.ai” organised the “Tackling Climate Change with Machine Learning” [1] event on February, 4, 2020. On the look for earth system science experts, the organisers invited with Maria Moreno de Castro and Karsten Peters two DKRZ colleagues for a keynote presentation.

More than 100 members of the meetup.ai community discussed during the event the application of ML to tackle climate change. The first two industry-based speakers focused on supporting the ML adoption [2] and how strategies based on ML might aid humanity, e.g. to reduce CO2 emissions [3]. Maria Moreno de Castro and Karsten Peters presented the third talk [4] which drove into the current challenges of earth system science, the possible applications of ML [4] and ongoing and future efforts of DKRZ to offer state-of-the-art IT-services and support in that field [6,7].

Karsten Peters introduced the interdisciplinary audience to the challenges of understanding the earth system with models simulating physical processes at a global scale. He further highlighted the suitability of the IT-infrastructure at DKRZ to meet these challenges, especially to handle and analyze big data volumes. He explained that ML methods are suitable for sifting through such large data amounts to reveal underlying relationships, and also offer possibilities for everyday life, e.g. short-term weather forecasts.

Maria Moreno de Castro then focussed on the adoption of Deep Learning methods, which have shown a great potential in accounting for spatio-temporal relations. She introduced the fundamental problems of inconsistency and interpretability and the advantages of hybrid approaches, that is, the combination of physical models and data-driven ML-based models. Maria Moreno de Castro also introduced the two recently established ML research groups at DKRZ: the ML research group [6], led by Christopher Kadow, and the local HAICU unit AIM [7], established in collaboration with HZG.

The 30-minute discussion following the DKRZ’s presentation covered the entire spectrum addressed in the presentation and showcased an inspiring motivation of the audience to learn more about the processes, foundations, and challenges of ML and ESS that researchers are facing in their daily work.

Further information:

Document Actions