To overcome limitations in computing and data analytics related to Earth System science, the uptake of artificial intelligence (AI) and machine learning (ML) methodologies is currently being explored. Multiple initiatives are now emerging to tackle open challenges such as subscale parametrization, detection of patterns and in-situ analysis, adoption of ML for alternative process models or dedicated fast prediction systems to address specific end-user needs. This also leads to further concerns that need to be addressed with regard to the verifiability and reproducibility of results, efficient and effective use of computing and storage resources and the necessary practical software environments.
Among these initiatives, the Helmholtz Association initiated the Helmholtz Artificial Intelligence Cooperation Unit (HAICU). From the beginning of 2020 on, HAICU will form a strong collaboration across multiple disciplines to bring AI into practice. It will develop, implement, and disseminate methods of Artificial Intelligence for purposes including the analysis of complex systems in the fields of health, energy, transport and earth and environment. Specific to Earth sciences, the Helmholtz Digital Earth project has also already paved the ground for large-scale data analytics and initiated actions to foster AI adoption.
With Digital Earth, HAICU and other initiatives, there is now an accelerating momentum to tackle these challenges with a diverse set of approaches and stakeholders, which opens up a rich area of future opportunities for collaboration. This workshop will bring multiple key stakeholders together, exchange current experiences, and foster further community actions. The goals of the workshop are to assess the state of the art, identify gaps in knowledge or services, and build future community collaborations. (Source: Introduction Workshop)
The Workshop was held on February 3-4, 2020 at the German Climate Computing Center (DKRZ) in Hamburg.
Helmholtz-Zentrum Geesthacht (HZG) is one of 5 local HAICU units and following colleagues contributed with talks/posters at the workshop:
- Fitting Interpretable Scientific Models with Machine Learning (David Greenberg)
- Neural interpretation of European summer climate ensemble predictions (Julianna Carvalho Oliviera)
- Machine Learning Tools for fitting Interpretable Models to Data (Marcel Nonnenmacher)
- Ability of Neural Network in reproducing Chemical Transport Model Estimates based on meteorological data (Andrey Vlasenko)
- Feed-forward backpropagation Neural Network (Kathrin Wahl)
- Application of Machine Learning methods for reconstructions of past climate (Eduardo Zorita)