Find the EGU on

Tag your tweets with #EGU18

Please note that this session was withdrawn and is no longer available in the respective programme. This withdrawal might have been the result of a merge with another session.

ESSI3.3

Leveraging data-driven workflows to accelerate Earth Science research
Convener: Jens Klump  | Co-Conveners: Gustau Camps-Valls , Jess Robertson , Hua-Liang Wei , Tobias Weigel 

Machine learning methods have revolutionized many aspects of our daily life and are poised to revolutionize far more. Some areas of Earth and Space Science have long established histories of use, but most have seen only limited application. Exascale computing and data workflows challenge the currently established practices across the Earth Sciences. Advances in new sensor technology and data collection, storage and processing systems are driving an explosion in the variety and volume of data streams available to geoscience researchers. This data bonanza presents geoscientists with the new challenge of deriving valuable knowledge out of an ever increasing variety and volume of data streams without drowning in noise.

Data-driven science promises a novel paradigm in which scientific advances are gleaned directly from relationships hidden in the data. In particular approaches such as machine learning, causal inference and data mining promise a richer space of models and tools for researchers to use to interrogate large volumes of data. However, effective use of these tools generally requires close collaboration between both computational and geoscientific domain experts.

This session aims to share learnings from researchers and teams who have deployed effective data-driven workflows to Earth Science research. We encourage submissions on topics such as novel data-driven methods (including machine learning and statistical approaches) for data reduction, model parameter retrieval and estimation, prediction/classification, and discovering previously unknown relationships in data through feature selection and causal inference approaches. We particularly welcome insights on how to use machine learning effectively as part of larger research projects in the Earth Sciences.