Over the last decades, the magnitude and complexity of climate and biogeochemical data from satellite sensors, stations, and climate models has substantially increased. This is starting to overwhelm the relatively simple tools and methods currently used to analyze the data. On the other hand, recent developments in machine learning have led to powerful new methods which promise to yield novel insights - if they properly cope with the particular challenges of such data. The new field of Climate Informatics could contribute to substantially enhance our understanding of the Earth system and confidence in future climate projections. This session invites contributions of new statistical and machine learning methodology including (but not limited to) causal discovery, deep learning, probabilistic and Bayesian inference, computer vision, and advanced model-data integration applied to Earth system models and observations.
[mehr]
