Theory-informed and Data-driven Approaches to Advance Climate Sciences

The understanding of the fundamental dynamical and statistical properties of the climate system is a grand scientific challenge of contemporary science. Despite substantial progress, reaching a complete theory able to relate climate variability and climate response across scales is proving elusive. While a lot of progress has been made, various fundamental theoretical and practical problems still need to be addressed. In the context of numerical weather forecast, accurate and efficient parametrizations are key to achieving high skill in prediction. We are now living in the time when machine learning is revolutionizing the way we look at complex systems, in terms of both modelling and data processing. A critical appraisal of its potential and pitfalls in the context of climate-related research is extremely timely. The goal of this special issue is to collect contributions that, using theory-informed and/or data-driven methods, aim at advancing: i) our understanding of the dynamical properties of the climate system and of its individual components, ii) our ability to model the climate system and of its individual components and to achieve accurate predictions at different spatial and temporal scales; iii) our ability to merge optimally data and model outputs using innovative data assimilation methods across scales and/or involving one or more components of the climate systems.

Guest Editors: Valerio Lucarini, Davide Faranda, Alexander Feigin, Juan Restrepo, and Nikki Vercauteren