AI for Knowledge Discovery in Geosciences exact replica watches tag heuer replica fake watches rolex replica watches fake rolex replique montre replica relojes replica watches rolex replika imitasyon saat

AI for Knowledge Discovery in Geosciences


The rapid evolution of Earth Observation capabilities (e.g., Copernicus Sentinels, Earth Explorers, meteorological missions, national and commercial satellites) offer a unique opportunity to address several urgent problems facing our humanity and the planet, by providing global and synoptic observations at unprecedented spatial and temporal scales across the full spectrum of the Earth system domains. Multi-variate EO data jointly, in situ observations and advanced models are a key source of new knowledge and insights on poorly known processes and complex interactions across the Earth system components. However, due to the complexity of problems in geosciences, neither data-only nor physics-only approaches can provide sufficient representation for knowledge discovery. Geosciences unique challenges are rarely found in traditional applications, requiring novel problem formulations and methodologies in Machine Learning. Recent ML/AI advances offer a huge potential to contribute to answering grand scientific questions and can play a major role in accelerating knowledge discovery by automatically learning patterns and models from the data, while taking into account the wealth of knowledge accumulated in physics-based model representations of geoscience processes. This session addresses some of the specific geoscience challenges (e.g. long range spatio-temporal dependencies of the observed phenomena, high-dimensionality of processes, latent variables, interpretability, etc.) and presents various ML to address these challenges (e.g. mining for relationships between variables, causality analysis, Deep Learning frameworks, integrating observational data in physics-based models, etc.), as well as new technologies to facilitate analyses of multi-variate datasets (e.g. Earth System Data cubes).