The challenges of AI for weather and climate, or how to use big data to succeed in the 21st century

Summary

Meteorology and climatology were among the first big data generators and were an important driver of innovation in the development of high-performance computers. 

Already in the 1970s to 1990s, the application of statistical methods (in the post-processing of the outputs of numerical weather forecasting models), the ancestors of machine learning approaches (analog, perfect prog, MOS, UMOS etc.), were used with the big data of the time. The last two decades have seen an explosion in the volume, accessibility and diversity of data, as well as an improvement in the accessibility of high-performance computing systems at low cost. This has led to an acceleration of machine learning applications and the introduction of more sophisticated artificial intelligence techniques.  

This presentation will describe the “playground” that is atmospheric and climate science. It will also stimulate reflection on aspects associated with big data and the global infrastructure