Eumetnet workshop for the application of Artificial Intelligence to weather and climate studies
Weather and climate studies are based on large amounts of heterogeneous observations that are combined by online data assimilation with physical models to produce weather forecasts or climate change analysis with high societal impact. Both observation data volume and data diversity are expected to increase exponentially in the future, while meeting the future requirements for forecast reliability and timeliness needs 100-1000 times bigger high-performance computing and data management resources than today.
Artificial Intelligence (AI), of which Deep Learning (DL) is an important part, is an explosively growing field of technology, which is expected to transform many aspects of society in a fundamental way. DL is a generic data driven technique that is able to discover very complex relationships between input and output data in an automated way by the offline training on large training datasets of input and corresponding output data. Being given 1) the data driven nature of weather and climate studies, 2) the remarkable outperformance of DL compared to more traditional methods in other application fields, 3) the above mentioned growing user requirements that are difficult to meet with the more traditional methods, the application of AI and DL to weather and climate studies is logical and unavoidable.
The goal of this workshop is to investigate the potential of Artificial Intelligence (AI) techniques to address e.g.:
• the optimal quality control, data fusion and data extraction of both traditional (weather station, precipitation radar, satellite, ...) and new (private weather station, smartphones, car sensors, ...) observation types;
• the optimal nowcasting and seamless short-term forecasting of weather fields (precipitation, visibility, ...);
• the optimal parametrisation of sub-grid physical processes in NWP models;
• the optimal weather type and model dependent correction of systematic errors (post-processing) of NWP output.
The workshop will start on 26/2 at 10 pm and end on 27/2 at 4 pm. It will comprise plenary presentations of existing first examples of the application of AI to weather and climate studies, as well as a discussion on the way forward towards a further development and sharing of this AI expertise.
Experts interested to participate are invited to register and submit a short abstract proposing a contribution regarding an existing or potential weather and/or climate application of AI. The workshop program is expected to be finalised by December 20th 2020.