Description
Keywords: Digital Twins, Machine Learning and AI, Data management
Climate Change has been leading to an exacerbation of Extreme Weather Events (EWEs), such as
storms and wildfires, raising major concerns in terms of their increase of their intensity, frequency
and duration. Detecting and predicting extreme events is challenging due to the rare occurrence of
these events and consequently the lack of related historical data.
Machine Learning (ML) approaches represent emerging solutions for dealing with extreme events
analysis, providing cost-effective and fast-computing methods that can complement or replace
traditional methodologies. Such solutions require huge amounts of heterogeneous data for properly
training and running the models, which in turn pose big challenges in terms of data management,
computing/memory resource requirements, workflow orchestration and software infrastructure
needs. A Digital Twin for EWEs integrating data and models could provide scientists and policy
makers with a system for conducting prompt analysis and evaluating what-if scenarios.
In the context of the EU-funded interTwin project, a Digital Twin for the analysis of extreme events,
targeting tropical cyclones and wildfires, on future climate projections following a data-driven
approach is being developed. The interTwin project aims at defining a Digital Twin Engine for
supporting scientific Digital Twins applications from different fields. This contribution will present
the initial work behind the design of this machine learning-powered Digital Twin for extreme events
studies as well as some preliminary results.
| Key Topic | Digital Twins |
|---|
