How can previous year sea surface temperatures in the North Atlantic influence the fire seasons in Europe? The groundbreaking ESA-funded project “SeasFire” aspires to find out.
SeasFire “Earth System Deep Learning for Seasonal Fire Forecasting in Europe” commenced in March 2022 and it is a 12-month project, funded by the European Space Agency and coordinated by the National Observatory of Athens, that aims to develop a prototype system to predict the seasonal burned areas for Europe, using global environmental variables and artificial intelligence to model their impact on local fire regimes.
To achieve this, the project leverages the availability of a huge amount of satellite data that can be used as fire drivers called “Earth System variables”, combined with innovative approaches in Deep Learning and particularly Graph Neural Networks (GNNs), which have proved capable of capturing the spatio-temporal interactions of the Earth System variables, treating the Earth as a unified interlinked system and opening a new path for Earth System Science research. SeasFire represents the Earth as a graph and models the underlying relations that exist between fire drivers in different areas of the planet, uncovering teleconnections and memory effects in the Earth System.
SeasFire will create a prototype system that could help Fire Management Services across Europe to better prepare for the coming fire season. The service will include Europe-wide model predictions of burned areas, as well as explainability diagrams for understanding what drives the predictions.
SeasFire has been conceived as a “spin-out” of the Horizon 2020 project DeepCube (https://deepcube-h2020.eu/). It is coordinated by the research group “Orion Lab” of the National Observatory of Athens and it is being implemented in collaboration with the Max Planck Institute for Biogeochemistry (Germany) and the Harokopio University of Athens (Greece).
The consortium had the project’s kick-off meeting on 24 March 2022 and on 19 April 2022 Dr. Ioannis Papoutsis was invited to present SeasFire to ESA’s Φ-lab (https://philab.phi.esa.int/).