Research project
Contact
Predicting Embankment Fire Risk Along Railway Lines Using Machine Learning – BurnML
12/2022 - 06/2024
Project volume:99.793 EURO
Competence field:Ecology, Sustainable ecosystem management, Sustainability in decision-making processes and accountability structures
Cooperation:Composite partnership
Funding partner:State funding
As a result of human-induced climate change, fires along railway lines in Germany are becoming more frequent, sometimes with severe consequences for railway operations and affected infrastructure. In the absence of systematic studies, little is known about the causes of such fires. As these risks are expected to increase further, there is a need for sound knowledge of the triggers and drivers of fires, as well as methodological expertise in the use of predictive tools.
The objectives of BurnML are
- To build a dataset for training a model, including data on past embankment fires.
- To train a machine-learning model to predict fire risks.
- To pilot the model in collaboration with DB Netz AG.
The project is led by adelphi research gGmbH.
Fires along railway lines in Germany are becoming more frequent, sometimes with severe consequences for railway operations and affected infrastructure.
Map showing the spatial distribution of reported railway embankment fires recorded by Deutsche Bahn (March to October, 2017–2022).
Germany’s railway infrastructure is highly exposed to climate-related hazards. Embankment fires are becoming more frequent as a result of climate change and have significant impacts on rail transport and infrastructure. High-resolution meteorological and land-cover data, particularly information on grassland, help to identify vulnerable areas and take appropriate measures. The findings of the research reveal complex relationships underlying the development of embankment fires. The results of the BurnML project have demonstrated the applicability of the methods used and approaches implemented, including for predicting embankment fires.
Prof. Dr Jan-Peter Mund
Funding partner
Federal funding
