Application of graph neural networks to forecast urban flood events: the case study of the 2013 flood of the Bow River, Calgary, Canada

ABSTRACT

Mitigating the harmful effects of flooding is essential in the current climate change and urban growth scenario, where these natural phenomena are expected to occur more often. Addressing this problem, we present a novel graph-based forecasting model for predicting urban flooding and showcase its application in the Bow River in the City of Calgary, Alberta, Canada. The proposed model is based on graph theory and deep learning paradigms and was used to forecast flooding occurrences up to 24 h ahead. Our new approach developed the SAGE algorithm, a hybrid learning and planning approach that accurately predicted floodings for various analyzed forecasting horizons, increasing the forecasting performance by up to 44% compared to the persistence model. The SAGE model also returned superior or competitive results compared to other models in the literature. The superiority of the SAGE model was highlighted for longer forecast horizons of 24 h ahead, where it reached a maximum improvement of 79%, suggesting its superiority in capturing spatiotemporal information in the dataset. These results indicate that the SAGE approach is a cutting-edge tool for flooding forecasting, and it could be used to develop early flood warning systems to reduce potential flooding impacts.

KEYWORDS:

• Flooding
• Bow River
• forecasting
• machine learning
• graph neural networks
• deep learning

Disclosure statement

No potential conflict of interest was reported by the author(s).

Data availability statement

The Bow River flow rate data used in this study were provided by the Government of Alberta Environment and Parks (Government of Alberta Environment and Parks 2016). The code used in our work can be found at the following GitHub repository https://github.com/victor0s/Bow-River/tree/master.

Additional information

Funding

This research was funded by the Natural Sciences and Engineering Research Council of Canada (NSERC) Alliance [grant number: 401643], in association with Lakes Environmental Software Inc., and by the Conselho Nacional de Desenvolvimento Científico e Tecnológico—Brasil (CNPq) [grant number: 303585/2022-6].

Notes on contributors

Paulo Alexandre Costa Rocha

Paulo Alexandre Costa Rocha has a doctorate in Civil Engineering from the Federal University of Ceará. He completed a postdoctoral fellowship at the University of California San Diego through CAPES' “Visiting Professor Abroad Program” from 2019 to 2020. He is currently doing his second postdoctoral fellowship on a joint project between the Universities of Guelph and Waterloo, Canada. Dr. Costa Rocha has experience in the area of Fluid Mechanics and Solar Thermal Energy, with emphasis on Renewable Energy Systems. Dr. Costa Rocha is currently an Associate Professor IV of the Department of Mechanical Engineering at the Federal University of Ceará. He acts as a reviewer of several high impact factor journals, and is a Guest Editor of the Atmosphere Journal (MDPI) for the Special Volume “Solar Irradiance and Wind Forecasting”.

Victor Oliveira Santos

Victor Oliveira Santos has a bachelor degree in Mechanical Engineering from the Federal University of Ceara, and a master's degree on Mechanical Engineering from the Federaul University of Ceara (renewable energy). Currently, he is Ph.D. candidate at University of Guelph. Has experience on Machine Learning and Deep Learning applied to time-series problems. Currently researching Quantum Computing applications to Machine Learning.

John Scott

John Scott is an undergraduate student in his final year of study at Queen's University. His research is focused on application of machine learning paradigms to spatiotemporal environmental modeling and prediction.

Jesse Van Griensven Thé

Jesse Van Griensven Thé, Ph.D., P.Eng., is a distinguished academic and engineer with a prolific career spanning over two decades. Holding a Ph.D. in Mechanical Engineering from the University of Waterloo, his expertise in Computational Fluid Dynamics has led to significant contributions in environmental management systems. He is an Adjunct Professor at both the University of Waterloo and the University of Guelph while also serving as Executive Editor for the ENERGY Journal. He is the founding president of Lakes Environmental Consultants Inc. His work in emissions inventory systems and air dispersion modeling packages has garnered international recognition. His innovative research and leadership have earned him numerous awards, patents, and the title of CEO of one of the “25 Best Risk and Compliance IT Companies in the World” by CIO Magazine.

Bahram Gharabaghi

Bahram Gharabaghi is a professor of engineering in the School of Engineering at the University of Guelph in Guelph, Ontario, Canada. His research has primarily been focused on watershed scale water quality models to improve the accuracy of calculations of these management tools for protection of source waters from further degradation and to develop effective strategies for improvement of the quality of impaired water bodies in Ontario. He has more than twenty years of experience in research and development in hydrologic modelling and non-point source pollution control in Canada and internationally. He is currently leading several research projects in collaboration with Conservation Authorities, Ontario Ministry of Environment and Climate Change, Ministry of Transportation, Ministry of Natural Resources, and Ministry of Agriculture, Food and Rural Affairs.