Predicting Coastal Flooding in the Mediterranean with Remote Sensing and Machine Learning

Due to its historically low tidal variations, the Mediterranean Sea basin has seen significant coastal urbanisation, exemplified in the densely populated Italian region Liguria. However, the region faces increased vulnerability to extreme sea level changes and coastal flooding due to potential climate change-induced storminess.

Machine learning has recently received increased attention in the literature as regards the ability of data-driven approaches to solve flood-related problems, including the identification of areas potentially susceptible to inundation in support of risk preparedness and resilience in coastal cities. This work explores the application of machine learning using widely available remote sensing datasets to predict the inundation extent for a modelled 100-year return period coastal flooding event in Liguria. Numerical simulations produced by local administrations in the context of the EU Floods Directive serve as ground truth due to the absence of post-event inundation maps. Various pre-processed remote sensing datasets are employed as predictors, including land cover data, spectral indices and high-resolution DEM.

The results highlight challenges in integrating diverse timescales and data types and can be used to assess the influence of predictors on coastal resilience. The study also addresses the benefits and drawbacks of different machine learning algorithms in evaluating coastal resilience within this approach.

-------------------------------------------------------------------------------------------------------------------------------------

References

Woznicki, S. A., Baynes, J., Panlasigui, S., Mehaffey, M. and Anne Neale. “Development of a spatially complete floodplain map of the conterminous United States using random forest.” Science of the Total Environment 647 (2019): 942-953.

Ireland, Gareth, Michele Volpi, and George P. Petropoulos. "Examining the capability of supervised machine learning classifiers in extracting flooded areas from Landsat TM imagery: a case study from a Mediterranean flood." Remote sensing 7.3 (2015): 3372-3399.

Fogarin, S., et al. "Combining remote sensing analysis with machine learning to evaluate short-term coastal evolution trend in the shoreline of Venice." Science of The Total Environment 859 (2023): 160293.

Tsiakos, Chrysovalantis-Antonios D., and Christos Chalkias. "Use of Machine Learning and Remote Sensing Techniques for Shoreline Monitoring: A Review of Recent Literature." Applied Sciences 13.5 (2023): 3268.

Reichstein, M., Camps-Valls, G., Stevens, B., Jung, M., Denzler, J., Carvalhais, N., & Prabhat, F. (2019). Deep learning and process understanding for data-driven Earth system science. Nature, 566(7743), 195-204.

Camps-Valls, Gustau, et al., eds. Deep learning for the Earth Sciences: A comprehensive approach to remote sensing, climate science and geosciences. John Wiley & Sons, 2021.