Predictive modeling of severe weather impact on individuals and populations using Machine Learning

Identificador:  imarina:9368063

Handle: https://hdl.handle.net/20.500.11797/imarina9368063

Autores:  Iglesias, Jordi; Cuesta, Ildefonso; Saluena, Clara; Sole, Jordi; Prevatt, David O; Fabregat, Alexandre

Resumen:
In this work, Machine Learning (ML) techniques are used to develop tools capable of accurately predicting the impact of severe weather events. We use readily accessible predictors, including daily meteorological data, basic demographics, geographic and terrain features, along with the number of daily meteorological incidents reported to the emergency services. The model was built using disaggregated data from the 947 municipalities from the region of Catalonia between January 1, 2015 and June 30, 2021. Catalonia's region is situated in the northeastern part of Spain along the Mediterranean Basin, and frequently experiences storms with strong winds and intense rainfall. In 2020, such weather events resulted in 64 injuries and damages amounting to approximately 70 million USD. The ML-based model presented in this study shows a predictive capacity for extreme weather risk superior to that of the daily meteorological warning system of the Meteorological Service of Catalonia. In addition, the model has been used to estimate how urbanization modifies the extreme weather impact. Given the expected increase in the intensity, frequency, and duration of extreme weather events in the context of global warming, the methodology presented in this work could be helpful in developing tools to assist emergency service managers and policy makers in making rapid and effective decisions.