Climate change impact on hurricane storm surge hazards in New York/New Jersey Coastlines using machine-learning
Mahmoud Ayyad, Muhammad R. Hajj, Reza Marsooli
Abstract
Abstract Recent hurricane losses in the New York Metropolitan area demonstrate its vulnerability to flood hazards. Long-term development and planning require predictions of low-probability high-consequence storm surge levels that account for climate change impacts. This requires simulating thousands of synthetic storms under a specific climate change scenario which requires high computational power. To alleviate this burden, we developed a machine learning-based predictive model. The training data set was generated using a high-fidelity hydrodynamic model including the effect of wind-generated waves. The machine learning model is then used to predict and compare storm surges over historical (1980–2000) and future (2080–2100) periods, considering the Representative Concentration Pathway 8.5 scenario. Our analysis encompassed 57 locations along the New York and New Jersey coastlines. The results show an increase along the southern coastline of New Jersey and inside Jamaica, Raritan, and Sandy Hook bays, while a decrease along the Long Island coastline and inland bays.