Coastal regions worldwide face escalating flood hazards, with hurricanes causing particularly devastating storm surges. The New York Metropolitan area, covering the coastlines of both New York and New Jersey, is highly vulnerable to these hazards. As climate change intensifies, accurately predicting storm surge levels becomes crucial for effective long-term development planning and implementing robust risk mitigation strategies.
To overcome the computational challenges associated with simulating numerous synthetic storms under specific climate change scenarios, researchers have leveraged the power of machine learning. A recent article published in the journal Nature Communications assessed the impact of climate change on hurricane storm surge hazards in the New York and New Jersey coastlines by developing a machine learning-based predictive model.
Study: Riding the Storm: Machine Learning Unveils Climate Change's Impact on New York/New Jersey Coastlines. Image credit: TippaPatt /Shutterstock
Methodology
To generate a comprehensive training dataset, the researchers employed a high-fidelity hydrodynamic model that incorporates the effects of wind-generated waves. The advanced circulation and simulating waves nearshore (ADCIRC + SWAN) coupled model simulated thousands of synthetic tropical cyclones (TCs). Various TC parameters were considered, including maximum sustained wind speed, latitudinal and longitudinal distances, and the minimum distance between the study site and the TC eye. This rich dataset enabled the machine learning model to capture the complex interactions between storm characteristics and resulting storm surge levels.
The machine learning model underwent training, validation, and testing using the training dataset generated from the ADCIRC + SWAN model. Researchers utilized machine learning algorithms, specifically the Adaptive Boost (AdaBoost) algorithm with support vector regressor (SVR) as the base estimator. Hyperparameters were carefully tuned using a cross-validation grid search to ensure optimal model performance. The dataset was divided into training, validation, and testing subsets, allowing for an accurate assessment of the model's predictive capabilities.
Results and discussion
The study's findings indicate a projected increase in storm surge levels at the southern edge of New Jersey's coast within specific bays, such as Jamaica, Raritan, and Sandy Hook. This increase is attributed to the combined impact of climate change and potential shifts in hurricane tracks. By analyzing historical and future storm surge levels, researchers identified spatial variations and assessed the potential risks posed by climate change.
Additionally, the study emphasizes the significance of considering the effect of wind-generated waves on storm surge heights and return periods. Incorporating these factors into the predictions enables a more accurate assessment of future storm surge scenarios. This holistic approach provides valuable insights into the potential consequences of climate change on storm surge hazards.
Future flood levels
Accurately predicting low-probability, high-consequence flood levels is crucial for effective risk management and resilient planning. To address this, the machine learning model was employed to forecast flood levels for the 100- and 500-year return periods. The analysis reveals that climate change would increase flood levels along the southern part of the New Jersey coastline. Specifically, the 100-year return period exhibited an up to 4% increase, while the 500-year return period experienced a 12% increase. These projected changes highlight the heightened risks climate change poses and emphasize the need for proactive adaptation strategies.
Comparison with previous studies
The findings of this study align with previous research conducted by Garner et al. and Marsooli et al. Garner et al. predicted a negative change in storm surge height return periods at the Battery, consistent with the results of the current study. However, slight variations in percentage changes can be attributed to differences in data sets, domains, and the inclusion of wave effects. Marsooli et al. also found an average 6% increase in the 100-year return period along the southern part of the New Jersey coastline, further supporting the impact of climate change on storm surge hazards.
Conclusion
This research demonstrates the effectiveness of a machine learning-based predictive model in assessing the impact of climate change on hurricane storm surge hazards. The study provides valuable insights into future flood levels along the New York and New Jersey coastlines by integrating high-fidelity hydrodynamic models with machine learning techniques. The findings underscore the need for proactive planning and adaptation strategies to mitigate the increasing risks posed by climate change.
As the challenges of climate change continue, it is essential to prioritize proactive measures to protect coastal communities. Policymakers, urban planners, and infrastructure developers can leverage the insights provided by machine learning models to implement effective strategies that mitigate the escalating risks associated with storm surge hazards. By accurately predicting storm surge levels and understanding their potential changes due to climate change, we can work towards resilient coastal communities and better adapt to the evolving risks we face in the future.
In summary, this research highlights the importance of evaluating the impact of climate change on storm surge hazards using machine learning. By harnessing the power of advanced modeling techniques, decision-makers can make informed choices and take proactive measures to protect vulnerable coastal regions. The integration of machine learning in risk-informed coastal planning and development ensures a sustainable future in the face of climate change. By embracing predictive models and leveraging the insights they provide, we can strive for a resilient coastal landscape that is prepared to withstand the challenges of a changing climate.
Machine Learning Models Predict Arsenic Contamination