As extreme weather events become more frequent and severe, port and coastal infrastructure owners are looking for better ways to assess and mitigate risks in a changing climate.
When a port facility is built, it is expected to remain operational for many years. However, the rapidly changing climate makes this more difficult to achieve. Weather events once considered ‘100-year’ events are now occurring in 10-year cycles. As a result, historical climate data is no longer reliable.
Despite this, many port and coastal infrastructure designs being developed today still rely on outdated information. Engineers typically analyse past data, find outliers and base their designs on the most extreme historical scenarios. In cases where data does not exist – common for many new port developments – design decisions are often guided by industry benchmarks and prior experience.
Predicting future climatic events for ports is challenging. Traditional methods like Extreme Value Analysis (EVA) assume a stationary climate and rely on downscaled climate models based on emissions scenarios. These models are computationally intensive, often taking months to run, and analyse a limited set of variables due to the effort required to run multiple scenarios.
Waves and storm surges are particularly complex to forecast, requiring additional models for wave generation and propagation. This significantly increases resource demands, making extreme wave projections impractical – even for academic centers with supercomputers, where a single model can take months to process.
A better understanding of risk
At Ausenco, our experts tackle the challenges of marine and coastal engineering. By improving analysis and extreme weather forecasting, we help our clients mitigate risks and implement stronger, more resilient designs.
We searched for a better way to predict climatic events. Reflecting on our experience, we wondered if Non-Stationary Extreme Value Analysis (NEVA) could be applied to better predict extreme surges and wave activity in ports. In basic terms, NEVA is a Bayesian statistical tool where the distribution parameters evolve over time. Our team uses CO2 concentration levels as an explanatory variable for these parameters to predict extreme events, such as wave height or river flow.
In other words, NEVA predicts future extreme events using CO2 levels, and because this data is already readily available, this methodology provides predictions in a shortened timeframe. Moreover, it is a highly efficient statistical tool that avoids the high computational costs of traditional methods and provides flexible, reliable projections. Multiple scenarios can be run simultaneously, in just a matter of minutes or hours at a small fraction of the cost – all with a high level of confidence.
We put it to the test
The Maipo river, located in Chile, is a critical source of irrigation and potable water in the region, making extreme flow events a major concern for residents, government and businesses.
We examined extreme flow events in the river, beginning with traditional EVA methodologies. The analysis showed that recent extreme flows of the Maipo were negatively correlated with CO2 concentrations in the atmosphere. These methodologies assumed that all floods followed the same probability distribution, which was clearly not the case here. This presented an opportunity to apply the NEVA methodology.
To do so, we developed a Bayesian inference model to incorporate expert knowledge about physical processes and the parameters that represent them. This allowed us to assess the distribution of the parameters and their variations over time while analysing different CO2 projections associated with different future emission scenarios. Expert knowledge and high-quality data used to define the parameters were key factors in achieving a high-quality projection.
This approach allowed us to better assess the exposure of the river infrastructure to future risks under a changing climate. Using EVA to produce the same results would have added an additional 6-12 months to the analysis.
A simpler solution to a complex problem
While other methods exist to achieve similar results, the greatest advantage of the NEVA approach is its simplicity and its ability to uncover trends in the observed data. Its speed and cost-effectiveness mean that it can be applied at various stages of project design to provide a clearer risk assessment based on more precise scenarios.
As climate-related extreme events are expected to increase in frequency and unpredictability, we believe that the NEVA approach can add significant value in resilience planning for coastal infrastructure. Using NEVA, our team can help asset owners find a better way to understand and manage these evolving risks.