The Most Cost-Effective Approaches to Protect Against Rising Seas Are Flexible, Adaptable

Seagrass plantings on the Atlantic Ocean
Scientists are analyzing approaches to decision-making on preserving coastal infrastructure over time to find optimal solutions.
Matthew Drews

In an artificial intelligence study, Rutgers and Princeton scientists conclude that solutions with built-in flexibility are superior to one-shot tactics


Public officials designing seawalls, levees and other safeguards against rising seas can save money if their solutions are flexible, adapting to sea-level increases over time, according to an analysis by scientists at Rutgers and Princeton universities.

Such an approach is superior to one where planners and engineers produce a single fix meant to last, they said.

Employing an analysis powered by an artificial intelligence (AI) technique known as reinforcement learning, researchers modeled the process of making decisions on preserving coastal infrastructure over time – steps that ultimately will be undertaken by humans, not computers. Conclusions from their study published in the Proceedings of the National Academy of Sciences bear directly on future efforts by planners and engineers working in places such as New York City and coastal New Jersey to combat flooding, they said.

The analysis is designed to assist public officials faced with the dilemma of taking actions – ones requiring expensive, long-term investments – without knowing the exact extent of increase in tidal levels brought about by climate change.

 “If you think there’s some chance, as we do, of a very high-end, sea-level rise and your response is to act only once, you have to decide how much weight to give that outcome,” said Robert Kopp, a Distinguished Professor in the Department of Earth and Planetary Sciences at Rutgers School of Arts and Sciences and an author of the study. “If you don’t give sufficient emphasis to it, you run the risk that you are under-adapted and will have very large damages. But if you over accentuate the importance of that outcome, you may have over-adapted, and you will have spent too much money.”

Dune grass built on sand dunes facing the Atlantic Ocean

Beachgrass plantings, such as this plot in the dunes of Seaside Park, is a strategy New Jersey Shore communities are employing to mitigate beach erosion. Beachgrass stabilizes sand dunes and strengthens coastal resilience to storms.
Matthew Drews

The researchers analyzed extensive data on flooding, which has caused increasing damage along the coastal U.S. and throughout the world.

“Defenses are being built to protect coastal regions for the next few decades or longer,” said Ning Lin, a professor of civil and environmental engineering at Princeton and an author of the study. “However, climate projects are largely uncertain over long time horizons.”

To deal with this unpredictability, Ling said planners must be flexible and ready to adapt their plans to future observation of climate conditions. Although this can be extremely challenging because of the complexity of climate science, harnessing advances in data science will provide effective solutions, she added.

The difficulty for public officials in confronting these challenges is enhanced by the large number of variables that are likely to shift in unforeseen ways.

“Flexible approaches, like those we model with reinforcement learning, allow planners to protect against high-end rise without incurring excessive costs,” Kopp said.

Flexible approaches, like those we model with reinforcement learning, allow planners to protect against high-end rise without incurring excessive costs

Robert Kopp

Distinguished Professor, Department of Earth and Planetary Sciences

In the study, the researchers simulated efforts to defend Manhattan against sea-level rise through the end of the century. The goal was to determine whether any decision-making process that systematically incorporates observations and updates would prove superior to others over such a long period of time. 

To do this, the researchers simulated decisions by city planners in 10-year intervals up to the year 2100. The researchers compared various hypothetical efforts, including a dynamic approach that added to a seawall’s height over time based on interpretations of new data, and a static approach where the height of a seawall constructed was based on historic 100-year flood projections.

They found that a dynamic approach costs less than other methods, and more effectively reduced the risk of very bad outcomes.

Reinforcement learning is a type of machine learning in which a computer software program makes decisions and receives positive reinforcement based on results. Designers train the program by running it through vast amounts of simulated decisions, enabling it to learn by trial and error rather than through explicit instructions from programmers.

A berm of rocks facing the Delaware Bay

Scientists at the Rutgers Cape Shore Laboratory have installed a rock berm to protect structures from encroaching waters. Berms are an example of adaptation to preserve shoreline.
Matthew Drews

Defending Manhattan and other coastal areas from the onslaught of storms and rising sea levels is not only complex, it requires making difficult decisions under uncertain conditions, the researchers said. For each time interval in the study, planners made decisions based on observed sea-level rise as well as roughly 80,000 differing scenarios of future sea-level rise and associated decisions made in response.

While climate adaptation decisions are not simple, reinforcement learning is a highly efficient system for incorporating observations and updating plans to derive optimal solutions for limiting impacts from extreme events, the researchers said.

“The analysis of New York City’s situation is by no means unique,” said Michael Oppenheimer, a professor of geosciences and international affairs at Princeton and an author of the study. “The method can be applied widely, although its benefit compared to other systems of analysis would vary from place to place.”

The authors are part of the Megalopolitan Coastal Transformation Hub, a Rutgers-led, National Science Foundation-funded consortium of research institutions working to advance the science of how coastal climate hazards, landforms and human decisions interact to shape climate risk and to advance climate adaptation in the New York City, New Jersey and Philadelphia regions.

Explore more of the ways Rutgers research is shaping the future.