Sea-level rise stands to be one of the most serious consequences of climate change, but we don’t yet have great tools for predicting exactly where and when seawater will encroach onto land. Abigail Bodner, a Junior Fellow with the Simons Society of Fellows who recently began a postdoctoral appointment with Laure Zanna at New York University’s Courant Institute, hopes that machine learning can help experts make better predictions of sea-level rise — and perhaps help mitigate the toll that this rise is set to have on coastal communities.
Bodner holds a doctorate in earth sciences and a Master of Science in applied mathematics from Brown University, where she studied models of ocean water flow. She also holds a master’s degree in atmospheric sciences and a bachelor’s in mathematics from Tel Aviv University. We recently spoke about her ongoing work to build tools to better understand the impacts of climate change. Our conversation has been edited for clarity.
Why is it important to model ocean surface flow?
There are many reasons to understand ocean dynamics. Some are very practical: Someone who fishes for a living might need to know when a heat burst within the ocean will affect their shellfish catch, for example, or when an algae bloom will mean that they can’t fish at all for a while. This is like keeping tabs on the weather; if it’s snowing outside, you know to dig out your snow boots.
My focus on ocean flows during my graduate work was somewhat different, and it was motivated by the 2010 Deepwater Horizon oil spill in the Gulf of Mexico. That was a massive disaster, with oil flowing in all directions throughout the Gulf for months after it happened. At the time, we didn’t have a way to predict precisely where that oil would flow, or where oil would be most likely to flow if another spill like that happened in the future. This knowledge gap — which still exists today — makes cleanup even more costly and time-consuming than it already is.
Working with Baylor Fox-Kemper at Brown, we studied how to better model these flows, and published that work in 2019 in the Journal of Fluid Mechanics and in 2020 in the Journal of Advances in Modeling Earth Systems. We also developed methods to better represent these flows in global climate models (currently under review). We believe this work is far-reaching. Ocean surface flows play a major role in atmosphere-ocean interactions, so better modeling these flows is an important way to improve understanding of the overall climate system.
How does modeling ocean flows relate to climate change?
Many people think of the atmosphere with respect to climate change — the terms greenhouse gases and the ozone layer most often come to mind. And yes, understanding the impacts of those forces is important. But studying the ocean in relation to climate change is critical as well; it stores much of the carbon released into the atmosphere, and it can keep that carbon stored for a very long time. A carbon deposit that goes into the ocean today may suddenly be released years from now — thus altering atmospheric carbon levels. So, any kind of climate change prediction needs to take the ocean into account.
There’s a lot we still don’t know about the ocean, especially what’s happening below the surface. How much heat is stored in one part of the ocean, and how much carbon is in another? And when this heat and carbon is eventually released into the atmosphere, how will it affect the climate? The surface flows I modeled during my graduate work can help determine how properties such as heat, momentum and carbon are transferred between the atmosphere and ocean. These are all unknowns right now.
What are you focusing on now at the Courant Institute?
My postdoctoral appointment started in September 2021, so I’ve only just gotten started. My focus will be on using machine learning — basically, training a computer to detect patterns in ocean flows and properties and to make predictions based on that input — to improve our understanding of the impact sea-level rise may have over the next several decades. Such information could help inform mitigation efforts by government officials.
At present, most sea-level rise predictions only resolve down to 100 kilometers, meaning that we can predict the broad likelihood of sea-level rise patterns to within a 100-kilometer area. That’s far too large to help local officials make plans; all of New York City fits within a 100-kilometer grid, and there’s wide variability in the impact of sea-level rise depending on what part of the city you’re in. Our vision is for machine learning models to fill in this gap so that sea-level rise predictions can be made at a much more granular level.
My NYU adviser, Laure Zanna, is a pioneer in using machine learning for climate science — and the method we are using is still very novel. Machine learning techniques can be especially useful for the field today; climate science generates so much data that it would be hard to make sense of it any other way. Our group at NYU uses a variety of modeling and machine learning techniques to advance our knowledge of climate science in ways that are not possible with more traditional tools. I am already learning so much!
What drew you to this area of research, and what do you hope can be achieved?
I’ve always liked being outdoors in nature, but I didn’t always know I would go into environmental work or consider myself an environmental activist. My undergraduate degree was actually in theoretical mathematics, so not environmentally focused.
As time has gone by, it’s become clear that the threat of climate change is real, it is looming, and it plays a critical role in many areas of science. I decided to combine my interest in math with my ability to write computer code toward a better understanding of climate change and its impacts.
It’s a very diverse and interesting field, from people like me who study the physics and theory of how ocean surfaces flow and sea levels rise, to scientists who study different elements of the climate system, such as the chemical composition of ice cores or biomarkers in leaf waxes or identifying storm tracks from satellites. We are each working on our own puzzle pieces that, when fit together, could lead to policies that blunt the worst of climate change’s impact.