The Massachusetts Institute of Technology on Monday announced five multi-year pilot projects in its first-ever Great Climate Challenge, a new initiative to tackle complex climate problems and deliver breakthrough solutions to the world as quickly as possible. This article is the first in a five-part series that highlights the most promising concepts that have come out of competition, and the multidisciplinary research teams behind them.
With improvements in computer processing power and an increased understanding of the physical equations that govern Earth’s climate, scientists are constantly working to improve climate models and improve their predictive power. But the tools they’re refining were originally designed decades ago with only scientists in mind. When it comes to developing concrete climate action plans, these models remain a mystery to the policy makers, public safety officials, civil engineers, and community regulators who desperately need their predictive vision.
Noelle Selin, professor in the Institute for Data, Systems and Society and the Department of Earth, Atmospheric and Planetary Sciences (EAPS), and co-lead with Professor Raphael Ferrari on the MIT Climate Grand Challenges pilot project “Bringing Account to the Climate Challenge.” “How can we use new computational techniques, new understanding, new ways of thinking about modeling, to really bridge that gap between the latest scientific and modeling advances, and the people who really need to use these models?”
Using this as a leadership question, the team will not only attempt to improve existing climate models, but build a new one from the ground up.
This kind of game-changing progress is exactly what the major MIT Climate Institute is looking for, which is why the proposal was chosen as one of the five key projects in the ambitious institute-wide program aimed at tackling the climate crisis. The proposal, which was selected from among 100 submissions and was among 27 finalists, will receive additional funding and support to further their goal of reimagining the climate modeling system. It also brings together contributors from across the institute, including the MIT Schwarzman School of Computing, the School of Engineering, and the Sloan School of Management.
When it comes to the pursuit of high-impact climate solutions that communities around the world can use, “it’s great to do that at MIT,” says Ferrari, EAPS Cecil, and Ida Green professor of oceanography. “You won’t find many places in the world where the cutting-edge climate science, cutting-edge computer science, and cutting-edge policy science experts we need to work together are available.”
Climate model of the future
The proposal builds on work Ferrari began three years ago as part of a joint project with the California Institute of Technology, the Naval Postgraduate School, and NASA’s Jet Propulsion Laboratory. Called the Climate Modeling Alliance (CliMA), the Consortium of Scientists, Engineers and Applied Mathematics is building a climate model that is able to more accurately predict future changes in critical variables, such as clouds in the atmosphere and turbulence in the ocean, with uncertainties at least half the size of those in current models.
To do so, however, requires a new approach. For one thing, current models are too coarse in terms of accuracy — on the scale of 100 to 200 kilometers — to solve for small processes like cloud cover, precipitation, and sea ice extent. But also, Ferrari explains, part of this limitation in accuracy is due to the basic architecture of the models themselves. The languages in which most global climate models are coded were first created in the 1960s and 1970s, largely by scientists for scientists. Since then, advances in the worlds of corporate computing and computer games have led to the emergence of new dynamic computer languages, powerful graphics processing units, and machine learning.
For climate models to take full advantage of these developments, there is only one option: to start over with modern, more flexible language. Written in Julia, part of Julialab’s machine learning technology, and led by Alan Edelman, a professor of applied mathematics in the MIT Department of Mathematics, CliMA will be able to harness far more data than current models can handle.
“It’s finally been fun working with people in computer science here at MIT,” Ferrari says. “Previously it was impossible, because traditional climate models are written in a language that their students cannot read.”
The result is the so-called “Earth’s digital twin,” a climate model that can simulate global conditions on a large scale. That in itself is an impressive feat, but the team wants to take this a step further with their proposal.
“We want to take this large-scale model and create what we call a ‘simulator’ that only predicts a set of variables of interest, but is trained on a large-scale model,” Ferrari explains. Simulators aren’t a new technology, but what’s new is that these simulators, referred to as “digital Earth cousins,” will take advantage of machine learning.
“We now know how to train a model if we have enough data to train them on,” Ferrari says. Machine learning for projects like this has only become possible in recent years as more monitoring data becomes available, along with improved computer processing power. The goal is to create smaller, more localized models by training them with Earth’s digital twin. Doing so will save time and money, which is key if digital cousins are able to tap into stakeholders, such as local governments and private developers.
Adaptable forecasts for ordinary stakeholders
When it comes to making informed climate policy, stakeholders need to understand the likelihood of an outcome within their regions — the same way you would prepare for a different altitude if there was a 10 percent chance of rain versus 90 percent. Earth’s smaller digital cousin models will be able to do things the larger model can’t, such as simulating local areas in real time and providing a wider range of probabilistic scenarios.
“Currently, if you want to use an output from a global climate model, you usually have to use outputs designed for public use,” says Celine, who is also director of the Policy and Technology Program at MIT. With the project, the team can take into account the needs of the end user from the start while incorporating their observations and suggestions into the models, helping to “democratize the idea of running these climate models,” as she puts it. Doing so means building an interactive interface that will eventually give users the ability to change input values and run new simulations in real time. Ultimately, the team hopes, Digital Earth Cousins will work on something as ubiquitous as a smartphone, although such developments are currently outside the scope of the project.
The next thing the team will work on is building relationships with stakeholders. By partnering with other MIT groups, such as the Joint Program on Global Change Science and Policy and the Climate and Sustainability Consortium, they hope to work closely with policymakers, public safety officials and urban planners to give them predictive tools tailored to them. Needs that can provide practical and important outputs for planning. In the face of rising sea levels, for example, coastal cities can better visualize the threat and make informed decisions about infrastructure development and disaster preparedness; Communities in drought-prone areas can develop long-term urban planning with a focus on water conservation and forest fire resistance.
“We want to make the modeling and analysis process faster so that people can get more direct and helpful feedback for near-term decisions,” she says.
The final part of the challenge is to motivate the students now so they can join the project and make a difference. Ferrari has already had a lot of luck getting students interested after co-teaching a class with Edelman and seeing students’ enthusiasm about computer science and climate solutions.
“With this project, we intend to build a climate model for the future,” says Celine. “So it seems really appropriate that we also train the builders of this climate model.”