Sustaining Life:Fighting disaster with math

In search of an answer, Suckale earned a master’s in public administration at the Harvard Kennedy School and then a PhD in geophysics at MIT. She kept looking for inflection points at the intersection of science and governance—places where collaborative science might help a community head off a crisis before it struck.

She made it a policy to always start a new risk modeling project with a very specific problem, and someone in need of a solution. If she’s modeling tree cover, it’s at the request of low-income families who are tired of their communities being passed over for tree-planting grants. As Suckale explains, more conventional environmental assessment models declare that their census tracts already have enough vegetation, while failing to note that most of it is under 2 feet tall.

A model is, of course, broad by design—but that doesn’t mean they can’t be sharpened. If Suckale is modeling climate change’s effect on glaciers, it’s with the Exploratory Ice Preservation Working Group, which is studying ways to slow the melting of glaciers. Doing so at scale could ultimately reduce sea level rise and give coastal communities time to adapt. Or, if she’s modeling coastal flooding, it’s to show how much flooding would increase the number of financially unstable households in the coming decades.

“Thinking we can save time or money, we tend to reuse models for things that they were not intended for,” says Suckale. “Algorithms and models are good at certain things, but if you try to use them for something else, chances are that’s not going to work. They’re still going to give you some sort of answer. But the question is: How valuable is that answer, and what are the assumptions going into it?”

While teaching a class in 2023, Suckale met some officials from East Palo Alto, a low-lying, working-class, majority non-white city. The officials were worried—a plan was afoot to widen a bridge on the San Francisquito Creek and thus reduce the risk of flooding in Palo Alto, a wealthy community upstream. The officials suspected doing so would lead to more flooding in East Palo Alto. But they couldn’t prove it. 

At their disposal was a model for predicting the effect of the bridge widening, called the Hydrologic Engineering Center’s River Analysis System, or HEC-RAS. It had been created before climate change began reshaping the United States. There were signs that something was off. 

Suckale and her team could have created a new model of the creek and run projections on that. But generations of hydrologists have been trained on HEC-RAS; it was written into legislation. Instead, Suckale and her students wrote a different kind of program—an AI assistant that plugged into HEC-RAS and spun it into 50 different futures of the creek’s next 1,000 years.

For most hypothetical floods except for the one used in the planning process, the bridge widening increased flooding to East Palo Alto. The city officials had been right—and now they had proof. 

Suckale enjoys working with public officials who use models like HEC-RAS. 

“What I learned at the U.N. was to really appreciate the constraints they operate under, and what a tough position everyone is in,” she says. “Few public officials specialize in math and neither should they. People are overworked. I have a special place in my heart for public officials who try to do the right thing, and then everyone beats them up.”   

The Joint Powers Authority overseeing the creek project had an explicit goal to distribute risk fairly to everyone along the creek’s path. The group had tried to incorporate sea level rise into its model, but hadn’t included a change in rainfall patterns. Once the mistake was spotted, the group redesigned its plans. 

California is full of infrastructure not designed for its new realities. Eight recent floods in the city of San Jose have all been dam overspills. Suckale has been working on new models to determine when dams need to release water, so the process can begin before heavy rainfall even starts. “That’s why I like working in this space,” she says. “The solutions are quite simple. We can do better.”

Recently, California’s water board asked Suckale to help it use the HEC-RAS add-on to model futures on more creeks and rivers across the state. Doing so means reviving an old project after the students involved have moved on to new things. Still, Suckale said yes; the outside world rarely adheres to academia’s timelines.