Defusing Disaster

Computing Resilience in an Era of Uncertainty

In an era of frequent, powerful storms, fast-spreading wildfires and global pandemics, communities are discovering their vulnerabilities when they can least afford it.

“We need to rethink what it means to be resilient. I use the boxing analogy ‘roll with the punches’: the ability to absorb the shocks of extreme events and recover quickly,” says Michel Boufadel, the director of NJIT’s Center for Natural Resources. “But to do so, the whole system needs to work together. It doesn’t matter if the power stays on, but 90% of the roads are closed.”

With collaborators at Rutgers and Princeton, Boufadel has developed a Community Intrinsic Resilience Index (CIRI) that will help cities and regions reduce the potential impacts of natural disasters by making strategic investments in infrastructure. The researchers assess resilience in four key sectors — transportation, energy, health care and socio-economics — they deem important to remaining productive and quickly returning to normal.

The team evaluates disaster-relevant factors in each sector and assigns values to them. In transportation, the amount of roadway per square mile and population, and the availability of public transit are some of the elements; in health care, the number of hospital beds, doctors and nurses per patient population and the percentage of residents with health insurance; in energy, the amount of underground wires and the availability of backup power. By analyzing socio-economic demographics, they can project the number of jobs potentially affected.

“We set thresholds for each sector. Going under one could potentially shut down a community,” notes Firas Gerges, a former Ph.D. student of Boufadel’s who is now a postdoctoral researcher in computer science at Princeton. Hani Nassif, director of the Rutgers Infrastructure Monitoring and Evaluation Group, is another collaborator.

The group recently computed CIRI scores for counties in New Jersey, which ranged between 63% and 80%, based on preliminary data. In the study, a post-disaster CIRI calculation following projected major flooding revealed that the transportation and socio-economic attributes of two coastal counties would fall below specified thresholds due to projected road closures and harm to local economies.

Their goal, Boufadel says, is to help local leaders and other policy makers integrate resilience within the planning and design phases of disaster management. “There is not enough money to avoid harm entirely, and decision-makers need numbers in order to prioritize spending,” he notes. “Should they spend money protecting the golf course or the high school?”

The team is currently collecting more energy data from local, nonprofit and government stakeholders so they can better assess energy resilience throughout the state and identify communities in need of both improved capacity and rapidly deployable energy supplies. Suggested solutions include mobile energy storage and megawatt-scale batteries that could be charged off-peak and placed where needed, such as close to commuter lines during weekdays.

“This is a crucial step to tackle energy budget deficits and attain energy equity, particularly in underserved communities within New Jersey,” Boufadel says.

Because natural disasters transcend political boundaries, the group has developed measures to assess resilience on the census tract level. Gerges explains: “We can calculate CIRI census block by census block and compile that to see how a region of a state would perform.”

They are currently developing and validating a new model that combines social and engineering concepts to measure the resilience of areas and infrastructures under different degrees of stress and at different temporal scales.

“We want to be able to say what’s resilient for a 10-year storm versus a 50-year storm,” Gerges says. “Using machine learning, we can predict climate variables decades into the future in New Jersey, such as the amount of precipitation in the Hackensack-Passaic Watershed, wind speed, temperature and solar irradiance. The latter will allow us to locate the best spots for solar farms under different climate scenarios.”

Uncharted Territory: 
The Fate of Forests Amid Climate Change

The world’s forests are at the front lines of the climate crisis, absorbing nearly 30% of carbon emitted globally each year. But how are these vital ecosystems responding as climate change-driven wildfires and droughts intensify?

Nearly a decade ago, a bark beetle infestation tore through Southeast Wyoming, transforming the lush landscape of Medicine Bow National Forest into a tinderbox of dead lodgepole pine. In September 2020 it ignited, and what became known as the Mullen Fire raged beyond the parkland across 176,000 acres over the next month, fueled by the decimated trees and unusually dry conditions.

Xiaonan Tai, an assistant professor of biological sciences and director of NJIT’s Ecohydrology Lab, is now investigating the fate of the national forest. She has been developing models used to unravel the complex ecological and hydrological processes taking place in North American forests amid historic climate changes.

“Groundwater flow has been missing from ecosystem modeling and future projections, largely because it is difficult to directly observe, and it is computationally challenging to solve … But we have better capability now.”

The Mullen Fire touched off in the midst of a 20-year climate trend in the Rocky Mountains — its high-elevation forests are experiencing reduced snowfall and greater wildfires than at any point in the past 2,000 years.

“The disaster reflects a clear trend in increasing fires across Western U.S. forests, but the big question now is how these forests recover,” says Tai. “The response of Medicine Bow’s forests could give us a window into how the region’s forests respond to other climate change-related disturbances going forward.”

Tai, who three years ago studied the national park’s bark beetle epidemic, is returning to collaborate with University of Wyoming researchers through a U.S. Department of Energy grant. The team is conducting a full workup of the land that includes the collection of microclimate, vegetation and hydrological measurements over the next three years.

Tai’s ecohydrology modeling is incorporating the field data to paint a picture of the interaction between Medicine Bow’s groundwater and vegetation health, answering questions about how the area’s subsurface water shapes forest response and recovery from the fire.

It’s a tricky equation to calculate.

“The park’s wetlands are where the land’s groundwater tends to accumulate most, so we hypothesize that these are likely sites during a fire that are critical to the forest’s response,” says Tai. “A big challenge is that this is an enormous groundwater system where water can transport and affect the fate of trees from thousands of kilometers away. Rainwater doesn’t remain local, but travels based on a mix of regional factors such as topography and substrate properties.”

In the journal “Environmental Research Letters,” Tai recently established a new state-of-the-art ecohydrological model to quantify the impacts of subsurface groundwater on forest mortality risk for the first time. The NSF-funded study highlighted a key, unseen mechanism affecting how forests respond to drought and elevated CO2 concentrations — the way in which water flows laterally beneath the forest floor.

“Past studies focused on solving water fluxes only in the vertical dimension, ignoring the role of lateral subsurface water exchanges. But with this model, we get more complete descriptions of ecosystem water dynamics than ever before,” she says.

Tai says the findings have challenged a previously held mainstream expectation by climate scientists — that elevated CO2 levels ameliorate drought stress experienced by trees.

“We showed that most water stored doesn’t stay local, but rather, begins to travel laterally to areas that may otherwise be dry. … It shows we need to reassess future forest predictions by incorporating subsurface flow.”

As Tai’s modeling charts Medicine Bow’s forest-hydrology connections, ecologists could gain a vital roadmap of where and how groundwater influences forest recovery regionally.

“There are similarly impacted forests throughout the Rocky Mountains. It’s our hope that our results will inform effective post-disturbance land management strategies beyond Medicine Bow,” says Tai.

Robot Swarms to the Rescue

When the Champlain Towers condominium collapsed in Surfside, Fla. in 2021, more than 100 residents were trapped under debris so unstable that rescuers were hampered from entering it immediately.

“A common source of danger in disasters is the risk of rubble shifting, endangering both emergency responders and the people they’re trying to rescue. This calls for great caution, slowing efforts and potentially increasing the death toll,” says Petras Swissler, a robotics engineer who joined the NJIT faculty last fall.

Swissler envisions a far more efficient rescue: a rapid-response team composed of hundreds or even thousands of self-assembling robots swarming into the disaster area and using their own bodies to form ramps, support columns and bridges. In the case of a collapsing bridge, teams of robots would serve as both the scaffolding used in construction and the repair force.

What distinguishes these future robots from current models is their autonomy to determine what structures are needed in a particular environment, amid shifting loads and emerging operational needs. They’re given high-level tasks, but no blueprints.

“The robots themselves would decide what structures to build and how best to build them by using their sensors to determine environmental conditions and then converge on an optimal structure based on inter-robot communication,” explains Swissler, who is co-developing a hardware-algorithm system to operate them.

A primary challenge, he notes, is enabling robots to attach to each other, while also giving them the ability to move about a structure during construction.

The robots he’s currently building look like three black tennis balls arranged in a triangle. A novel attachment mechanism called “continuous dock” that Swissler developed as a Ph.D. student at Northwestern University allows the robots to attach to each other wherever they come into contact. Microchips in each ball drive motors that enable them to flip about one another, attaching and detaching with each step. The connections are strong enough to hold the weight of around 60 robots.

Prior work on self-assembling robotic systems has generally focused on the construction of latticed, blueprint-based shapes that include detailed instructions, thus limiting the extent to which robots can adapt structures to conditions. They have other built-in vulnerabilities. When building a bridge from both sides of a canyon, for example, it is difficult to achieve the required perfect alignment.

Swissler says he’s looking to nature for “messy” construction designs that adapt spontaneously to the shape of the environment they’re in, analogous to the process by which certain species of ants form bridges and rafts. He traveled last spring to an island in the Panama Canal with members of Simon Garnier’s NJIT Swarm Lab to study nomadic army ants that search out new hunting grounds day to day. They form clusters of hundreds of thousands of ants called bivouacs, instead of the more typical underground ant nests.

“With ants, cells and bees, they’re more or less just grabbing onto neighbors arbitrarily, while robots tend to attach at very specific locations on each other’s bodies and with very specific intent,” says Swissler, who is exploring radically different robot designs. “By designing new types of algorithms along with the robot hardware, we’re aiming for what we believe is realistically achievable with current robot technology.”

The robots and algorithms would be particularly useful in resource-constrained and remote environments, such as Antarctica or outer space, where some would form and reform on-demand structures, while others would form tools, such as wrenches, hammers and screwdrivers.

   “In Apollo 13, where there was a mismatch between the shape of the CO2 scrubbers in the command module and the lander, NASA had to scramble to improvise that connection. Future swarms of these robots could have autonomously analyzed the problem and self-assembled into an adapter,” he says, adding, “When you don’t have what you need, when you need it, there’s a cost associated with that. Sometimes that cost is monetary, but in disasters, it’s also potentially human lives.”