Bellwether and Kansas City
Severe weather is becoming increasingly unpredictable, but the traditional, manual methods used to assess its impact remain slow and resource-intensive. City officials need faster ways to evaluate structural damage so they can immediately deploy relief and secure vital federal and state aid.
Using Bellwether’s AI-powered geospatial tools to analyze drone imagery, Kansas City’s emergency teams calculated neighborhood damage with 96% accuracy in just 20 minutes, demonstrating technology’s potential to accelerate and augment emergency response.
The Data Delay in Disaster Recovery
After a cataclysmic storm strikes the Kansas City metropolitan area, city officials immediately begin the critical work of evaluating its impact and mobilizing a response effort—but legacy workflows hinder their ability to act quickly.
Historically, volunteers would visit affected neighborhoods with pen, paper, and a camera to gauge the severity of the destruction. They would record their observations, take photos, and return their findings to the city’s Emergency Management Team to sift through manually.
While Kansas City is working to digitize its system so volunteers can upload field reports to an online platform, analyzing that influx of data and allocating assistance remains a time-consuming effort. Every minute spent surveying a disaster’s aftermath is a minute lost to sending relief.
“We can forecast the weather, but we can’t predict how severe it’s going to be,” says Andrew Ngui, Kansas City’s Chief Digital Officer. “That’s only going to increase. We had tennis-ball-sized hail recently that wasn’t in the weather forecast.”
Not only does efficient damage assessment lead to faster response on the ground, Ngui explains, but it also makes it easier for the city to secure an official “disaster designation”: the threshold required to receive much-needed aid from FEMA and state agencies.
“My work with the city is focused on identifying problems and exploring the role technology might play in helping to address them,” Ngui says. “I started thinking: ‘How can we get in front of this?’"
A Kansas City emergency responder reviews drone footage
From Pen and Paper to a Prediction Engine
Ngui and his team recently partnered with Bellwether, X’s moonshot to understand and anticipate changes across our planet, to explore how its technology could accelerate Kansas City’s ability to evaluate the impact of severe weather events.
Designed to function as the first prediction engine for the Earth and everything on it, Bellwether's advanced geospatial analysis tools use AI to more deeply understand the planet’s behavior, from mapping wildfire risks to improving weather forecasting. The team trains Bellwether’s models on massive Earth observation datasets, enabling the system to process incoming visual data and identify granular changes to the physical world in seconds.
“We’re building our technology with the mission of helping cities and people be more resilient,” says Bellwether Operations Manager Josh Jeffery. “If they know something’s coming, can they build their infrastructure better?”
In a disaster simulation exercise, Kansas City’s Emergency Management Team designated random homes in a neighborhood as “damaged” and then used AI imagery tools to make pictures of these houses look destroyed. The Kansas City fire department flew a single drone across a designated area at 200 and 400 feet to capture overhead imagery of 120 homes, replaced some images with the doctored photos, and then uploaded the pictures to Bellwether.
Bellwether’s AI-powered geospatial tools immediately went to work to identify structural damage. At the same time, 60 volunteers set out on foot for several hours to survey the area manually, chronicling their findings on both paper forms and in the new digital platform.
The drones mapped the entire neighborhood from above in 25 minutes, applying accurate GPS coordinates to each image they took. In the span of 20 minutes, Bellwether’s technology then analyzed footage of the entire subdivision and calculated damage within 96% accuracy of its boots-on-the-ground volunteer counterparts.
“Technology does some things very well, and humans do other things very well,” Ngui says. “While expert human assessments are essential, I’m excited by the prospect of using new AI tools to augment their approach.”
Bellwether analyzed drone footage and calculated the damage to hundreds of homes according to FEMA's damage classifications. Dark red indicates the home is destroyed, orange indicates major damage, yellow indicates minor damage, and green means no damage.
A Blueprint for Resilience
Kansas City isn’t the first public-sector operation Bellwether has partnered with to help speed up disaster relief. The team is also collaborating with the U.S. National Guard to analyze aerial imagery of natural disaster zones, identifying infrastructural damage and creating labeled maps so first responders can deploy resources more quickly and effectively.
Ngui says Kansas City’s successful emergency drill represents “just the tip of the iceberg of what is possible” for damage assessment, both in his own city and beyond.
“There’s a huge opportunity for Kansas City to demonstrate how Bellwether’s technology can transform an entire workflow,” he says. “Ultimately, we don’t want it to stay here. I hope we can act as a blueprint for other cities.”