- Climate-driven natural disasters are upending lives and economies across the globe. Focusing on adaptation is necessary to minimize their impacts
- Wildfires today burn an average of 7.7 million acres every year in the U.S., up from 2.1 million acres 40 years ago, putting 4.5 million homes at high or extreme risk
- The disaster mitigation and response sector encompasses myriad agencies, jurisdictions, and organizations, from NOAA to small rural fire departments. It is a sector that has historically been underserved by the technology industry, and it is hungry for innovation
- The technological building blocks for better, faster disaster response—from smart cameras and drones to sensors, 5G networks, satellite connectivity and imaging, and AI—already exist. A growing cohort of innovative companies is starting to put them to work to save property, ecosystems, and lives
When Hurricane Ian crashed into southwest Florida in 2022, it caused massive and widespread damage. The Category 4 hurricane caused more than 150 direct and indirect deaths, and more than $112 billion in damage. It was the costliest hurricane in Florida’s history to date, and the third most costly in the U.S.
But one town remained mostly unscathed. A traffic light fell, and so did some palm trees. Lights flickered in some homes, but none lost power. That’s because Babcock Ranch, located north of Fort Myers, was designed and built from the ground up, starting in the 1990s, to withstand natural disasters like hurricanes—all without losing power or water.
The resilience of Babcock Ranch underscores two important ideas. First, the city is a testament to the power of adaptation: taking specific action to adjust to the impacts of natural disasters whose strength and frequency are increasing. Second, adaptation can start now. Many of the technologies we need already exist. It’s a matter of putting them to use in innovative applications and products that can make us safer and more resilient.
Harnessing the spirit
Indeed, nothing that Babcock Ranch did required years of intense research or product development. The community buried power lines and fiber optic cables underground, shielding them from strong winds; it uses smart grid technology that can keep local failures from spreading; it engineered its streets so they could absorb floodwaters; and the community is built 30 miles inland in order to avoid coastal storm surges.
The spirit of Babcock Ranch is what’s driving our work at Pano AI. As the threat from natural disasters has increased, it has become clear that we must change the way we approach disaster mitigation, preparedness, response, and recovery, including by leveraging modern-day technology tools. We’ve chosen to start by focusing on wildfires, but similar approaches could be put to use to help our world adapt to a range of natural disasters.
Disasters of unprecedented scale
As a California company, our team at Pano had experienced the devastation of wildfires firsthand, and we were eager to apply our technology backgrounds to build tools for the firefighting community. The urgency has never been higher. One-third of the 15 largest wildfires in U.S. history took place in the last decade. In a matter of hours, some of these wind-driven infernos have decimated entire towns—or large portions of them—such as Santa Rosa and Paradise, California, and most recently, in Lahaina, Hawaii.
The catastrophic 2019-2020 Australia bushfires significantly damaged human and animal lives, ecosystems, and properties. An estimated 24.3 million hectares burned, destroying more than 3,000 buildings and killing 34 people. Air quality became hazardous, and the smoke impacted weather conditions even on other continents. In 2023, Canada experienced its worst wildfire season on record. The resulting smoke exposed more than 100 million Americans from Minnesota to New York to sooty air, which has been associated with higher rates of hospital admissions for cardiovascular disease, diabetes, kidney failure, and has been linked to other ailments. Similar disasters have caused unprecedented damage from Australia to Chile to Greece.
There is a growing consensus in the industry that early detection and an aggressive initial response is often the best chance at containment. The first hour or two of a fire are critical, and the initial response can determine the ultimate threat to lives and property. And yet most emergency responders still rely on a bystander calling in a fire when they see one. This detection method can be unreliable and lack key information like fire location and severity.
To catch fires while they’re still small, we built an integrated technology solution that utilizes ultra-high definition cameras and AI. The cameras sit atop a tower and look out for smoke in a 10-mile range—much like a person in a lookout tower would. Computer vision technology detects plumes of smoke and differentiates between smoke and fog or clouds. At night, the cameras switch into near-infrared mode that detects heat from a fire.
“The first hour or two of a fire are critical, and the initial response can determine the ultimate threat
to lives and property.”
When the AI determines there’s reason enough to sound the alarm, it sends a video with a bounding box around the potential smoke or heat to our 24/7 Pano Intelligence Center for human review. If a fire is confirmed, an alert gets pushed to the appropriate end users: emergency services, fire departments, utilities, local officials, and others. The alerts include GPS coordinates (calculated by triangulating the location from multiple cameras) and real-time camera footage. The system generates a live incident page that’s updated every minute and can be shared at the click of a button. To remove friction to broad-based information sharing, there’s no login required for the first 24 hours.
This system was built entirely on existing technologies that were developed for other uses. Indeed, I like to think of us as technology scouts: We go out and find proven tech and harness it to build elegant integrated solutions. Above all, we make sure the tech is cutting-edge and fresh—and we are using it in new and innovative ways. Sometimes it comes from an unexpected place. For our push notifications, for example, we use a software tool that was developed for marketing purposes.
For our tech stack, we leverage the latest technology to create an easy-to-use product with an elegant user interface. This includes everything from Linux distribution tools to Kubernetes to scale our cloud computing, as well as advanced AI chips to process the massive amounts of data our cameras generate: 360,000 megapixels per hour and more than 2,000 GB of image data per month and counting. Our AI fire detection models use the same object-detection computer vision technology that powers many autonomous vehicles.
We strive to be early adopters of new technology, so that customers can benefit from cutting-edge innovations. That’s why we moved from microwave connectivity to 4G cellular to 5G, as each new technology allowed us to push our cameras deeper into the forest. A Starlink option is available for the most remote locations, where even 5G won’t reach.
A cohort of like-minded innovators
Of course, there is no single solution that will help us adapt to the growing threat and severity of wildfires and other natural disasters. That’s why I’m excited by the growing cohort of like-minded innovators who are also developing tech-enabled solutions for a better, more adaptable, and resilient future.
Take Burn Bot. The robotics startup is bringing technology to controlled burns—one of the most proven methods to prevent megafires. It uses drones and remotely operated vehicles to initiate burns in remote areas, acting as a force multiplier that vastly expands the reach and precision of this critical wildfire prevention tactic. Another company, Rain, also uses manned and unmanned remote-controlled helicopters that can be deployed to douse a fire with fire retardant as soon as it is detected. And Overstory uses AI to analyze satellite imagery and pinpoint places where vegetation is a hazard to power. Its machine learning models can predict how trees and shrubs will grow and when they might fall, allowing utilities and others to mitigate threats before they develop.
To gain the upper hand on the growing threat posed by wildfires and other disasters, we need even more investment, more ingenuity, and more innovation. The sooner we start working on technologies to adapt to an unpredictable planet, the more we’ll be able to do to ensure safer communities, good quality air, vibrant ecosystems, and resilient infrastructure—and we’ll all be able to breathe a little easier.