By Russ Banham
Better ways are needed to predict the risk of a wildfire and devastating floods causing property damages, loss of life and devastating floods—not just to prevent such disasters but also to insure them.
In 2021, the insured losses from global natural disasters totaled $130 billion, with the United States accounting for 71% of the losses. A sobering case in point is Hurricane Ida in August 2021, which due to record rainfall levels, shut down much of the subway system in New York and produced $36 billion in total insured losses.
Global insurers and reinsurers consider climate change to be the industry’s number one concern, making it a concern of everybody, since ever-increasing insured losses are passed on to businesses and people in the form of higher insurance premiums. Unless zero carbon emission targets are met by 2050, climate change is estimated to reduce global economic output by 11% to 14%.
Putting aside this immense financial toll, disasters like the November 2018 wildfire in Paradise, California, in which 85 people died and more than 19,000 structures burned to the ground, are a hammer blow to the heart. To better predict the location and impact of future natural disasters to mitigate tomorrow’s ungovernable wildfires and floods, more than a score of technology startups have been launched with this mission in mind.
A new approach to calculating risks
The startups, classified by the insurance industry as “insurtech” firms, are tapping into a growing range of insightful data, including satellite imagery, topological records, flood maps and information on the age and condition of physical structures. Leveraging artificial intelligence (AI)-enabled, machine learning (ML) algorithms to sift through hundreds of thousands of data points, the insurtech platforms enhance the risk modeling of droughts, wildfires and floods—the repetitive climate change-induced cycle responsible for incalculable environmental and economic damage.
Insurtech companies like Delos, for instance, use a combination of refined satellite imagery and proprietary AI algorithms to model the risks of a wildfire in the wildland-urban interface, the transitional zone between unoccupied wilderness and developed land containing more than 44 million housing units. Delos is focused first on California, the state that accounts for over half of the country’s wildfires. Over the last 10years, one of every eight acres in California has burned to the ground.
“Using our models, an insurance company can discern the specific risk of wildfire damage on one block versus the next block,” says Kevin Stein, CEO and co-founder of Delos. “This is a crucial consideration for humanity, since insurance is a financial dampening mechanism, smoothing out the peaks and valleys of risks that a single entity alone could not bear.”
The insurance industry’s traditional underwriting approach is based on an assessment of current building risks, such as the condition of the structure, the quality of materials used, and location of the nearest fire station, among other factors. Insurers further rely on past losses to predict future losses, says Stein. “This doesn’t work in predicting possible losses due to climate change, since the risk is growing too fast to make sense of the future potential for loss,” he asserts.
Data-driven risk considerations
Five years ago, Stein launched Delos to enhance the status quo of predictive capabilities. “My co-founder (Delos chief data officer Shanna McIntyre) and I are both from the aeronautics industry and were involved in predictive modeling,” he says. “We partnered with an environmental think tank and a spatial information group, working alongside nearly 100 post-doctoral scientists, researchers and university professors studying climate change, to understand peak temperature risks on an extremely granular level.”
Stein and McIntyre subsequently contracted with the California Department of Forestry and Fire Protection to create Delos’s wildfire risk model. More than 100 different data points are provided by CalFire on different geographic-related fire risks. The data is fed into Delos’s platform, which leverages the firm’s proprietary AI algorithms to create a 3D picture of wildfire risks on a building location basis.
Each block [of homes] has a different fire risk level than the next block. In some cases, our model can discern this risk on homes right next door to each other.
―Kevin Stein, CEO and co-founder, Delos
The data informing the risk assessment includes traditional information on area vegetation, wind speed, rainfall, temperature and land slope, as well as data culled from satellite images, topological maps, drought records and other inputs that Stein says make up the company’s secret sauce. “All I can say is that if you have a tree behind your home that’s a certain number of feet from single-pane windows on your house, it’s reflected in the model,” he says.
By personalizing the risk of fire losses to each person’s home, Delos makes underwriting decisions for insurers more secure. “Each block [of homes] has a different fire risk level than the next block,” he says. “In some cases, our model can discern this risk on homes right next door to each other.”
Biblical floods on tap
Another insurtech firm looking to make a difference is reThought, which is focused specifically on the flood insurance risks of commercial buildings valued anywhere from $2 million to $1 billion.
“Flood risks are related to wildfire risks, in the sense that a wildfire leaves the ground charred, causing the roots of trees and plants to die for a period of time,” says Cory Isaacson, CEO and co-founder of reThought. “When the next big rainfall comes along, there’s nothing to absorb and hold the water back.”
Isaacson speaks from personal experience. In November 1993, he and his family lived in Old Topanga Canyon, a rural area in Los Angeles near the Malibu beaches. A wildfire that started at his next-door neighbor’s property skipped his house and burned more than 24,000 acres down to the beach. “The smoke plume looked like an atom bomb had hit; hundreds of homes were destroyed,” Isaacson says.
Months before, the area experienced the heaviest rainfall in more than 100 years, followed by a severe drought; ripe conditions for a massive wildfire. “After the ground burned, a geologist I know persuaded us to build a 10-foot-high concrete wall above the hill out back,” says Isaacson. “He was concerned about this huge rock outcropping above our property coming loose in the event of another major rainfall.”
Two years later, torrential rains fell on the hill, and a massive boulder came loose from the outcropping and careened towards the Isaacson home. “It crashed through the wall and stopped a few feet from the house; had it not been there, someone would have died,” he says.
ReThought’s technology platform filters wide-ranging hydrological, topological and geospatial data from traditional catastrophe risk assessment models provided by firms Verisk Analytics and RMS. “We take these historical events and other data like historical satellite imagery to create new events that could happen just like them,” he says. “We’re then able to make a more accurate prediction of the risk of a flood for insurance pricing purposes.” The satellite data is provided by a partnering organization, Hydronos Labs.
By having a more accurate understanding of potential flood losses, insurers are more apt to provide flood insurance. Isaacson cites an example of a large corporate campus with several buildings. Previous estimates of potential flood damage losses ran as high as $6 million. “Our formula had it a little over $1 million,” Isaacson says. “The more granular we get, the better the ability to offer more complete and higher-value coverage limits.”
That’s good news for property owners. Since 2000, flood events have cost U.S. taxpayers more than $850 billion. Flooding is responsible for approximately two-thirds of the cost of all natural disasters. According to Flood Defenders, a non-profit, non-partisan flood research group, around 6 million Americans unknowingly live in a 1-in-a-100-year floodplain, a disaster in waiting.
We take these historical events and other data like historical satellite imagery to create new events that could happen just like them. We’re then able to make a more accurate prediction of the risk of a flood for insurance pricing purposes.
―Cory Isaacson, CEO and co-founder, reThought
Both Delos and reThought monetize their businesses as Managing General Agencies, a type of insurance intermediary that receives a commission from insurance companies for their risk assessment and pricing services. As the planet continues to warm, other insurtech startups like Cape Analytics, FutureProof Technologies and zesti.ai also are taking stock of the rising cost of natural disasters and the impact on insurance availability and pricing.
As Stein puts it, “In a time when so many insurance companies are pulling out of communities at risk of wildfire, we’ve built an innovative data analytics platform that enables us to deeply understand wildfire risk.”
The invention is good news for anyone with a home in the wildland-urban interface.
Lead photo courtesy of Shutterstock