Sophie Yeo is an independent journalist covering climate change, the environment, and conservation. She has written for the Guardian, the Washington Post, Earther, Hakai, and others. She was formerly a staff writer for Carbon Brief.
Hi-res images taken from orbit can more accurately track damage from extreme weather events.
A new fleet of satellites could become the ultimate foil for fraudsters looking to make a quick buck in the wake of a natural disaster.
That’s the intention of Planet, a San Francisco company that has sent around 380 satellites into space since 2013. Every day, these satellites take high-resolution images of the entire globe, creating an enormous data set of every significant change to the surface of the Earth.
By taking the images and processing them using artificial intelligence, these satellites can track anything from deforestation to offshore oil rigs. And as climate change causes extreme weather events to become more severe, Planet’s executives believe these images could have another vital application: detecting disaster insurance fraud.
Planet currently has more than 175 “Planetscope” satellites in orbit, which photograph the world daily at three meters per pixel (meaning lots of detail). In addition, the company has 13 “Skysat” satellites, which can zoom in even closer, taking photos at 0.8-meter resolution. With images this detailed, it’s possible to see whether a building has burned down, a farm has flooded, or a road become blocked.
When it comes to tracking crime, the satellites are like CCTV on steroids. And instead of tracking individuals, the satellites will be able to gather physical evidence to help investigators determine whether a crime has been committed.
“In the environmental and social sectors, we often have organizations solving 21st-century problems with 19th-century tools,” says Andrew Zolli, vice president of global impact initiatives at Planet. “These satellites are continuous, and they tell the truth. What we see coming is a world in which the combination of very sophisticated analysis and continuous data feeds allows people to take some categories of fraud and make them so difficult to perpetrate that they cease altogether.”
For those looking to profit from a disaster, fraud can be an easy win. Readily available government aid in the wake of a disaster provides an incentive to break the law, while the need for quick pay-outs means that claims can be difficult to monitor. Sending professionals into the field to verify whether claimants are telling the truth is expensive and time-consuming.
Following Hurricane Katrina, federal prosecutors charged more than 1,300 disaster fraud cases; the Federal Bureau of Investigation estimates that insurance fraud may have accounted for as much as $6 billion of the $80 billion in government funding appropriated for reconstruction. To date, the government’s National Center for Disaster Fraud, set up after Katrina, has received more than 70,000 complaints involving over 50 disasters.
It may seem like a victimless crime—sticking it to the government, or else faceless insurance companies to whom you’ve paid high premiums for years—but, in reality, disaster insurance fraud can delay payments to genuine victims, and takes already-limited funding away from those in need.
Satellite images could suddenly make it a lot riskier to perpetrate these crimes. In property insurance scams, for instance, someone might falsely claim that their house was damaged by floods. Armed with high-resolution images from before and after the event, insurance companies and law enforcement agencies could quickly examine whether the claimant is telling the truth.
Another application could be in the case of droughts, where the government has limited a given region’s use of water. Such is the case with California, which is currently considering permanent restrictions in the wake of its severe and regular droughts. If the satellites captured an image of one green field or backyard amid a sea of brown, it could be a clear indication that someone is illegally overusing their sprinklers.
The difficulty with capturing so many images is that it quickly becomes too complex for a human staff to analyze. For that reason, Planet is developing algorithms, relying on artificial intelligence, to analyze these stacks of images and automatically detect change, sending out alerts when it registers some kind of disturbance.
“This machine learning tool, it’s constantly looking at the imagery and classifying things it finds within them: that's a road, that’s a building, that’s a flood, that’s an area that’s been burned,” Zolli says. “The system of change-alerting is currently done in ways that are somewhat automated and somewhat manual, but their overall direction is to become completely automated.”
That is just one way in which artificial intelligence technologies can help people to survive the disasters and emergencies caused by extreme weather. A recent report from PricewaterhouseCoopers suggests further applications, including weather event and risk prediction, automating floodgates, and choosing flight paths that are better informed by the weather forecast.
As well as these potential applications for insurance fraud detection, Planet is already working with international development groups to develop new insurance products to protect vulnerable people in the developing world.
In continents like Asia and Africa, people can easily enter a cycle of poverty when a natural disaster hits. In 2017, some 92 percent of economic losses in Asia related to natural catastrophes were uninsured, according to the insurance group Munich Re. In Africa, it was 87 percent.
Some assets are practically uninsurable under regular operations: Verifying the claims of a rural farmer in Kenya whose crops have been destroyed by drought, for instance, would simply be too time-consuming and expensive for a regular insurance company to consider.
To combat this trend, in 2015 the G7 adopted the InsuResilience Initiative—a new partnership aiming to offer 400 million new people access to climate-related insurance in the developing world by 2020. One of the projects that’s helping the G7 move toward this target is a partnership between the German government and Cloud to Street, a startup that conducts flood modeling using data collected by Planet’s satellites.
In Tamil Nadu, India, Cloud to Street has developed a flood detection scheme where payouts from insurers are automatically triggered when satellites identify the presence of flood water on the ground. The index is already functional, and the company hopes to deploy it with an insurer during the monsoon season this fall.
“There are shopkeepers and smallholder farmers who, in many cases, are living on the edge of poverty,” says Bessie Schwarz, co-founder of Cloud to Street. “When floods hit, they take their kids out of school and sell off critical assets so they can still afford medicine, still afford to eat during monsoon season. The satellite imagery can detect that potentially as it’s happening, or near real-time, and then do those payouts very quickly, so we can provide that capital to people there.”
While Planet uses just under 10 satellite fleets to track flood data, the company’s offerings could be particularly valuable because its high-resolution images capture every detail of the globe, from Beverly Hills to Bangladesh, allowing users to track changes to individual properties in any country. Other satellite groups tend to be too coarse, too infrequent, or too specific to be useful to insurers.
Still, some observers and analysts are still skeptical of the benefits of climate insurance, worried that companies may seek profit at the expense of the poor.
“There are hardly any examples to show that insurance has helped large populations of poor people in recovering from the climate impacts,” says Harjeet Singh, global lead on climate change for the nonprofit ActionAid. “On the contrary, several reports indicate that corporations gain more out of insurance than poor people. For instance, in India, the percentage of outstanding claims from crop insurance rose to 32 percent in the last few years, but the insurance companies made huge profits from the premiums. [Climate insurance] only plays a limited role in transferring risk for a small segment of people and is not a viable instrument for dealing with slow-onset impacts such as sea level rise, increasing salinity, desertification, and so on.”
Meanwhile, others have raised concerns around privacy, as well as the potential for cluttering space with too many satellites.
“To be sure, the trend, and it’s not just from Planet but from other satellite-based imaging companies, is moving toward imaging the whole Earth's surface every day,” says Laura Grego, an expert on space security at the Union of Concerned Scientists. “However, Planet images are low resolution compared to the most advanced satellites in orbit, meaning Planet satellites can't see small details as well as other, bigger satellites”—including military reconnaissance satellites, which “have resolution in the tens of centimeter range.”
“Planet fields a large number of satellites to do this job, which I know can be concern from a space debris and space traffic standpoint,” Grego says. “I do know the founders have a personal commitment to the sustainable and peaceful use of space, and that they aim to be responsible space users.”
Relentless sentinels of law and order, or guardian angels with a direct line to your wallet—your perspective could well depend on how many laws you’re breaking—these satellites could play an important part in the tech industry’s answer to disaster and emergency response.