A decade ago, large reinsurers were the first entrants from the insurance sector in the market for satellite Earth Observation (EO) imagery, according to Isabelle Flückiger, the program director for new technologies and data at the Geneva Association, a think tank run by insurance industry CEOs.
Reinsurers are businesses that insure insurance companies. Reinsurers need to know, in the wake of large natural catastrophes like hurricanes, wildfires, or flooding, what the losses are on an aggregate basis, she says. Conventional data gathering methods on the ground can take months to provide a complete picture of the damage as investigators grapple with disaster conditions and destroyed infrastructure.
Reinsurers, who set aside capital to meet potential losses, need answers more quickly. They had the resources to interpret EO data in a very manual way, employing human analysts to pore over bespoke imagery bought at great expense from one of a handful of satellite operators selling in the commercial marketplace at that time.
Direct insurers then followed suit, also using human analysts to interpret imagery, and still only seeking aggregate data. “This is all about estimating the scale of total losses more accurately and quickly. It was not linked to products,” Flückiger says.
Four or five years ago, that all started to change. Insurers now use data derived from satellite imagery both to assess individual claims and to offer whole new categories of products.
This data revolution has cut claim assessment times from weeks or months down to days, Flückiger explained — and enabled insurers to expand a category of products called parametric insurance. Unlike traditional insurance — which covers customers for actual losses — parametric insurance pays out based on trigger events, such as a flood which rises more than a certain number of inches.
There are many of these kinds of insurance products in the agriculture sector. Companies offering insurance or finance to farmers can use processed EO data to assess their potential profitability and creditworthiness by looking at crop yields, and meteorological data like rainfall — as well as figuring out their losses in an event, or when the trigger points for parametric payouts are passed.
In another use case Flückiger described as “early maturity,” algorithms are used to augment the work of risk engineering teams conducting on-site inspections of insured commercial property. Artificial Intelligence (AI) programs highlight potential hazards and pre-existing damage by analyzing imagery. “The teams go onsite with already quite a good picture of where a property might need repairs or might have other issues,” she says.
“What has changed is the availability of data,” Flückiger concludes. Interpreting EO imagery is no longer a manual process. Today it is more and more automated, with Artificial Intelligence algorithms translating imagery into spreadsheets, indicating for example the depth of floodwater in a series of locations — and then analyzing the impact on crop yields, industrial production, or property damage. There are also historical data sets that can be used to index current measurements and feed predictive models, for instance of wildfire spread.
The data revolution Flückiger describes isn’t limited to the insurance and financial services sector. Analytics and other data products from satellite imagery are projected to be a $2.2 billion annual business by the end of the decade, according to NSR’s Big Data Analytics Via Satellite annual report published in November 2020.
According to NSR, the services sector — which includes insurance, banking and other financial services, along with commodity supply chain monitoring — is only the fourth-largest market for satellite data after government, transportation, and energy. Services currently represent only 7 percent of the revenue from satellite data; but that’s expected to grow to 22 percent by 2029.
“The EO market is still dominated by the government and military sectors,” says Shivaprakash Muruganandham, NSR senior analyst and author of the satellite data report. “But the number of players has grown, the number of ways to pull value from that imagery has grown.”
That growth has created a market opening for one-stop-shops like SpaceKnow, Ursa and Descartes Labs, who collate imagery from multiple providers into a single platform — and output data in use-case friendly formats that banks, insurers, and other financial service companies can plug into to existing business processes.
Muruganandham calls it the Netflix model: “You sign up, subscribe, and then when you log on, you have a menu of options. You can buy the imagery, you can buy the analysis.”
Another use case of great utility to the financial services sector is data about economic activity. Analysis of EO imagery can provide data that’s more timely, more objective, more detailed, and more easily disaggregated than conventional data sources like government reporting, according to Anu Murgai, vice president for economic and commodity products at SpaceKnow Inc., one of the new players in that EO application layer.
“I can count finished cars in a factory parking lot. In mining, I can size the piles that sit outside of either raw materials or finished product.” The company can capture key performance indicators and other activities for many industries, says Murgai — in a very timely fashion and with complete objectivity. “The way we do our data, it is at most four days old ... Government data is [typically subject to] a month plus delay. And in some emerging market countries, or for certain industries, it's longer.”
Government data also tends to be survey-based — “you are asking people about their output and how it’s trending. But our data is simply what we're seeing on the ground. It also provides a different quality of information,” Murgai concludes.
From the point of view of a customer in the banking sector, that quality of information powers decision advantage, says Adam Maher, CEO and founder of Ursa Space Systems. “This is really about understanding what is happening in the market more quickly than your competitors.”
Ursa provides a wide sheaf of on-demand data analytics solutions for government and private sector decision makers, and is known for its weekly figures on global crude reserves. But Maher says it will take as long as a decade for the most recent innovations in orbit and on the ground to work their way into the marketplace.
“This data piece is just starting. In the next couple of years, the infrastructure that’s going overhead and in space is just starting to come on stream. It’ll take something on the order of five or 10 years to really bring it all to market,” he says.
In the meantime, existing overhead architecture is powering a whole new set of use cases, driven by the growing impacts of climate change, and the increasing proliferation and worsening intensity of extreme weather events it is driving. Both short and long-term impacts are of grave concern to the insurance industry, which is increasingly turning to data derived from EO imagery to predict, warn, and hopefully mitigate extreme weather events.
“Geospatial intelligence fueled by EO data can help the financial services sector monitor the effects of climate change in a number of different ways,” says Alex Diamond, product marketing lead at Descartes Labs. “In the short term, they need to do market forecasting and assess the damage from large physical events like floods or hurricanes in near real time. In the longer term, they need to be able to assess climate risks to different businesses. What’s going to happen to South American coffee production, for instance, if average global temperatures rise by three degrees?”
This kind of work, Diamond says, takes "massive compute to power the analysis of the world’s largest physical systems, create models of continental or planetary size, and use them to craft a complete business solution for predictive modelling.”
Diamond calls Descartes Labs a “data refinery” — pulling in raw imagery and “cleaning and calibrating it, to make it analysis ready. Then running whatever analysis the customer needs, to produce the data most useful to them.”
“We’re not just sending them pictures,” Diamond says. “It’s a fully analyzed data set provided as a complete operational solution … Our job is to get as close as we can to our customers’ business needs.”
To this end, Descartes Labs actually builds bespoke models for some of its larger customers, he adds.
Another climate change area Descartes Labs specializes in is data for so-called carbon accounting — a key element of environmental, social and governance (ESG) investment. When companies make a net-zero emissions pledge, it typically involves both reducing emissions and offsetting them against measures designed to absorb carbon from the atmosphere, like planting trees or enriching soil carbon.
Descartes Labs offers companies a way of monitoring, reporting and verifying both emissions — their own and others’ — and offsetting activities like tree planting. This enables banks and other financial services companies to verify emissions pledges and satisfy the demand of their ESG investment customers to balance money-making with environmentally sustainable business practices.
Such use cases are going to get bigger and bigger, says SpaceKnow’s Murgai.
“You have the ability to measure various forms of pollution, monitor water reserves, vegetation [coverage], deforestation,” she says. “These are all things that matter, not just to the financial services industry that's looking at ESG investing, but also anybody who needs data about climate change related issues of any kind, which is everybody.” VS