“Don’t use history as a way to plan your future!” - Greg Dee
A Florida meteorologist said this recently in the runup to Hurricane Helene. He was referring to assuming your property will not flood simply because it hasn’t in your memory. It has stuck with me as it seems like the central theme of our times. This advice is particularly relevant today, especially in industries like insurance, where the stakes are higher than ever. Traditionally, catastrophe modeling and insurance pricing have leaned heavily on historical data to predict future risks and set premiums. However, with the increasing unpredictability of natural disasters, climate change, and the frequency of extreme weather events, relying solely on the past is no longer enough. This is where satellite data and remote sensing, powered by AI, can revolutionize the landscape.
The Limits of Historical Data in Catastrophe Modeling
For decades, catastrophe modeling has been built on retrospective data: historical loss events, weather patterns, and demographic changes. While this approach has worked well and is still incredibly valuable, it’s increasingly ill-equipped to handle today’s complex and evolving risk landscape as a standalone activity. The patterns of the past are no longer reliable indicators of future risks. Consider, for instance:
Climate Change: Weather extremes are becoming more frequent and severe, leading to events outside the scope of historical data.
Urbanization and Infrastructure Changes: Rapid urbanization changes the exposure and vulnerability of areas to disasters. What was once a low-risk zone may now face higher flood or wildfire risk due to development and changing infrastructure. We saw this play out in Hurricane Sandy (2012) when a combination of meteorological confluences and more extensive infrastructure led to extreme losses.
Emerging Risks: Natural events like wildfires, hurricanes, and floods behave differently due to shifting environmental conditions, making historical models less predictive. Take, for instance, the 2017/8 wildfires in California. A very wet winter, causing abundant plant growth, combined with a very dry summer led to a severe wildfire season.
The “rearview mirror” approach falls short in anticipating these new realities. That’s where the fusion of satellite data and AI can provide a much-needed update to traditional models.
Why Satellite Data and Remote Sensing Are Game-Changers
Satellite and remote sensing data offer near real-time insights into the world’s changing landscapes at a better temporal resolution and coverage than aerial data alone. By capturing everything from current weather patterns to vegetation growth and urban sprawl, insurers can understand risk dynamically—no longer having to rely solely on historical assumptions.
Key Applications of Satellite Data in Insurance: It’s not just for claims
Real-Time Risk Monitoring: Using satellite data, insurers can monitor changing risk conditions in real-time. For example, data from satellites can detect developing flood conditions, wildfire progression, or storm paths.
Assessing Property Vulnerability: Geospatial and hyperspectral imaging technologies provide high-resolution images that allow insurers to assess vulnerabilities more accurately. For example, an AI-powered model combined with satellite data could assess roof conditions, vegetation around properties, and changes in terrain that could increase the risk of flooding, roof damage, or fire risk.
Climate Change Adaptation: Satellite data is particularly effective in monitoring environmental factors linked to climate change. Remote sensing can coastal resilience, soil moisture, and drought conditions, giving insurers real-time insights into how risks are evolving geographically and climatically.
Improving Underwriting Accuracy: Using high-resolution satellite data, insurers can create more accurate risk profiles. AI can analyze this data to refine underwriting models, taking into account real-time environmental variables that wouldn’t have been captured using static historical data.
Post-Disaster Assessments: After a catastrophe, satellite imagery provides more rapid visibility into affected areas. This enables insurers to rapidly assess damages, estimate losses, and process claims faster and more efficiently. Remote sensing also helps identify areas with fraudulent claims by providing precise before-and-after comparisons.
Parametric Insurance: Remote sensing data can be used to determine or validate triggering events.
The Role of AI in Enhancing Remote Sensing Data
While satellite data is powerful, the sheer volume of information it generates is often too vast for traditional analytics methods to process. This is where artificial intelligence comes in—analyzing, interpreting, and making sense of the data at a speed and scale that humans alone could not achieve.
AI can leverage machine learning models trained to recognize patterns of risk and vulnerability in satellite data. For instance, a machine learning model might:
Detect changes in vegetation density and predict wildfire risk.
Interpret property images (computer vision) to understand vulnerability and hazard risk, or update secondary modifiers at scale.
Cross-reference historical data with real-time imagery to identify shifts in risk patterns.
By combining AI with remote sensing, insurers can develop predictive models that aren't just reactive but truly forward-looking—continuously updated with the latest information to provide more precise risk assessments.
A New Paradigm: The Future of Insurance Pricing and Catastrophe Modeling
The integration of satellite data and AI represents a shift from reactive, historically based modeling to a dynamic, proactive approach. Insurers who adopt these technologies can:
Anticipate Risks Early: Rather than waiting for an event to occur and relying on historical analogs, insurers can model real-time conditions to anticipate risks before they materialize.
Enhance Customer Confidence: Offering more accurate and responsive insurance products based on real-time data helps customers feel secure, knowing their coverage aligns with current realities, not outdated assumptions.
Optimize Claims Management: Real-time disaster assessment means insurers can handle claims more efficiently and fairly, with rapid validation of the extent of damages.
In the insurance industry, we like to say “all models are wrong, but that doesn’t mean they aren’t useful.” Backwards looking models still hold great value, but Greg Dee’s advice resonates strongly: planning for the future based on historical data alone no longer cuts it. Insurers must adapt to the evolving risk landscape by leveraging satellite data and AI. Remote sensing provides a real-time, precise view of the world, while AI processes these vast amounts of data, offering the foresight necessary to anticipate risks and price them accurately.
In this new age of insurance, it’s no longer about predicting outcomes based on yesterday’s patterns—it’s about embracing cutting-edge technologies to understand the world as it is today and where it’s headed tomorrow.
Very interesting issue! Can I recommend it in my next newsletter issue? :)