Shifting weather patterns challenge future flood prediction

New technology juggles storm water risk, urbanization and climate change.

Floods of the past do not give you the flood risk of the future. (Credit: Sergey/Adobe Stock)

With global catastrophe losses surpassing $100 billion for the sixth consecutive quarter, insurance companies are scrambling for better risk assessments.

Flood risk modelling is at the top for many in the wake of Hurricanes Helene, Milton, severe convective storms in the U.S. and recent flooding in parts of eastern and southern Spain.

PropertyCasualty360.com recently spoke to the co-founder and CEO of 7Analytics, Helge Jorgensen, on a new computer model that provides insight into storm water risk and how it spreads in tandem with urbanization and climate change.

Prior to co-founding 7Analytics, Jorgensen held several geology roles and has a master’s degree in structural geology from the University of Bergen.

PropertyCasualty360.com: Why is flood risk modeling from the past not helpful for the future/? What’s changed?

Jorgensen: Flood risk modeling is a wide field covering both pluvial, fluvial and coastal flooding events. There are traditional approaches that are still very relevant – But the question is: how can we innovate flood risk understanding to improve future predictions?

Floods of the past do not give you the flood risk of the future.  Shifting weather patterns are making floods more unpredictable, bringing them to previously unaffected areas and increasing the frequency in regions once considered low-risk.

If flood risk models rely only on historical data, they will not accurately capture future risks—highlighting the need for a different approach. A significant challenge is the lack of detailed, location-specific flood claims data, which is essential for enhancing model accuracy. Additionally, the use of low resolution and outdated terrain data limits the ability to assess flood risk accurately.

Meanwhile, urban flooding is a pressing issue due to city growth. Old infrastructure struggles to handle increasing water volumes, and constant changes in land use and terrain add complexity to creating accurate flood risk models.

PropertyCasualty360.com: Where are the riskiest locations in the U.S. for flooding outside coastal states and places along rivers/?

Jorgensen: This is a complex question, and addressing it requires innovative methods to assess the initial risk of urban flooding. If an area has never experienced flooding, how can we determine if it’s high or low risk? Our data allows us to pinpoint the area’s most likely to face damage in the near future. We will not pick any specific areas, but we see both cities and neighborhoods with highly varying overall risk.

We are seeing in 2024 that catastrophic flood events such as Helene and the earlier flooding event in June striking Iowa, Minnesota, Nebraska and South Dakota are hitting inland areas where major flood events have never happened in the past and where residential flood insurance coverage uptake is sadly typically less than 3%.

Also important to consider that flooding risk spikes after wildfires burn away vegetation, causing rain to run off faster. In 2023, California experienced this with flash floods and mudslides following severe fires, as charred soil couldn’t absorb rainfall.

PropertyCasualty360.com: How can an accurate weather model be implemented when weather has become so erratic/?

Jorgensen: For this reason, our flood risk prediction models do not rely on historical weather data. The increase in unpredictable weather patterns adds a new layer of complexity to flood model prediction.

Areas that have never experienced flooding before are now at risk, making historical data insufficient for predicting future flood events.

PropertyCasualty360.com: What is 7Analytics and how does the technology predict flooding?

Jorgensen: Founded by four geologists with over 100 years of combined experience in complex subsurface fluid systems, 7Analytics is advancing flood risk prediction through models deeply rooted in geoscience, including hydrology, geology, and geomorphology.

This data centric approach allows for the development of specialized datasets that reveal the complexities of flood vulnerability. By analyzing hundreds of physical parameters for each building, 7Analytics applies natural science insights to identify why a building may or may not be at risk, offering a precise understanding of flood risk on an individual level.

We leverage our market-leading terrain mapping capabilities in conjunction with advanced machine learning techniques trained on extensive claims data. This innovative combination allows us to deliver cutting-edge precision in flood risk assessments.

By integrating detailed terrain analysis with real-world claims insights, we can enhance the accuracy of our predictions, enabling insurers to better understand and mitigate risks. This holistic approach not only improves risk evaluation but also supports the development of tailored insurance solutions that address the unique challenges faced by property owners in flood-prone areas.

PropertyCasualty360.com: Can the technology’s accuracy lead to more insurance carriers removing coverage in higher risk areas?

Jorgensen: On the contrary, it’s actually the presence of precise data that enables insurance coverage. With 7Analytics’ granular, building-level flood risk data, insurers can move beyond broad, zip-code-based assessments and evaluate risk for each property individually.

Helge Jorgensen

This level of detail allows carriers to differentiate flood risk between neighboring buildings, creating opportunities to re-enter markets previously considered too risky. For buildings with elevated risk, additional mitigation measures might be necessary to secure coverage, as some could otherwise lose access to affordable premiums.

In the U.S., many major property insurers currently avoid offering flood coverage, largely due to the lack of reliable, high-resolution flood risk data to accurately assess their exposure. Carriers have indicated in meetings that this level of data is critical to their return to the market. Addressing this need is central to our mission in the U.S. – through our precise building-specific flood risk assessments, we aim to empower insurers to offer tailored flood protection in areas where coverage has long been limited or unavailable.

PropertyCasualty360.com: Can this technology impact property values? How?

Jorgensen: High-resolution flood risk data can actually have a stabilizing effect on property values. When homeowners and insurers have access to precise risk information, it becomes easier to maintain insurance coverage in areas prone to flooding, which, in turn, supports property prices.

Storm Damage from Hurricane Helene, Florida. (Photo: Mark Rankin/U.S. Army Corps of Engineers, Jacksonville District)

However, many homeowners today face significant challenges where repeated flooding is already depressing property values. For example, in parts of Florida, sellers often have to repeatedly lower their asking prices to attract buyers – if they can sell at all.

Granular data also raises awareness about specific properties at risk. Many homeowners and businesses depend solely on FEMA-designated caution areas, assuming they are safe if located outside these zones. Yet, as recent hurricanes have shown, many homes outside these areas still experienced severe flooding.

Estimates in North and South Carolina suggest that only 2 to 4% of flooded homes were covered by flood insurance, and many homeowners were unaware they lacked coverage, leading to devastating personal and financial consequences.

Better awareness of flood risks enables homeowners and insurers to collaborate on preventive measures that reduce future flood risks. Globally, we see the significant impact of flood risk on property values. In Australia, for instance, new flood mapping in Melbourne’s Kensington Banks has sparked concerns about potential value losses. Similarly, climate-related risks are increasingly affecting real estate markets across Australia and New Zealand.

See also: