New hail forecasting research helps predict the most unpredictable of perils


News  

Historically, hail is known to be one of the more uncertain of natural perils. It is difficult to predict its occurrence, its size, its location and consequently the extent of its damage on the ground. However, thanks to a collaborative partnership between Suncorp and the Bureau of Meteorology (BOM), recent research is available to help.

Share

New hail forecasting research helps predict the most unpredictable of perils

Meteorologists around the world might unfairly wear criticism over unpredictable forecasts that ruin plans and holidays, but innovation in weather radars has greatly improved over the years.

Weather radars are the primary meteorological tool for observing the location, volume, and type of precipitation (e.g., rainfall, snow, or hail). However, while radar has been critical in monitoring and providing warnings on severe weather, it is incredibly difficult for these instruments to tell the difference between rain and hail in approaching storms.

Realtime warnings on hail have therefore been limited and despite its massive potential for damage, innovation in this space has not advanced much in 20 years – until now.

Suncorp collaborated with BOM’s Science and Innovation Research Group on a more accurate radar forecast solution

Dr Rhys Whitley, Lisa Ye and Dr Nina Ridder from Suncorp’s Natural Perils team co-authored a research paper with BOM’s Science and Innovation Research Group titled ‘Radar and environment-based hail damage estimates using machine learning.’

The article was published in the ‘Atmospheric Measurement Techniques’ scientific journal in January 2024, and detailed how BOM and Suncorp successfully developed a new hail radar solution based on radar recordings and insurance claims data.

“If there are very large hail stones, they’ll typically fall straight down. But other sizes of stones might deflect elsewhere within the storm, or high winds might accelerate their falling pattern. Melting can also occur before it hits the ground,” he said.

“To date, the best weather forecasts can do is warn the public of thunderstorm activity over very broad geographical areas and that there is a potential for hail.

“This hasn’t really changed much in Australia for the last 20 years.”

Combining radar recordings and insurance claims data for a more accurate indication of hail risk

The research project analysed where damaging hail had fallen historically, using machine learning to combine large amounts of data from BOM radar recordings with Suncorp insurance claims data. Together, they created a Hail Damage Estimate (HDE) score.

Suncorp Scientific Specialist, Pricing & Analytics, Dr Rhys Whitley

Our collaborators at BOM were measuring what is happening in the atmosphere. And we, as an insurer, can measure the impact of extreme weather on the ground. This research was about how to connect the two for a more accurate measurement of hail risk.

The research set out to close the gap in information about what happens with hail between forming in the sky, and when it hits the ground, using machine learning to process large amounts of data.

Dr Whitley said it means radar can potentially be more accurate at observing the impact of severe thunderstorms in real-time. “We can say, for example - ‘there is damaging hail developing in this storm system, and it is likely to hit the following areas,” he said.

Previously, hail forecast gave broader, ‘city-wide’ indications of hail, whereas the new radar can provide a more detailed, suburb-level indication.

The increased accuracy of hail risk also has potential benefits for future insurance customers

Homes and vehicles are at risk of significant damage from hail. Hail can smash roof tiles, causing a collapsed roof or interior flooding, in addition to broken windows, punctured cladding, and even punctures through Colorbond have been reported. In regional Australia, hail can wipe out a crop in seconds, devastating a season.

On 31 October 2020, a devastating hailstorm struck residents in a small pocket south-west of Brisbane. Power was cut to thousands of homes, while giant hail smashed through roofs of houses across Springfield Lakes and Greenbank.

The new research’s ability to more accurately pinpoint hail at a suburb level can help communities better prepare their homes and vehicles for a potentially damaging hail event in the future.

There are potential benefits to customers through Suncorp’s disaster response teams too – the more accurate indication of hail risk will be able to help optimise preparations and deploy resources to the most at-risk communities.

The hail risk research closely aligns with Suncorp’s four-point resilience plan that underpins decisions, investments and how the insurer shows up in the community and for its customers.


#Subscribe for Suncorp updates