New research highlights flaws in cyclone risk assessment

A new systematic review reveals serious shortcomings in cyclone risk assessments in Australia and around the world. Analyzing 94 studies on cyclone risk, admonishes that existing approaches may not provide a full picture of the risks facing communities.

A study titled “A Critical Review of Hurricane Risk Assessment Models and Prediction Frameworks” was published in the journal Geoscience Frontiers. This represents the first comprehensive review of cyclone risk assessments.

Over 80 cyclones, typhoons and hurricanes form around the world every year. Australia faces some of the most powerful and harmful systems. They threaten lives and wreaking havoc in infrastructure and the economy.

This study identified six major factors that influence cyclone risk. Land use, slopes, rainfall, rise, population density, and soil quality. Incorporating these variables into risk models can improve prediction accuracy and lead to standard-based policy decisions.

Professor Biswajeet Pradhan, director of the Center for Advanced Modeling and Geospatial Information Systems at Sydney Institute of Technology (UTS), said failure to improve risk assessments could put the community at risk.

“Our review shows that risk assessments narrowly focus on specific dangers, such as storms and floods, rather than how multiple threats interact. This may mean that the community is not ready to the full extent of cyclone-related destruction,” Professor Pradhan said.

“Another important concern is that current assessments prioritize cyclone frequency over actual damage, despite their greater usefulness for policymakers. Only 5% of studies have investigated the effectiveness of mitigation measures and exposed blind spots in disaster restoration plans.”

Mitigation measures include actions such as improved building codes, coastal defense, early warning systems and land use planning. All of these can help reduce the impact of cyclones and protect the community.

The economic impact of cyclones is another area where existing assessments are lacking. The study points out that despite its potential for long-term economic harm, indirect effects are often overlooked, such as disruptions in business operations.

This Geoscience Frontiers study follows another study by Professor Pradhan, published in the Earth System and Environment, on the possibility of AI and machine learning-based risk assessments of cyclone-induced flood damage.

“The possibility of using machine learning in cyclone risk assessments is undeveloped,” Professor Pradhan said. “Integrating AI and machine learning can significantly improve prediction accuracy and resilience planning.

“Some studies use artificial intelligence, including random forest models and neural networks, but they have a range of exploring more advanced technologies such as ensemble models, allowing for greater accuracy and adaptability in different regions.

“These findings provide important insights that could shape future research and policies, and will help Australia and other cyclone-prone regions prepare for an increasing threat of extreme weather events in climate change,” he said.