Evrim Yazgin
Cosmos science journalist
As Europe, South America and Africa begin to cope with devastating floods, attention is turning to mitigation and response.
University of Melbourne researchers have developed a computer model which they say can predict flooding more accurately and 1000 times faster than current simulations.
The new model reduces flood forecast times from hours or days to just seconds. The Low-Fidelity, Spatial Analysis and Gaussian Process Learning (LSG) model is detailed in a paper published in Nature Water.
“Currently, our most advanced flood models can accurately simulate flood behaviour, but they’re very slow and can’t be used during a flood event as it unfolds,” says the University of Melbourne’s Professor Rory Nathan, who has 40 years of experience in engineering and environmental hydrology.
“This enables highly accurate modelling to be used in real-time during an emergency. It’s a game-changer.”
The new LSG model was tested on two vastly different, yet equally complex Australian river systems.
It was able to predict floods with 99% accuracy on southern Australia’s Chowilla floodplain. It did this in just 33 seconds, as opposed to the 11 hours it would take current advanced models. The new model spent just 27 seconds predicting floods in the Burnett River system in Queensland, compared to 36 hours for present models.
The new model also allows responders to adapt to changing and unpredictable weather conditions.
Using mathematical transformations and machine learning, the LSG model makes it possible to simulate uncertainty in weather forecasts, rather than focusing on the most likely scenario as current models do.
“This new model also has potential benefits in helping us design more resilient infrastructure. Being able to simulate thousands of different flooding scenarios, instead of just a handful, will help design infrastructure that holds up to more unpredictable or extreme weather events,” says Nathan.
In February–March 2022, floodwaters swept through New South Wales and Queensland. The disaster destroyed thousands of homes and saw at least 16 lives lost in a week. Last month, Tropical Storm Hilary drifted into parts of northwestern Mexico and California – regions which have had little exposure in recent decades to such torrential downpours.
As severe weather events become more common due to climate change, the need for early warning systems and real-time models which can inform emergency responses could be beneficial to the community and save lives.
