Scientists have taken a big step forward in planning for bushfires by developing a computer model that could help predict when and where a fire will occur.
Dr Ritaban Dutta, of CSIRO's Data61, and research partners from the University of Tasmania say they used NASA satellite data on the location and intensity of fires across the world to create the prediction model.
It can detect and predict bushfires with more than 90 per cent accuracy in both grasslands and forests, two types of Australian vegetation prone to fire.
The model could be used to alert authorities about likely blazes and help develop plans for early evacuation, saving lives and properties.
The research, published on Wednesday in the journal Royal Society Open Science, reveals bushfires have surged in Australia by 40 per cent since 2007.
"Estimated average bushfire event frequency over the whole continent of Australia was 3284 per week in 2007, in comparison with the frequency of 5484 events per week in 2013," the paper states.
Dr Dutta says the figures - equating to 469 fires a day nation-wide in 2007 and 783 a day in 2013 - included various types of fires and everything from the smallest flare-ups to major blazes.
The scientists say the increased bushfire frequency reflects "a serious change in climate".
Further development is needed to turn the research model into a practical system.
