Finding fires faster

The development of new and innovative algorithms are supporting near-continuous active fire surveillance from space unlike any other satellite hotspot products previously available.  

Using the latest geostationary satellite-based earth observation systems and the Himawari satellite, the Fire surveillance and hazard mapping team from RMIT University, led by Prof Simon Jones and Dr Karin Reinke, will help fire managers with early fire detections to hone in on bushfires.

Most satellite-based fire detection algorithms are susceptible to the effects of clouds, as well as the accuracy of the land surface temperatures observed around a potential fire. But thanks to the research into an algorithm designed to take advantage of the 10-minute observations available from Himawari-8, and that is specifically tuned to Australian conditions and seasons, a robust and computationally rapid method for early fire detection across Australia has been developed.

Simeon Telfer is a fire manager from South Australia’s Department of Environment and Water, and says the research can make a difference to operations.

“Due to the increased availability of the satellite data and faster processing, there is an opportunity for earlier detections, and for ongoing remote observations of fires to be made,” Simeon says.

This means some fires could be detected hours earlier than was previously possible, leading to quicker deployment of firefighters and firefighting aircraft, as well as warnings to the public. 

This is being tested with a trial by the New South Wales Rural Fire Service over summer 2018/2019. Currently bushfires are primarily detected when a member of the public calls Triple Zero, and occasionally from other satellites that may be passing over the area. Working with the research team, RFS will assess how much faster the new algorithm can detect fires compared to current methods. 

The project is also improving the accuracy of vegetation monitoring for flammability, as well as saving critical personnel hours, through the development of a smartphone application. The Fuels3D app combines off-the-shelf digital cameras and/or smartphones with computer vision and photogrammetric techniques to calculate vegetation structure and fuel hazard metrics. This reduces individual bias in estimating bushfire risk and ensures more accurate and consistent data is collected as individual bias is completely removed. Fuels3D allows anyone to take a vegetation fuel sample; it has potential for pre- and post-burn mapping, and can provide inputs into fire behaviour modelling and risk assessment and planning.