Extract from ABC News
In late 2019, as bushfires lit up Australia's east coast, Simon Jones received an urgent call from the Victorian Country Fire Authority (CFA).
The RMIT remote sensing expert and his colleagues, along with the NSW Rural Fire Service (RFS), were testing an algorithm to automatically detect bushfires from a feed of weather satellite image-data.
"I got a call from Victoria saying, 'We need this now'", Professor Jones recalled.
"They asked, 'What do we have to do to get it?'"
Soon the CFA had access to the algorithm's fire map, which it used to detect and monitor bushfire activity that summer.
The 2019/20 fire season was followed by a period of higher rainfall and lower fire risk, and attention turned to fighting the fires of the future.
Experts told the bushfire royal commission that Australia needs to invest in fire-detecting satellite technology.
A month later, Andrew Forrest's Minderoo Foundation announced a plan for rapid fire detection using satellites, to be ready by 2025.
Under the plan, satellite technology would be able to detect fires in a fraction of the time of the RMIT-developed algorithm.
Now, almost four years on from the Black Summer, another dangerous fire season is brewing for large parts of the country.
So, how are the new technologies progressing?
'Antiquated methods' can't fight climate-fueled fires
Government agencies, industry and academics gathered in Brisbane last week for the annual Australasian Fire Authorities Council (AFAC23) conference.
As the fire agencies sounded the alarm about the upcoming fire season, delegates wound their way through the product display area, where robotic dogs stamped their feet, new fire trucks sparkled, and satellite vendors showed the latest in orbital technology.
The keynote speaker, Jen Beverly from the University of Alberta, called for a radical rethink on how to predict, detect and manage bushfires, which global warming is making more dangerous.
"Historical data and antiquated methods can't be relied on to plan for the future," she said.
Extreme fires are expected to increase globally by up to 14 per cent by 2030 and 50 per cent by 2100, according to the UN.
Bushfire records have tumbled through the 2023 northern summer.
Experts say fires need to be detected more quickly, so they can be extinguished before they grow too large.
In Australia, most fires are still reported through triple-zero calls, and can burn for hours before authorities are alerted.
In that time, a lightning strike can become an inferno.
The RMIT-developed fire-detection algorithm, trialled in 2019/20, demonstrated the potential for rapid space-based detection.
It meant fires anywhere in Australia could be detected within less than half an hour of ignition, or at least before the triple-zero call.
It's now been rolled out to all fire-fighting agencies in Australia.
Several commercial systems have also hit the market, with access to data from more satellite feeds.
But fires still aren't being detected as fast as they could be.
All space-based fire detection shares a common problem.
And it isn't the software, it's the hardware.
Rapid fire detection hits a roadblock
In September 2020, Minderoo announced an audacious goal.
The non-profit would develop the technology and capabilities for Australia's firefighting agencies to extinguish dangerous fires anywhere in Australia within one hour.
This speed of response would require detecting the fires with a few minutes of ignition.
Minderoo said it would achieve this by 2025, pledging $70 million to what it dubbed mission "Fire Shield".
The plan was spruiked as the bushfire equivalent of the "moonshot" Apollo missions of the 1960s.
In the years since, Minderoo has made significant progress, Fire Shield project manager Rania Poullos told the ABC.
FireShield piloted a system of fire-detection cameras mounted on mobile phone towers, collaborated with fire agencies on a high-risk lighting alert system, fast-tracked the delivery of a national bushfire spread simulator, and tested new technology to map fuel loads, Ms Poullos said.
By 2025, Minderoo will have "tested and proven" the technologies needed to detect fires faster, she added.
But experts say it won't reach its 2025 goal.
Professor Jones, who's not involved with Fire Shield, said the technology was "nowhere near" good enough at the moment.
Being able to detect dangerous fires anywhere in Australia within minutes of ignition appears to be at least four years away, he said.
In June this year, Minderoo sponsored a four-year $16 million XPRIZE challenge to develop technologies necessary to achieve the Fire Shield mission.
The global competition has two categories:
- Teams will have one minute to accurately detect all fires across a landscape "larger than entire states or countries"
- Teams have 10 minutes to autonomously detect and suppress a high-risk fire in a 1,000 square meter area
Both these goals were "immensely challenging," Professor Jones said.
"It’s not just sensing, but AI and integration and getting information to the end user," he said.
Bushfire detection and response like Minderoo proposed for 2025 may be ready in four years' time, by 2027, he added.
"The XPRIZE is a four-year competition. This level of technology is where they're hoping to be in four years."
Why space-based fire detection is so hard
Pointing a satellite at the ground to detect fires is not as simple as it may sound.
For one, the satellites that are parked in one spot above the Earth, locked in geostationary orbit, are very far away.
The geostationary satellite that Australia uses for bushfire detection, Japan's Himawari-9, takes one photo of half of the Earth every 10 minutes.
It's so far out that a single pixel of this image is about one kilometre across (and even larger in the infrared bands).
This spatial resolution means it's difficult to detect a small, not very intense fire.
Then there's the other problem: time between images.
Factoring in the delay as the information is passed from Japan's metrological agency to Australia's, the time between images is about 20 minutes, Professor Jones said.
"It only takes 40 seconds for the algorithm to compute," he said.
"The lag is in the data provision."
The apparent solution to the spatial resolution problem is to deploy satellites that are closer to Earth.
Low-Earth orbit satellites (LEOs) are already used for communications, such as space-based internet services.
Unfortunately, fire-detecting LEOs have their own problems.
Their low and fast orbital path only takes them over the same spot on the surface once every few days.
Services like Starlink deploy constellations of thousands of satellites, so that a LEO is always overhead.
But there are only a handful of Earth observation satellites available for fire detection. This means there's long gaps in critical coverage.
"If you have a LEO satellite go overhead and it's cloudy at that point in time, you miss an observation," Professor Jones said.
Speaking at AFAC23, Rania Poullos said existing satellite capabilities were "not fit-for-purpose" and called for a global constellation of fire-sensing LEOs.
"Low-Earth orbit satellites with fit-for-purpose fire monitoring sensors and near-real time image processing and analytics can provide unprecedented situational awareness of bushfires at state, national or even global scales,” she said.
A history of disappointments
Constellations of satellites at this scale have been proposed before.
In March 2021, Queensland company Fireball announced plans to launch the country's first purpose-built fire-detecting LEO. It said it would have a constellation of 24 within five years.
But the promised launch never went ahead.
The company changed its name to Exci and backed away from the idea.
Speaking last week, Exci CEO Christopher Tylor said space-based detection using LEOs worked out to be too expensive.
To ensure constant coverage of Australia, the constellation would need tens of thousands of LEOs, he said.
"Satellites have a lot of applications in Earth observations, but fire detection is not one of them ... not right now," he said.
Other proposed constellations have also failed to launch.
In September last year, Minderoo CEO Adrian Turner proposed a constellation of hundreds of LEOs to be operational by 2025.
He reportedly said Minderoo was close to committing to using a fleet of LEOs, and the first satellites could be launched in 2022.
"We're really far down the line on this," Mr Turner said at the time.
Those satellites never launched, and, one year later, the future of their constellation idea appears less certain.
Ms Poullos said Minderoo would create opportunities for others to pursue the idea.
She said the non-profit was "supporting a constellation of global satellites" by assessing how well they work for bushfire detection.
"This work is still in progress," she said.
This year's devastating northern hemisphere fire season may finally push companies and governments to build a large LEO constellation for bushfires.
In June, the German company OroraTech said it would launch an eight-satellite constellation, to be operational by next year's northern summer.
In July, China announced plans for 300 remote-sensing LEOs by 2030.
Lack of funding holding back deployment of new technologies: Minderoo
But a global constellation that's large enough to be useful at detecting bushfires is still many years away.
It almost certainly won't be ready by 2025.
Ms Poullos said there was a "consistent lack of funding for national-scale projects".
"Through no fault of their own, our state-based fire agencies do not have access to the full remit of necessary systems and information to develop effective strategies for fire management, resource allocation and emergency response planning," she said.
"The challenge now is for policymakers to fund the development and ongoing costs of implementing those technologies for fire agencies."
Exci's Christopher Tylor agreed.
State and territory governments are primarily responsible for disaster response, while climate adaptation is a federal responsibility.
He said it it wasn't clear who was responsible for improving the country's capacity to rapidly detect bushfires.
"Fire detection was never a problem because up to now we've been relying on the public," he said.
"So it's a complicated subject."
A spokesperson from the National Emergency Management Agency (NEMA) said federal, state and territory governments have co-invested $94 million this year in bushfire mitigation and preparedness.
They said the federal government has also upgraded the emergency management platform used by government agencies and non-government organisations during a crisis.
"[This includes] obtaining satellite imagery from partner agencies as well as predictive analysis to anticipate likely challenges and associated impacts on communities," the spokesperson said.
In late September, the federal government will convene a national bushfire preparedness summit for the first time, bringing together governments, emergency services, industry and not-for-profit organisations.
"As natural hazards increase in intensity and frequency, the Australian Government has a responsibility to be prepared for any threats or risks that could become crises of national significance," the NEMA spokesperson said.
No comments:
Post a Comment