Technology: Can ‘Big Data’ help fight big fires? Fire fighters are betting on it
As out-of-control wildfires in the US West grow more frequent and more intense, fire departments in Southern California are looking to big data and artificial intelligence to enhance the way they respond to these disasters. The marriage of computing, brawn and speed, they hope, may help save lives. “In Los Angeles, with our population density, a fire is going to be burning houses down right away,” said Ralph Terrazas, the chief of the Los Angeles Fire Department. “There’s smoke, there’s fire, there’s sirens and we have to make decisions in just minutes.” With that urgency in mind, for about 18 months the department has been testing a programme developed by the WiFire Lab at the San Diego Supercomputer Centre that makes fast predictions about where active fires will spread next. The programme, known as FireMap, pulls together real-time information about topography, flammable materials and weather conditions, among other variables, from giant government data sets and on-the-ground sensors.
When fire fighters across the city are dispatched to respond to brush fires, the department’s leaders at headquarters now run the WiFire programme as part of their initial protocol. Then, WiFire’s servers at the San Diego Supercomputer Centre in La Jolla crunch the numbers and the programme turns out a predictive map of the fire’s expected trajectory. Those maps can then be transmitted electronically from headquarters to incident commanders on the ground.
Wildfires are still won and lost through gruelling, terrifying work in the field and some fires simply move too quickly for fire fighters to contain, with or without a supercomputer. But fire chiefs in the region believe the predictive technology could provide an extra tool when they need to make decisions quickly.
The programme can make sophisticated calculations in minutes that would take hours to run manually, said Ilkay Altintas, the chief data science officer at the San Diego Supercomputer Centre. The algorithm takes into account forecasts from the National Weather Service, vegetation readouts from the United States Department of Interior and even satellite data from NASA, among other data. “Human beings can probably do anything computers can do but not at the speed and scale that computers can achieve,” said Altintas. “We are limited by our brain and muscle capacity. The more data we can use, the more lives and property and businesses we can save.”
Computers have played an important role in emergency planning for decades, though the technology is used unevenly across fire departments. Now, new advances in computing and the expansive troves of public data, many of which were not easily available even a decade ago, are greatly expanding how these programmes can be used during emergencies.
Terrazas, who first approached WiFire in 2015 about collaborating, said he has advocated for the programme to expand statewide. He worked closely with the WiFire Lab to specify how the programme would be most useful to fire fighters. WiFire builds on a programme called FarSite, which was developed in the 1990s and is considered the gold standard in fire-risk modelling. FarSite and other programmes like it can take several hours to use, rendering it impractical for real-time use, according to Jason Loomis, a regional fire management specialist National Park Service.
“For fast-moving brush fires, especially in Southern California, we’re not turning to those models because by the time we get the results, the fire has often done what it’s going to,” Loomis said. He has not used the WiFire programme but said he was intrigued to learn about it recently.
Terrazas at the Los Angeles Fire Department has made the WiFire programme an integrated part of the department’s fire protocols. “It’s not the only, singular solution for fighting fires,” he said. “But it is one piece that help us make better, faster, calculated decisions.” That promise has caught the attention of other fire departments in the region. The Ventura County Fire Department signed up for the WiFire programme earlier this year and plans to put it to use for the first time this summer.
The Orange County Fire Authority, the Los Angeles County Fire Department, San Diego Fire Rescue and the State of California Public Utilities Commission are also partnering with WiFire.
The need for such tools, especially those that work quickly and simply, has grown with the sheer intensity of wildfires in recent years, driven in part by climate change and by the spread of residential development. California suffered a devastating fire season in 2018, with nearly 1,7 million acres burned across the state.
The Mendocino Complex fire, which started last July in Northern California, burned nearly 500 000 acres to become the largest wildfire on record in the state. The Camp Fire, also in the northern part of the state, became the state’s deadliest when it tore across 150 000 acres in November, killing 85 people and destroying nearly 19 900 structures.
Martha Witter, a fire ecologist with the National Park Service in Southern California, said it was crucial to reduce the lag time between when a fire starts and when fire fighters can inform the public about potential evacuations. She said that need is especially great in Southern California, where dried-out vegetation and accelerating development along the “wildland-urban interface” combine to make the region ripe for devastating fires. “In Southern California, we have perfect conditions for a fire almost every single year,” Witter said. “So as these areas get developed, it just increases the number of targets.”
Max Moritz, a wildfire expert affiliated with the University of California, Santa Barbara, who has a long career in wildfire management, said improvements in predictive modelling are crucial for fighting fires, especially as the fires become more intense. But he noted that predictive technologies will not change the underlying factors of urban development and a warming planet that are making fires more intense. “We need better data and better models but we also need better preparation,” Moritz said. “We also have to make headway on all the other fronts, if we really want resilient communities in the face of climate change.”
Source: New York Times