Fighting Fires With AI

Written by Matt Harrison
Posted November 19, 2018

The people of California thought that 2017 was the worst wildfire season they'd ever seen. In 2017, 44 people lost their lives when the Santa Ana winds carried the flames through the urban areas of Northern California. Over 200,000 acres of land were destroyed.

Now, in 2018, the wildfire season's death toll has already risen to at least 80. And over a thousand citizens are still unaccounted for. That's 800,000 acres lost and counting. And 8,500 structures have already been destroyed. California's worsening air quality and increasingly dire droughts show that this will only happen on a larger scale, and more frequently, in the future.

The situation looks pretty grim. But there's still hope in the smog...

The Californian risk-assessment models underestimated the potential for fires in the 2018 dry season.

These underestimates of the 2018 wildfire season's capacity for destruction was the result of a few different factors. The largest of which was human error.

Fires are almost impossible to control but even harder to predict. No inspector can account for wind strength and direction. And they can't determine the flammability of the environments that the wildfire will encounter.

The maps created by the California Department of Insurance's (CDI's) fire hazard team don't account for homeowners' living situations on an individual basis. As in, the CDI's maps don't know which brush has been cleared, where lush gardens have been planted, or what other flammable changes have been made to the property.

What's more, the California maps in circulation are out of date. Due to the high cost and low efficiency of sending fire hazard inspectors from home to home, there haven't been new maps written since the early 2000s. And it's a lapse in accuracy that California is paying for today.

But there is a solution. When this disastrous wildfire season is over, Californians will be more prepared than ever before for the next fires. The availability of high-resolution aerial photography from drones could revamp how we assess fire hazards.

It's been a combination of three major West Coast brain trusts. The SILVIS Lab at the University of Wisconsin, the robotics labs at the University of California, San Diego, and the University of Corsica, have been working on aerial programs to identify potential fuel sources for wildfires.

It's the same sort of problem-solving infrared vision in driverless cars. The drones have been programmed to recognize potentially hazardous roof shapes, building density, and the presence of firebreaks. These drones will be able to understand how much vegetation has grown around a house. And they'll be able to check the quality and flammability of the vegetation. And of course, the drones are far cheaper and more effective than human inspectors.

These drones will be coupled with the wind and weather predictions from the U.S. Forest Service's (USFS') Santa Ana Wildfire Threat Index (SAWTI). Researchers believe that by using drone maps and the Santa Ana meteorological predictions, they'll be able to forecast wildfire threats up to six days in advance.

And drones will be at the forefront of the action. In addition to their mapping programs, these labs have developed programming that will let the drones map the safest routes for firefighters in real time. They'll be able to identify small fires through thick smoke with infrared. And they'll even be able to complete tasks that are as simple as dumping water.

When these current blazes die down, California may be able to look toward smog-free skies and a brighter, safer future.

Invest well,

Matt Harrison
Contributing Editor, Park Avenue Digest

Report: 5 Simple Rules
for Investing in a Bear Market