A Tool to Help Identify Wildfires Faster (Credit: CNN Business via MSN)

DISCUSSION:  In general, a bigger and/or more intense wildfire is more difficult to contain and put out.  Thus, catching fires early when they are still small is critical to putting them out quicker.  Traditionally, most fires in the U.S. have been spotted by people who live nearby or from people in planes or look-out towers.  This non-automated method of identifying fires can be slow and may suffer from other issues as well.  For example, the fire may not be identified at all if it occurs far from populated areas.  In addition, observing smoke likely indicates a fire is present, but its exact location may be difficult to determine from observations of smoke from the ground.
 
Another way to detect fires is to use satellite data.  Satellite channels that are sensitive to reflected visible radiation can sense smoke from fires.  From the high-level perspective of the satellite, observations of smoke can more easily be traced to their source, unlike smoke observations from the ground.  In addition, thermal IR satellite sensors may be sensitive to the high temperature of a fire relative to its surroundings.  However, just like ground-based observations of fires, satellite-based observations also have drawbacks.  For example, smoke is difficult to sense above a highly reflective surface like snow.  The spatial resolution of the IR channels on the latest National Oceanographic and Atmospheric Administr​ation geostationary satellites (GOES 16 and 17) is ~2 km.  When fires first begin, they are much smaller than this.  Thus, when the IR channels measure an average temperature over the 2-km pixel, the signal from the larger, cooler area around the fire will overwhelm the fire signal, making the fire difficult, if not impossible to detect.
 
Recently, the use of artificial intelligence (AI) techniques in meteorology has been rapidly increasing.  These techniques can automatically, efficiently, and accurately identify patterns in large datasets like satellite datasets.  A company in New Mexico, Descartes Labs, has recently applied AI to satellite data in order to identify wildfires faster.  The AI can key in on subtle cues and trends in the imagery that indicate smoke from fires or the high temperature of a fire, for example.  The company claims they can detect fires as small as 10 acres in size in as little as 9 minutes from the time the satellite imagery is collected.  For example, this technique was able to identify the precise location of the California Kincade fire (pictured above on 27 October 2019) shortly after the fire started.
 
Any tool that can help identify fires more quickly and more accurately may allow first responders to arrive on-scene quicker and put out the fire faster, thereby minimizing impacts to people and property.  The AI method of identifying fires developed by Descartes Labs may be a promising tool for such identification of wildfires.

To learn more about fire weather and impacts of fires, be sure to click here!

©2020 Meteorologist Dr. Ken Leppert II