It makes perfect sense that a horrific wildfire season would come in the year 2020 on the heels of a pandemic. Dozens of major fires burned across North America in September, including 85 large uncontained fires and six contained fires across 12 states.
Active fires have burned more than 3 million acres already, and 41,417 fires have burned almost 5 million acres year-to-date. The severity of the wildfire season is on track to surpass the 10-year average.
Better understanding wildfires
Global warming is often mentioned as a contributor to the wildfires, but there are other factors, too. Increasingly, researchers are looking to apply new approaches in address the risk of wildfires. They include tools such as deep learning and artificial intelligence (AI) to better understand wildfires and to control their intensity.
The model could be used to reveal areas of greatest risk for wildfires
A new deep learning model uses remote sensing and satellite data to trace fuel moisture levels across 12 Western states, in effect tracking the amount of easily burnable plant material and how dry it is. After additional testing is complete, the model could be used to reveal areas of greatest risk for wildfires and to plan the best areas for prescribed burns. Led by a Stanford University ecohydrologist, the research was published in the journal Remote Sensing of Environment.
Recurrent neural network
The model uses data from the U.S. Forest Service’s National Fuel Moisture Database, which amasses plant water content information from thousands of samples. Using a ‘recurrent neural network,’ the system leverages the fuel moisture data to corroborate measurements of visible light and microwave radar signals from spaceborne sensors that are tasked with estimating fuel moisture measurements.
Newer satellites with longer wavelengths allow sensitive observations about moisture content deeper into the forest canopy. Estimates from the model are used to generate interactive maps that fire agencies may one day use to identify patterns and prioritize wildfire control estimates.
Researchers are also working to analyze the impact of better and more efficient firefighting on the size and frequency of wildfires. The theory goes: When firefighters extinguish smaller vegetation fires, a consequence is the creation of an environment where wildfires are larger and/or more frequent.
Natural cycle of regeneration
Older woods will naturally catch fire from the sun’s heat to make way for fresh growth
The theory is based on the premise that wildfires play an essential role in the periodic regeneration of forests. Older woods will naturally catch fire from the sun’s heat to make way for fresh growth. However, more efficient firefighting can disrupt the natural cycle and, along with global warming, aggravate the broader likelihood of larger and more frequent fires.
Researchers at the WiFire Lab in California and the University of Alberta in Canada are using artificial intelligence (AI) to analyze the environment and provide recommendations for prescribed burns that can save some parts of the forest without interfering with the natural cycle of regeneration.
Providing early warning of wildfires
Equipment operated by Pacific Gas and Electric (PG&E) caused 2018’s Camp Fire, the most destructive wildfire in California history. Because of the threat of sparking a wildfire, PG&E this year shut off power to 172,000 customers in Northern California on Labor Day weekend, for example. A concern is the threat of winds tearing down power line or hurling debris into them. Southern California Edison (SCE), another utility, warned that about 55,000 customer accounts could lose power.
California utilities SCE, PG&E and San Diego Gas and Electric are helping to fund a network of ALERTWildfire video cameras in California that will help to provide early warning of wildfires. Video cameras keep watch throughout five Western United States to provide early warning, and the number of cameras is growing fast.
