A new study shows machine learning—computer algorithms that improve themselves without direct programming by humans—can improve lightning forecasts. A better understanding of where lightning can strike could help predict fires started by the bolts from the sky. “Combining remotely sensed data with information, such as ground truth from previous fires, vegetation health, and dryness, AI can offer the opportunity to improve wildfire monitoring and forecasting of wildfire propagation,” Scott Mackaro, vice president of science, innovation, and development at weather forecasting company AccuWeather, who was not involved in the study, told Lifewire in an email interview.
Predicting Danger
Improved lightning forecasts could help prepare for potential wildfires and improve safety warnings for lightning. “The best subjects for machine learning are things that we don’t fully understand. And what is something in the atmospheric sciences field that remains poorly understood? Lightning,” said Daehyun Kim, a professor of atmospheric sciences at the University of Washington who was involved in the recent study, said in a news release. “To our knowledge, our work is the first to demonstrate that machine learning algorithms can work for lightning.” The new technique combines weather forecasts with a machine learning equation based on analyses of past lightning events. The study’s authors said that the hybrid method could forecast lightning over the southeastern US two days earlier than the existing leading technique. Researchers trained the system with lightning data from 2010 to 2016, letting the computer discover relationships between weather variables and lightning bolts. Then they tested the technique on weather from 2017 to 2019, comparing the AI-supported process and an existing physics-based method, using actual lightning observations to evaluate both. AI can help process the data received from satellite systems, single out false alarms, and remove them, weather expert Yuri Shpilevsky of the app Clime told Lifewire in an email interview. “Besides that, AI can help track the weather parameters in different regions and detect the smaller areas where the weather conditions are the ‘most favorable’ for a fire to get started,” he added. This may help us automatically focus on the driest and thus most fire-prone places and conduct fire prevention activities there."
Putting Theory Into Practice
Artificial intelligence is already being used to help monitor for wildfire danger. The Aspen Fire Protection District uses an AI-driven program that leverages cameras to monitor smoke reports over 90 square miles in Colorado. The program is made by a California-based company called Pano AI and uses high-resolution cameras that can swivel 360 degrees. “We know that minutes matter when it comes to wildfire response,” said Arvind Satyam, Pano AI’s chief commercial officer, in a news release. “Our vision is to create a network of cutting-edge cameras, as well as integrate existing video feeds, that leverage our artificial intelligence and our intuitive software to provide timely and accurate alerts for situational awareness teams to prevent small flare-ups from becoming large infernos.” Many companies are using AI to improve weather forecasts. For example, Weather Stream uses AI to monitor precipitation from global satellite data, indicating drought regions. “AI and satellite data can be used at multiple stages of the wildfire cycle,” Richard Delf, a remote sensing scientist at Weather Stream, told Lifewire in an email interview. “We can use AI to interpret satellite data to establish regional fuel levels, levels of moisture at the surface, and canopy levels, which, along with local climate, are key indicators of the wildfire risk of a region.” Future advances in AI will make wildfire forecasting even more accurate, predicted Shpilevsky. Computer models will make predictions based on weather conditions and other data, such as a forest’s vegetation type, wind patterns, conditions favorable for lightning strikes. “This will help provide real-time forecasts on the way a wildfire is going to spread, predict the expected fire intensity, evaluate the possible damage, estimate the resources necessary for localizing the fire,” he added.
