The catastrophic wildfires that ravaged Northern California in August 2020 were sparked by lightning strikes, burning more than 1.5 million acres and claiming dozens of lives.
Israel, too, has seen destructive, fatal forest fires.
The most notorious was the one in December 2010 on Mount Carmel near Haifa. It spread quickly, consuming much of the Mediterranean forest covering the region. With a death toll of 44, it was the deadliest civil disaster in Israeli history until the 2021 Meron stampede.
While lightning has been blamed for numerous forest fires in Israel, the Mount Carmel one was apparently not caused by lightning. A 14-year-old resident of Isfiya admitted to inadvertently starting the fire after smoking a nargila and throwing a lit coal into an open area. He was so shocked by the result that he went back to school without telling anyone what he had done.
Now, a groundbreaking new artificial intelligence (AI) model developed by researchers at Bar-Ilan University (BIU) in Ramat Gan promises to revolutionize wildfire prediction, with a particular focus on lightning-induced blazes that are growing increasingly common due to climate change. The new AI model can predict where and when lightning strikes are most likely to cause wildfires, achieving over 90% accuracy – a first in wildfire forecasting that could transform emergency response and disaster management worldwide.Dr. Oren Glickman and Dr. Assaf Shmuel from the computer science department, in collaboration with experts from Ariel University and Tel Aviv University (TAU), worked with seven years of high-resolution global satellite data, alongside detailed environmental factors like vegetation, weather patterns, and topography, to map and predict lightning-induced wildfire risks on a global scale. Their research was recently published in Nature Publishing’s Scientific Reports under the title “Global lightning-ignited wildfires prediction and climate change projections based on explainable machine learning models.”
What makes this so significant is the ability to predict lightning-induced wildfires with remarkable precision. The AI model outperforms traditional fire danger indices by taking a global, data-driven approach. It integrates environmental factors, data from satellites, and weather systems to assess the likelihood of lightning-induced fires, overcoming the limitations of regional and data-restricted models.
It is fortunate, however, that in Israel, when the forests are dry in the summer, lightning storms are rare because there is usually no rain from May through September or even beyond.
This is not true in California or in Australia, which have suffered devastating forest fires that killed people and wildlife, Shmuel told The Jerusalem Post.
“Firebreaks in which trees and undergrowth are cleared between sections of trees can reduce the danger, but these are very difficult and expensive to maintain, he added, because they grow back quickly. Sending sheep to eat the undergrowth could help, however,” he said.
Wildfires becoming more frequent
As climate change accelerates, extreme weather events, such as lightning storms, hot and dry conditions, and shifting ecosystems, are contributing to more frequent and intense wildfires. While human activity is often responsible for igniting many fires, lightning remains one of the most unpredictable and deadly causes, especially in remote regions. These fires can smolder undetected for days, only to erupt into uncontrollable infernos before firefighters can respond.
Glickman added: “Lightning-ignited wildfires are a global challenge, and our models show they’re likely to intensify with climate change. We hope this research empowers countries around the world to better anticipate and mitigate these fires. With an improved ability to predict lightning fires: meteorological services, fire departments, and emergency planners can respond earlier and more effectively, potentially saving lives and protecting ecosystems. This model also addresses a key gap in existing wildfire prediction models: While many models are effective for fires caused by human activity, they struggle to predict lightning-induced fires that behave very differently and often start in hard-to-reach areas.”“We are at a critical moment in understanding the complexities of wildfire ignitions,” Glickman said. “Machine learning offers the potential to revolutionize how we predict and respond to lightning-ignited wildfires, providing insights that could save lives and preserve ecosystems.”While the AI model is not yet integrated into real-time forecasting systems, its development marks a critical step forward in wildfire prediction. “With the growing implications of climate change, new modeling tools are required to better understand and predict its impacts; machine learning holds significant potential to enhance these efforts,” Shmuel said.
“Our findings highlight significant global differences between anthropogenic [caused by humans] and lightning-ignited wildfires. Moreover, we demonstrate that, even over a short time span of less than a decade, climate changes have steadily increased the global risk of lightning-ignited wildfires,” the team wrote. “This distinction underscores the imperative need for dedicated predictive models and fire weather indices tailored specifically to each type of wildfire.The new machine learning models developed by the team have the potential to predict lightning-ignited wildfires worldwide, offering a powerful tool for fire mitigation and response. With an ever-increasing risk of wildfires driven by climate change, early detection and prediction are essential for protecting forests, wildlife, and human communities from the devastating effects of these fires.