Photo by
Shalev Shalom/TPS on 15 August, 2021
Lightning and Wildfires: How Israeli AI Research is Changing Disaster Prediction
Jerusalem, 3 April, 2025 (TPS-IL) -- Israeli researchers have developed an artificial intelligence model that can predict lightning-induced wildfires with unprecedented accuracy, potentially transforming wildfire forecasting and disaster management. Essentially, this is the first time an AI model has demonstrated such high precision in forecasting lightning-triggered wildfires on a global scale, they announced on Thursday.
The model, created by scientists at Bar-Ilan University in collaboration with researchers from Ariel and Tel Aviv Universities, demonstrated an accuracy rate exceeding 90% in predicting where and when lightning strikes could ignite wildfires. Dr. Oren Glickman and Dr. Assaf Shmuel from Bar-Ilan University’s Department of Computer Science spearheaded the project, utilizing seven years of high-resolution satellite data, along with environmental variables such as vegetation, weather patterns, and topography.
Their findings were recently published in the peer-reviewed Scientific Reports journal.
“Machine learning offers the potential to revolutionize how we predict and respond to lightning-ignited wildfires,” said Glickman. “With the growing implications of climate change, new modeling tools are required to better understand and predict its impacts.”
Unlike traditional fire danger indices that rely on regional and often limited data, the AI model incorporates a vast array of global datasets to more accurately assess the likelihood of wildfires triggered by lightning. The model’s effectiveness was tested using wildfire data from 2021.
Extreme weather conditions, including hotter and drier climates, shifting ecosystems, and more frequent lightning storms, have created conditions where wildfires can ignite and spread rapidly. Lightning, in particular, poses a unique challenge as it can spark fires in remote regions where they may smolder undetected for days before erupting into large-scale infernos.
One of the most devastating examples occurred in August 2020, when lightning strikes ignited wildfires in Northern California that burned over 1.5 million acres and resulted in significant loss of life and property. The new AI model could provide a critical tool for predicting and mitigating such disasters.
“This model addresses a major gap in existing wildfire prediction methods,” Shmuel explained. “While many models perform well in predicting fires caused by human activity, they struggle with lightning-induced fires, which behave very differently and often start in inaccessible areas.”
With the potential to aid meteorological services, emergency responders, and fire management agencies worldwide, the model could significantly enhance early warning systems. By identifying high-risk areas before ignition occurs, authorities may be able to allocate resources more effectively and implement preventive measures.
Although the AI model has yet to be fully integrated into real-time forecasting systems, researchers believe its development marks a crucial step forward in wildfire prediction. As climate change continues to drive extreme weather events, tools like this could become essential in safeguarding communities, ecosystems, and wildlife from the devastating consequences of wildfires.
“We are at a critical moment in understanding the complexities of wildfire ignitions,” said Glickman. “The insights gained from machine learning could help save lives and protect natural landscapes from the increasing threat of wildfires.”
As research continues, the team hopes to refine and expand the model’s capabilities, working towards a future where AI-driven forecasting can provide real-time wildfire risk assessments.
