Southern California Edison AI wildfire forecast underestimated Eaton Canyon fire by factor of 10
Southern California Edison's AI-driven wildfire forecasting system underestimated the size of the January 2025 Eaton Canyon fire by a factor of ten, predicting 404 hectares burned in eight hours when it ultimately consumed 5,600 hectares and killed 17 people. The utility used the inaccurate forecasts to inform power management decisions, and numerous lawsuits claim the decision to keep power flowing contributed to the fire's ignition.
What happened
Southern California Edison's internal wildfire forecasts underestimated the potential size of the Eaton Canyon fire in Los Angeles by a factor of ten in the days leading up to the January wildfire threat, according to documents reviewed by Reuters.
SCE's simulations predicted a January 7 ignition in Eaton Canyon that could scorch about 404 hectares within eight hours without fire suppression, but the Eaton fire ultimately consumed some 5,600 hectares.
The utility's fire spread simulations were factored into SCE's power shut-off decisions as strong seasonal winds and dry conditions escalated the looming wildfire threat.
Reported impact
The Eaton fire destroyed around 9,400 homes and buildings and killed 17 civilians, making it the centerpiece of one of the costliest natural disasters in US history.
Numerous lawsuits have claimed SCE's decision to keep power flowing to some lines and towers in the Altadena area led to the circumstances that triggered the fire.
Index analysis
SCE's fire modeling capabilities suggest potential weaknesses, as the system was upgraded with improved computing, datasets and artificial intelligence yet failed to accurately forecast fire spread.
Dr Joseph Mitchell, a scientist and wildfire expert witness for California utility regulators, said SCE's predictions missed the mark mainly because its models were only running simulations that extend eight hours after an ignition, with bulk of damage occurring well after that period.
Governance controls
SCE defended its modeling capabilities and said the cause and circumstances around the fire are under investigation. SCE told Reuters it would evaluate whether to use longer fire simulation periods.
Raymond Fugere, SCE's asset intelligence director, said variations in wind patterns and available fuels in hard-hit areas may not have been fully accounted for in the fire spread modeling.
Classification
Sources and evidence
This record was researched and written by the Index, based on the following original reporting.