The winter wildfires that rapidly ignited and spread through Los Angeles from January 7 to 8, 2025, marked a devastating event, with significant implications for communities, insurers and reinsurers. The fires caused widespread destruction, resulting in more than 18,000 destroyed structures and the tragic loss of 29 lives. Early estimates indicate insured losses could exceed $40 billion.
On the morning of January 7, following eight months of severe drought, strong Santa Ana winds reaching over 60 miles per hour quickly escalated a small brush fire in the Santa Monica Mountains into an extensive urban conflagration. Within half an hour, the fire grew twentyfold, spreading into Pacific Palisades, one of the world’s most expensive real estate areas (Figure 1).
Simultaneously, approximately 40 miles away in the suburbs of Altadena, the Eaton Fire ignited near a high-voltage transmission tower in Eaton Canyon. Southern California Edison, the local electricity company, had proactively shut power to residential lines due to high winds, but power at the high-voltage transmission towers in the hills remained on. Although the exact cause is still under investigation, preliminary analysis points to an electrical arcing event as a potential source of ignition.[1]
The emergency response was overwhelmed: Strong winds grounded firefighting aircraft, hydrants ran dry, fire crews were stretched beyond capacity, and evacuation routes were blocked by abandoned cars. Ultimately, the sheer speed, scale and simultaneous nature of multiple large fires made containment impossible. In total, nine fires ignited and burned an area of 39,000 acres from January 7 to 9.
The Los Angeles wildfires were catastrophic but not surprising; drier conditions, shifting seasons and expanding urban interfaces have steadily raised wildfire risk for decades, making it easy for anyone relying on wildfire models to be caught off guard if they’re not asking the right questions.
To ensure models remain relevant and robust in this fast-evolving risk landscape, insurers, reinsurers and risk managers must critically evaluate the models they use — whether deterministic weather risk scores such as the Fire Weather Index (FWI), physically-based spread and behavior models or catastrophe models[2] — by consistently asking these three questions:
01
The coastal part of Southern California has long experienced catastrophic wildfires, but the past two decades have seen a marked rise in both the total area burned and the frequency of fires exceeding 10,000 acres. Timing has shifted, too, with historically rare winter wildfires now becoming increasingly common. Already in 2025, California has recorded 182,197 acres burned across 3,938 fires,[3] up 119% and 8%, respectively, from the five-year average at this point in the year, with the Palisades and Eaton fires contributing significantly.
Research shows that with current global warming (around 1.3°C above preindustrial levels), extreme fire weather conditions, like those seen in Los Angeles, have become 35% more likely and about 6% more intense.[4] The dry season is now roughly 23 days longer than in past decades, creating overlap between prolonged dryness, fuel accumulation and intensified Santa Ana winds.
Models that fail to account for these climatic shifts, particularly those relying solely on historical baselines, will underestimate future wildfire losses, leaving (re)insurers unable to price and manage wildfire risk accurately.
02
Wildfire behavior is highly sensitive to vegetation conditions, land use and moisture availability, including factors such as fuel load, vegetation density and the vapor pressure deficit, which directly affects how quickly vegetation loses moisture and becomes more flammable. These characteristics are not static; urban expansion, deforestation and vegetation management practices continually reshape fuel availability and fire potential.
The Los Angeles fires illustrate how these conditions can combine in unexpected ways. Two preceding relatively wet winters allowed vegetation to flourish across Southern California, creating abundant fuel. This was followed by an exceptionally dry final quarter of 2024, leaving behind low fuel moisture, high fuel continuity and widespread dead vegetation.[4] When intense Santa Ana winds arrived, fires spread rapidly and proved exceptionally difficult to contain.
Models that rely on static or outdated vegetation and land-use data can miss precisely these kinds of compounding, fuel-driven fire dynamics, particularly when projecting future scenarios in rapidly changing urban and wildland-urban interface areas, such as suburban Los Angeles. Frequent updates are essential to ensure wildfire models accurately reflect present-day risk.
03
The most severe wildfire losses often occur at the wildland-urban interface, where embers ignite homes and fires spread through built environments. This transition remains underrepresented in many models, especially those designed for forecasting rather than risk assessment.
Urban fire spread behaves very differently from wildland fire. Dense housing, narrow streets, varied construction materials and the operational realities of emergency response (where saving lives may take precedence over protecting structures) all contribute to rapid escalation. Embers collect in such places as gutters, under eaves, in vehicles and in refuse bins. Fires can leap across roads and spread quickly from house to house. The Los Angeles fires showed how urban fuels such as fences, decking and cars can sustain fire in ways many models overlook.
Many models also rely on simplifications such as constant wind direction or static ember intensities throughout a fire day. These assumptions overlook the real-world complexity of shifting winds and fluctuating ember showers — factors that can significantly affect fire behavior and spread.
A comprehensive wildfire model should simulate how ember-driven fires enter urban areas, how suppression delays influence growth and how neighborhood layout affects propagation. It should also reflect regional differences in firefighting resources and response times.
Critically, models must be validated against real world events. Without testing assumptions and outputs against observed fire behavior and losses, even detailed models risk underestimating exposure, leaving (re)insurers vulnerable to unanticipated losses in high-density areas.
