Wildfire

The Fastest Growing Insured Peril

Of all the natural perils covered in this module, wildfire has undergone the most dramatic transformation in its insurance significance over the past decade. What was once considered a manageable, geographically contained peril — primarily affecting rural and semi-rural areas of California and Australia — has become one of the industry's most pressing cat modelling challenges. Insured wildfire losses in the United States averaged approximately USD 300 million per year through the 1990s. By the 2017–2022 period, that average had risen to over USD 10 billion per year — a more than 30-fold increase in two decades.

This transformation is not simply a function of more houses in fire-prone areas, though that is part of it. It reflects a fundamental change in wildfire behaviour driven by climate change, decades of fire suppression creating dense fuel loads, and the expansion of the Wildland-Urban Interface (WUI) — the zone where human development meets unmanaged wildland vegetation. Understanding this transformation is essential for cat modellers working in any market with significant WUI exposure.

The Fire Triangle — Conditions for Wildfire

Wildfire requires three ingredients, often visualised as the fire triangle:

• Fuel: Combustible vegetation — grass, shrubs, trees — that is dry enough to ignite and sustain combustion. Fuel moisture content is the single most important determinant of fire behaviour: dry fuel ignites more easily, burns more intensely, and spreads faster than moist fuel.
• Ignition: A source of heat sufficient to ignite the fuel. Wildfire ignitions are overwhelmingly human-caused — power line failures, vehicles, campfires, arson — with lightning the primary natural source in some regions.
• Weather: The atmospheric conditions that determine whether an ignition becomes a small contained fire or a catastrophic conflagration. The key weather factors are wind speed (spreading fire rapidly and carrying embers), low humidity (drying fuel and promoting combustion), and high temperature (pre-conditioning fuel and reducing its moisture content).

Fire Weather — The Critical Driver of Catastrophic Events

Fuel and ignition sources exist across vast areas of fire-prone regions throughout the year. What separates a routine fire from a catastrophic one is almost always fire weather — specifically the combination of extreme wind, low humidity, and high temperature that creates conditions where fire spreads faster than suppression resources can respond.

In California, the critical fire weather driver is the Santa Ana wind — a dry, warm, gusty offshore wind that blows from the interior desert regions toward the coast, typically in autumn and early winter. Santa Ana winds can gust to 100–130 km/h, drive relative humidity below 5%, and carry fire in any direction independent of normal terrain effects. The Thomas Fire (2017, then the largest in California history), the Woolsey Fire (2018), and the Camp Fire (2018) all occurred during Santa Ana or similar diablo wind events. In northern California, similar conditions are created by Diablo winds — offshore winds channelled through the Coast Ranges.

In Australia, catastrophic fire conditions are associated with northerly winds blowing hot, dry air from the interior desert toward the densely populated southern coast, combined with temperatures exceeding 40°C and relative humidity below 10%. The 2019–2020 Black Summer fires — which burned 18 million hectares, killed 33 people directly, and caused an estimated AUD 103 billion in economic impact — occurred during an extended period of record drought and exceptional fire weather.

The Wildland-Urban Interface (WUI)

The Wildland-Urban Interface is the zone where structures and human development meet or intermingle with unmanaged wildland vegetation. It is the primary arena where wildfire causes insured losses, and its expansion is the single most important driver of increasing wildfire loss trends in the United States, Australia, southern Europe, and beyond.

The WUI can be categorised into two types:

• Interface WUI: A distinct boundary where developed land abuts wildland vegetation — a suburb at the edge of a forest or chaparral landscape. Fire approaching the interface must cross a relatively clear boundary.
• Intermix WUI: Developed land and wildland vegetation are intermingled — homes within a forested or shrubby landscape. Fire can spread through the vegetation among the homes themselves.

Intermix WUI represents the highest-risk configuration. When fire is burning within and around a community — as happened in Paradise in 2018 — evacuation becomes extremely dangerous, ember showers ignite multiple structures simultaneously, and suppression resources are overwhelmed. The Camp Fire destroyed approximately 19,000 structures in a matter of hours, with fire spreading through the town at a rate that left residents with minutes to evacuate.

Wildfire Behaviour Modelling

Wildfire spread modelling must capture the complex interaction between fuel, terrain, and weather — three factors that all vary continuously in space and can change rapidly over time.

Fire Spread Models

The foundational physics of fire spread in surface fuels is described by Rothermel's fire spread model (1972), which estimates the rate of spread and fire intensity as a function of fuel characteristics (moisture content, load, depth, surface-to-volume ratio), wind speed, and terrain slope. This model forms the basis of operational fire behaviour tools used by fire managers and, increasingly, by cat model developers.

Key principles from fire spread modelling include:

• Slope effect: Fire spreads more rapidly uphill — heat and flames are convected upslope, preheating fuel ahead of the fire. Fire spreading upslope can move 2–4 times faster than on flat terrain.
• Wind effect: Wind accelerates fire spread by supplying oxygen, tilting flames toward unburned fuel, and carrying embers ahead of the fire front. The relationship between wind speed and rate of spread is non-linear — doubling wind speed more than doubles fire spread rate.
• Spotting: Embers lofted by intense convective columns or carried by wind can land far ahead of the fire front, creating spot fires that grow and merge with the main fire, dramatically accelerating spread.
• Fire weather interaction: Intense fires create their own local weather — convective columns that draw in surface winds, create downbursts, and in extreme cases develop pyrocumulonimbus (fire-generated thunderstorm) clouds that can generate lightning, extreme winds, and erratic fire behaviour.

Stochastic Wildfire Modelling

Cat model stochastic wildfire event generation must represent the full range of plausible fire scenarios — from small contained fires to catastrophic conflagrations. This involves:

• Ignition simulation: Generating stochastic ignition points based on historical patterns of human-caused and lightning ignitions, accounting for spatial variation in ignition probability
• Fire weather sampling: Sampling from historical and projected distributions of fire weather conditions (wind, humidity, temperature) to drive fire spread simulations
• Burn probability mapping: For each simulated fire, estimating the probability that fire reaches each location in the exposure database given the ignition point and sampled weather conditions
• Fuel mapping: Using satellite-derived vegetation maps updated regularly to capture changes in fuel conditions — including post-fire recovery and drought-driven fuel drying

Wildfire Vulnerability — How Structures Are Lost

Wildfire vulnerability is fundamentally different from other perils in that structure loss tends to be binary rather than continuous — structures either survive a wildfire or are completely destroyed. This binary outcome is driven by the ember ignition mechanism: a structure either resists ember ignition (through hardened construction features) or it ignites and typically burns to the ground.

Key Structural Vulnerability Factors

Post-fire research — particularly the systematic surveys conducted after California fires — has identified the construction features most strongly associated with structure survival:

• Roofing material: Class A fire-rated roofing (tile, metal, asphalt shingle with fibre mat) is substantially more resistant to ember ignition than wood shake or shingle roofing. Roofing is the most important single feature for structure survival.
• Vent screens: Vents (attic vents, crawl space vents) are the most common ember entry point. Fine mesh (1mm or less) ember-resistant vent screens dramatically reduce ignition probability from ember entry.
• Deck and porch materials: Combustible wood decks and porches adjacent to the structure can provide a direct path for fire to reach the building envelope. Non-combustible or fire-resistant deck materials are strongly associated with structure survival.
• Zone 0 — the immediate 0–1.5m zone: Research shows that the single most predictive factor for structure survival is the immediate non-combustible zone within 1.5 metres of the structure — no mulch, no wood, no combustible material immediately against the building. Embers accumulating in this zone and igniting combustible material are a leading cause of structure loss.
• Window glazing: Standard single-pane glass can fail from radiant heat or direct flame contact, allowing fire to enter. Dual-pane tempered glass performs significantly better.

Community-Level Effects

An important and unique feature of wildfire vulnerability is the community-level effect — in which the vulnerability of individual structures is influenced by their neighbours. A home with excellent fire-hardening can still be destroyed if adjacent homes ignite and provide radiant heat and direct flame exposure. Conversely, in a community where most homes are well-hardened, a single vulnerable home igniting from embers may not spread to well-prepared neighbours. This community-level interdependence makes wildfire vulnerability inherently spatial and portfolio-level — individual property risk cannot be assessed in isolation from the surrounding neighbourhood.

The Insurance Market Under Stress

The scale of recent wildfire losses has placed the insurance market for WUI properties under severe stress in California, Colorado, and parts of Australia. Several major insurers have announced withdrawal from or severe restrictions on their California homeowners business — State Farm, Allstate, Farmers, and others have either stopped writing new policies or non-renewed existing policies in high-risk areas. This has driven hundreds of thousands of homeowners to the California FAIR Plan — the state's insurer of last resort — which itself faces capital adequacy concerns given its concentrated wildfire exposure.

This market stress has important implications for cat modellers:

• Anti-selection: As admitted carriers withdraw, the FAIR Plan and remaining insurers inherit an increasingly adverse portfolio of the highest-risk properties
• Data quality: Non-renewed policies mean historical exposure data rapidly becomes stale, complicating model validation
• Regulatory pressure: California's Insurance Commissioner has resisted premium increases that actuaries argue are necessary to sustain a viable private market, creating a tension between regulatory and market forces that is itself a source of risk for remaining writers

Climate Change and Wildfire

Unlike the uncertain relationship between climate change and hail, the connection between climate change and wildfire is direct, well-documented, and already materialising in observed loss data. The primary pathways are:

• Longer fire seasons: Rising temperatures and earlier snowmelt are extending the fire season in many regions. The California fire season, once concentrated in August–October, now effectively runs year-round.
• Increased drought severity: Higher temperatures increase evapotranspiration — the drying of vegetation and soil — even without changes in precipitation. This increases vegetation moisture stress and fuel drying, making fires more likely and more intense.
• Bark beetle infestations: Warmer winters and drought-stressed trees have driven explosive bark beetle population growth, killing tens of millions of trees across the western U.S. and Canada. Standing dead timber represents an enormous, highly combustible fuel load.
• Precipitation pattern changes: Many fire-prone regions are experiencing more intense wet seasons (promoting vegetation growth and fuel accumulation) followed by more intense dry seasons (drying that fuel to high flammability) — a pattern sometimes called the "green-brown cycle" that promotes large, intense fires.

Global Wildfire Risk Beyond California

While California has dominated recent wildfire insurance loss discussions, the peril is genuinely global:

• Australia: The 2019–2020 Black Summer fires burned 18 million hectares across New South Wales and Victoria — an area larger than many European countries. Australia has a long history of catastrophic wildfire and a well-developed (if still evolving) wildfire cat modelling market.
• Southern Europe: Greece, Portugal, Spain, and southern France experience significant wildfire each summer, with the 2017 Portugal fires (114 deaths) and the 2018 Attica fires near Athens (102 deaths) among the deadliest in European history. Insurance penetration in high-risk areas is generally lower than in the U.S. or Australia.
• Canada: The 2016 Fort McMurray fire destroyed 2,400 structures in a major oil sands city, causing CAD 3.8 billion in insured losses — the costliest natural disaster in Canadian history. The 2023 wildfire season was the most destructive in Canadian history, burning over 15 million hectares.
• South America: Chile, Argentina, and Brazil experience significant wildfire, with growing exposure in WUI areas and very limited insurance penetration.