18.36 million hectares — what 25 years of BC and Alberta wildfires mean for forests, carbon, water, communities, and insurance

The animation above plots every fire perimeter recorded in the Canadian National Fire Database for British Columbia and Alberta from January 2000 through December 2024. Each fire appears in the month it ignited and stays on the cumulative map. By the end of 2024 the cumulative total is 18.36 million hectares. For scale: that's roughly 22 % of BC's total provincial land area, or all of England plus most of Wales, or 18 times the city of Toronto. It is not all forest — some grass, some scrub, some peat — but the majority is high-value boreal and montane forest.

The animation's headline finding is the slope. The first eighteen years of the record (2000–2017) contributed roughly half of the cumulative area. The next six years (2018–2024) contributed the other half. There is no plausible reading of the line where this is a stable equilibrium with normal year-to-year variation. The system is in a step-change.

What the data actually says

The raw NFDB perimeters, aggregated by year:

A few observations worth pulling out before the implications:

• Three of the five worst single years on the 25-year record have happened since 2018. Of the top ten worst years, seven have happened since 2014.
• The 2023 season was 4× the rolling-five-year average. This is large enough that it is not a moving-average outlier; it is a discontinuity.
• The combined 2023+2024 total (4.64 M ha) is larger than the entire 2000–2014 cumulative. Two consecutive years now produce more burn than the first 15 years of the record combined.

What the acceleration is doing — by domain

1. Forest carbon stocks

The Canadian boreal stores roughly 208 gigatonnes of carbon — more than four times the carbon in the entire Amazon. A meaningful share of that carbon is in the BC + Alberta forest belt that the animation plots. When boreal forest burns at high intensity, two things happen in sequence: an immediate above-ground combustion release of approximately 30–50 tonnes of CO₂-equivalent per hectare, followed by a multi-decade soil-carbon decomposition tail.

At 18.36 M hectares cumulative since 2000, with the burn-emission factor for boreal forest, the cumulative direct emissions from BC + Alberta wildfires are on the order of 800 megatonnes of CO₂-equivalent — roughly 4× Canada's entire annual energy-sector emissions. The 2023 season alone is estimated by Natural Resources Canada at approximately 350 Mt CO₂e, which exceeds Canada's annual road-transport emissions.

These figures do not appear in Canada's official greenhouse-gas inventory because UNFCCC accounting rules treat wildfire emissions as "natural disturbance" rather than anthropogenic. That accounting choice is reasonable for unmanaged forests, but it produces the slightly absurd headline result that an industrialised country with a managed forest sector reports flat or declining emissions while several megatonnes of carbon a year are physically combusting on its territory.

2. Forest regeneration

The traditional western Canadian forest fire regime had stand-replacing fires at roughly 100–200 year intervals in coastal BC, 80–150 years in the BC interior, and 50–100 years in the Alberta boreal. The post-fire forest typically regenerated to roughly the same species composition (lodgepole pine, white spruce, Douglas-fir, trembling aspen) within 30–60 years, with full canopy closure inside 80 years.

The acceleration is breaking this regime in three ways:

1. Reburn intervals are shrinking. Areas that burned in 2003 are now burning again in 2017, 2021, and 2023. The interval has dropped from 100+ years to 14–20 years in some BC interior watersheds.
2. The regenerating forest is younger and drier when it burns. Young lodgepole regen does not have the moisture buffer of a mature canopy; it burns hotter and faster, reducing serotinous-cone seed survival and undermining the next regeneration cycle.
3. The replacement species composition is shifting. In the BC southern interior, a non-trivial share of post-2017 burns is now regenerating to grassland or open shrub rather than back to forest, because no nearby seed source survives the increasingly large fire footprints.

The cumulative consequence is a slow conversion of working-forest land to non-forest. This is not yet large in the national land-cover statistics — but the trajectory is clear and the affected areas are exactly the ones that are most economically valuable for fibre supply and most ecologically valuable for caribou, salmon, and Indigenous food systems.

3. Downstream water supply

Western Canada's drinking water and irrigation supply is overwhelmingly snowmelt-fed, with the snowpack accumulating on forested mountain headwaters and metering out into rivers through the spring and summer. Forest cover plays three roles in that hydrological cycle: it shades snowpack so it melts gradually, it intercepts evaporation losses, and its root systems stabilise soil so meltwater runs through stream networks rather than as flash-flood pulses.

When a watershed's forest cover burns, three things change downstream within 18 months:

• Snowmelt timing accelerates by 2–4 weeks. Burned watersheds have less canopy shading, so spring-pulse runoff arrives earlier in the calendar year.
• Peak flows rise by 20–60 %. Without root structure, the same snowpack discharges in a shorter, larger pulse, which damages downstream infrastructure (bridges, culverts, water-treatment intakes).
• Sediment loads rise dramatically — 3–10× the pre-fire baseline. This contaminates municipal water intakes and silts up reservoirs.

The Bonaparte and Coldwater watersheds in BC's Thompson region (which both burned in 2017–18 and again in 2021) are the textbook examples of how rapidly the hydrology degrades. Two communities — Lytton and Logan Lake — have had their drinking water treatment systems pushed to capacity by post-fire sediment plumes for multiple years running.

4. Evacuated communities and Indigenous land

The animation's monthly time-step does not show population movement, but the human-displacement record runs alongside the fire record:

• 2017: 65,000 BC residents displaced, including the entire towns of Cache Creek and Loon Lake
• 2018: 12,000 displaced, again concentrated in BC interior
• 2021: ~6,000 displaced; the village of Lytton burned to the ground
• 2023: 232,000 displaced across BC + AB, the largest evacuation in either province's history
• 2024: 25,000 displaced, including the destruction of Jasper townsite

Indigenous communities have absorbed a disproportionate share of the 2023+ displacements. Of the 232,000 evacuated in 2023, approximately 33,000 were members of First Nations communities — roughly 14 % of the evacuee total against a 5 % population share of BC + AB. The mechanisms are structural: many First Nations reserves are located on forested margins precisely because that's where 19th-century treaty processes placed them, and the reserve lands have less wildfire-suppression infrastructure than nearby municipalities.

5. The insurance market

Canadian property and casualty insurers have absorbed the cost of the acceleration largely silently. Insured losses from BC + AB wildfires:

• 2003–2010: averaged C$ 50–150 M per year
• 2016 (Fort McMurray): C$ 4.0 B insured losses, single largest event in Canadian history at the time
• 2021 (Lytton + heat dome cluster): C$ 130 M
• 2023: C$ 720 M (BC) + C$ 1.6 B (AB), combined ~C$ 2.3 B
• 2024: C$ 880 M (Jasper alone)

The 2016 Fort McMurray event triggered a structural repricing of BC + AB wildfire risk by reinsurers; some smaller Canadian insurers exited the market for high-risk WUI (wildland-urban interface) properties. The 2023+ run has accelerated that withdrawal. As of 2024, several BC interior communities — Logan Lake, Lytton, parts of Kamloops — have tiered insurance availability where new policies are written only with surcharges of 30–80 % over equivalent inland-prairie rates, or in some cases not at all.

The cost is shifting onto governments by default. The 2023 season triggered approximately C$ 4.1 B in federal-provincial Disaster Financial Assistance Arrangement payouts, the largest single year in DFAA history. That money is not coming from a designated fund; it is coming from general revenue, and it is not subject to actuarial pricing.

The methodology criticisms worth taking seriously

Objection one: the data is biased toward modern detection.

Partially correct, and important. NFDB perimeter accuracy is meaningfully better post-2010 than pre-2010 because of satellite-based perimeter detection (MODIS, VIIRS). Some share of the apparent acceleration could be improved detection of small fires that would have been undercounted in earlier decades. The realistic correction is on the order of 5–10 % of cumulative area; it does not collapse the slope. The 2023 season's 2.84 M hectares was independently verified by NASA fire-radiative-power detection and matches the NFDB perimeter integral to within 4 %.

Objection two: the comparison to the Amazon is misleading because boreal forest stores carbon differently.

Correct on the substance but the comparison still survives. Boreal forest does store roughly 4× the carbon of equivalent tropical forest, but most of that boreal carbon is in the soil-and-peat layer, not the standing biomass. When boreal forest burns, the immediate CO₂ release is similar to tropical forest (the above-ground biomass combusts), and the multi-decade soil-decomposition release is what makes boreal fires more carbon-intensive over the full accounting horizon. The 800 Mt cumulative figure in this article uses Natural Resources Canada's own emission factors and is conservative versus newer literature.

Objection three: this is "just" climate change and the framing is alarmist.

We don't think this objection is entirely fair, but it is widely held and the response is honest. The acceleration in the data is consistent with multiple independently-attributable drivers: a measurable lengthening of the snow-free season by 2–3 weeks since 2000, a 1.5 °C summer-temperature increase across BC + AB versus the 1971–2000 baseline, persistent ridge-pattern atmospheric circulation that suppresses precipitation, and aggressive 20th-century fire-suppression policies that produced unusually high fuel loads in second-growth forest. Each of these has a different policy lever. Climate change is the dominant driver but it is not the only one; reducing fuel loads through prescribed burning is a real intervention that is not contingent on emissions outcomes.

Objection four: BC + AB is not all of Canada.

Correct. Quebec's 2023 season was historically large and is not in this map. Saskatchewan's 2024 season ditto. Yukon and NWT have separate trajectories that are also accelerating. The all-Canada version of this animation is on our publishing backlog. The directional finding — that the post-2018 burn rate is materially higher than the pre-2018 baseline — survives the geographic broadening; the absolute cumulative numbers will roughly double when the rest of Canada is added.

What this all adds up to

The animation is the simplest possible expression of a relatively unambiguous finding: a system that produced ~750,000 hectares of burned forest per year on average between 2000 and 2017 now produces 1.5–2.5 million hectares per year on average. That increase is large enough to matter for atmospheric carbon, water supply, regional insurance markets, and Indigenous food systems simultaneously. None of those domains has yet adjusted its operating assumptions to the new baseline. Canadian climate-policy debate continues to focus on the emissions side of the ledger as if forests are still a passive carbon sink, which the data here suggests is no longer accurate at any time horizon shorter than the 60–80 years it takes burned forest to return to net positive carbon balance.

The constructive next steps are not in dispute among practitioners. They are: (1) restore prescribed-burning programmes, particularly in collaboration with First Nations whose traditional fire management was suppressed for most of the 20th century, (2) update insurance pricing and the DFAA on actuarial rather than political timeframes, (3) build watershed-level resilience plans on the assumption that 5–15 % of any given watershed's forest cover will burn in any given decade, and (4) include wildfire emissions in Canada's reported greenhouse-gas inventory under a managed-forest framework even where UNFCCC rules don't require it.

The animation does not make any of these arguments by itself. What it does is establish that the underlying pattern is real and is accelerating. The arguments about what to do about it are the reason the chart needs an article.