Subscribe

Article · Environment · 4 min read

What Alberta's wildfire cause records actually show

You asked us for an arson versus lightning breakdown of the wildfires we mapped. Only Alberta publishes one. It shows arson is a rounding error by area, 95 per cent of it a single fire, and a fifth of what Alberta maps has no recorded cause at all.

Published 26 July 2026 · Open Data Canada Source Alberta Wildfire, BC Wildfire Service
Map of Alberta showing 8,728 wildfire perimeters 2006 to 2024 coloured by recorded cause, beside a ranked legend: lightning 4,632,028 hectares, no record 1,568,097, investigated with no cause found 1,339,487, industry 226,016, arson 209,177, other human 22,651
Every wildfire perimeter Alberta mapped, 2006 to 2024, coloured by the cause on file. Perimeters and cause records: Alberta Wildfire. Boundary: Statistics Canada.

When we mapped every wildfire perimeter in British Columbia and Alberta, the question that kept coming back was the one the map could not answer. How much of it was arson?

Only Alberta publishes a category for deliberately set fires. British Columbia separates a lightning strike from an arsonist with one word, "person", and stops. So we pulled Alberta's cause records, 26,551 fires over nineteen years, and joined them to the perimeters.

Arson is one fire

Alberta codes deliberately set fires as "incendiary", a category broader than arson that sweeps in every case nobody was charged for. There were 2,324 between 2006 and 2024, and the 285 big enough to leave a mapped perimeter burned 209,177 hectares, 2.6 per cent of everything the province mapped.

Almost all of it is one event. The McMillan Complex, northwest of Slave Lake in 2019, accounts for 199,073 hectares, 95 per cent of every arson hectare in the province. Every other arson fire in nineteen years comes to about 10,100 hectares. The label is soft even there: Alberta logged McMillan's activity as arson but left the true-cause field blank, so it was never confirmed.

Powerlines burn more than every arson fire but one

Alberta splits human causes into twelve categories. Ranked by hectares they look nothing like the argument online.

Cause Hectares burned, 2006 to 2024
Lightning 4,632,028
No cause on file 1,568,097
Investigated, no cause found 1,339,487
Industry 226,016
Arson (incendiary) 209,177
Other human 22,651

Industry burned 226,016 hectares, more than every arson fire combined, and not because six categories were bundled. Powerlines alone burned 138,051 hectares, thirteen times the arson record once McMillan is set aside, and unlike McMillan they are spread across the whole period and the whole province. Oil and gas adds 43,793. Forestry, blamed most often by readers last time, ignited 58 mapped fires for 17,849 hectares. Whether logging or decades of fire suppression change how a fire behaves once it starts is a question this data can't answer.

People keep asking about the government. Fires Alberta attributes to itself burned 772 hectares across 23 fires, escaped prescribed burns another 1,793 across five: 0.03 per cent of the mapped area, from Alberta's own file.

Map of Alberta showing all 8,728 mapped wildfire perimeters 2006 to 2024 coloured by recorded cause, with a ranked legend giving hectares and perimeter counts for each cause
All 8,728 mapped perimeters, ranked by area. Lightning covers the north, the two classes with no answer are the next two largest, and arson is one red mass west of Fort McMurray. Areas measured on dissolved perimeters in Canada Albers Equal Area (ESRI:102001).

Count fires instead of hectares and it flips

The counter-argument holds up. Inside the zone Alberta patrols, people start 61 per cent of the fires and lightning 35. Every headline saying most wildfires are human-caused comes from there.

Those fires also burn 7 per cent of the area. Human fires start near roads and towns, where crews reach them first: the average incendiary fire burned 101 hectares, the average lightning fire 495. We lead with area because area is what the season totals and the smoke measure.

Pie chart of Alberta burned area by recorded cause, 2006 to 2024, with the three human causes magnified into a bar breakout
All three human causes together are 5.7 per cent of the burned area.

A fifth has no cause on file

Alberta's detail comes with a hole. Of the eight million hectares it maps, 1,568,097 have no cause record at all, and another 1,339,487 were investigated without a conclusion. Alberta cannot name a cause for 36 per cent of the area on its own map.

The first number is jurisdictional. Alberta's cause database covers only the Forest Protection Area, so the 1,003 perimeters outside it have nothing to join to. Most of that is federal: 950,211 hectares sit inside national parks, dominated by Wood Buffalo. Parks Canada investigates those, so the causes exist, just not in Alberta's file.

The gap lands in the years that matter. The 2016 Fort McMurray fire, which pushed almost 88,000 people out of their homes, is still officially uncaused; investigators got as far as "probably human" and stopped. Ninety-five per cent of Alberta's 2016 burned area sits under investigation, and in 2023, the worst season on record, about 40 per cent carries no usable cause. Nothing in that hole supports a bigger arson number, and nothing rules one out.

Area burned by recorded cause per year in Alberta, 2006 to 2024
Alberta names a cause for 64 per cent of its burned area. The grey swallows its worst years: 2011, 2016 and 2023.

British Columbia has the opposite problem. It names a cause for 99.5 per cent of what burns and the record is still nearly useless, because lightning, person and unknown is the whole vocabulary.

What would fix it

Alberta already flags which fires burned on park land. It could carry the Parks Canada cause across. BC could publish what its investigators find and never release.

Until then, anyone putting a number on how much of Canada burned because someone lit it is filling a hole in the record with something they brought with them. Lightning burns most of this country. People start most of the fires and burn very little of it. Set aside the one fire carrying Alberta's arson total and powerlines burned thirteen times what arsonists did. And for a fifth of what Alberta maps, nobody wrote it down.

Sources & data

Sources

  • Alberta Wildfire, Historical Wildfire Perimeter Data 1931 to 2025 (shapefile and ESRI metadata) — alberta.ca
  • Alberta Wildfire, Historical Wildfire Data 2006 to 2025 (fire-level cause records) — open.alberta.ca
  • BC Wildfire Service, Historical Fire Polygons (WFS, PROT_HISTORICAL_FIRE_POLYS_SP) — openmaps.gov.bc.ca
  • Natural Resources Canada, Canadian National Fire Database, point records (NFDB_point_20260527)
  • Natural Resources Canada, Canada Lands Survey System, National Park legislative boundaries
  • Statistics Canada, provincial cartographic boundary file (lpr_000b21a_e)

Methodology. Window is 2006 to 2024, the period where both provinces publish comparable cause records. Areas are computed from dissolved fire perimeters in Canada Albers Equal Area (ESRI:102001). BC perimeters include the full extent of fires crossing into Alberta, so they were clipped to the BC boundary before measurement; Alberta perimeters exclude cross-border area by their own coding, apart from a residual 110,000 hectares (about one per cent) where in-province-coded perimeters spill over the line. Alberta perimeters were joined to fire-level cause records on fire number plus year, because Alberta reuses fire numbers annually. Perimeter area exceeds official burned-area statistics for both provinces because dissolved perimeters include unburned islands; use provincial season totals for headline burn figures and these numbers only for the cause split. Ignition-count and area shares in the "count fires instead of hectares" section are computed on fires inside Alberta's Forest Protection Area, the same population the cause database covers, so the two measures are like-for-like: 35.0 per cent of ignitions and 72.0 per cent of area are lightning, 61.3 per cent of ignitions and 7.1 per cent of area are human. The national-park share of the missing-cause gap was checked two ways, against Alberta's own BURNCODE park flag (60.6 per cent) and geometrically against NRCan park boundaries (60.5 per cent); the flag was read per polygon rather than collapsed to one value per fire, because 339 fires carry more than one code. The remaining 617,886 hectares outside the parks were then tested against the Canadian National Fire Database. A perimeter counts as federally recorded if a same-year national point with a cause of human, natural or prescribed burn falls inside it and its recorded fire size is at least 25 per cent of the perimeter's area, the size test excluding small unrelated fires that happen to sit within a large boundary. Seven perimeters totalling 71,132 hectares clear that bar, almost all of them Wood Buffalo fires Parks Canada holds, leaving 546,755 hectares with no cause at any level. Dropping the size test would remove a further 110,734 hectares from that uncaused total. Counts of arson and incendiary fires come from the fire-level records, not the perimeters. Average fire sizes are Alberta's own CURRENT_SIZE field.