# Fast fires: chart sources and scope

Source checked: 29 October 2026.

Balch, J. K., and colleagues (2024). *The fastest-growing and most destructive
fires in the US (2001 to 2020).* Science 386, 425–431.
[Published paper](https://doi.org/10.1126/science.adk5737).
[PubMed record](https://pubmed.ncbi.nlm.nih.gov/39446944/).
[Accessible copy of the published paper](https://eco-integrityalliance.org/wp-content/uploads/2024/10/balch-et-al-2024-fastest-growing-and-most-destructive-fires.pdf).

## Data used

The figures come from page 3 of the paper (journal page 427). The analysis covers
the contiguous United States during 2001–2020. A fast fire gained more than
1,620 hectares on at least one day. This is daily area growth, not the speed of a
flame front in kilometres per hour.

| Measure | Fast-fire share | Basis |
| --- | ---: | --- |
| Events analysed | 2.7% | 1,616 of 60,012 satellite-delineated FIRED events |
| Structures destroyed | 78% | 80,700 destroyed structures in fast fires, using ICS-209-PLUS incident records |
| Suppression expenditure | 61% | US$18.9 billion associated with fast fires in ICS-209-PLUS records |

These are three separate denominators. Do not add the percentages or interpret
them as shares of one common total. The chart preserves the paper's published
rounding. 1,616 / 60,012 × 100 rounds to 2.7%; the other shares are whole percentages.
The chart does not reconstruct unreported exact denominators from rounded shares.

“Structures destroyed” does not mean only homes. It also differs from structures
damaged, structures threatened, or the paper's separate damaged-or-destroyed
analysis. Suppression expenditure is firefighting spending, not total economic
loss, insured loss, or inflation-adjusted current costs. No monetary series is
plotted. Satellite-delineated events should not be presented as a census of every
US ignition. These associations do not measure the benefit of a detection product.

## Reproduction

`data.json` is the frozen input used by the article table and chart generator.
Run `python3 scripts/generate-fast-fires-chart.py` in an environment containing
Matplotlib. It creates the CSV, desktop and mobile SVGs, a standalone PNG and
the 1200 × 750 cover. Every chart uses a 0–100% scale.

The charts are original NuraGaze graphics using the published numerical findings;
no figure or photograph from the journal is reproduced.