Wildfire Black Box · Colorado · Historical evidence study
South Canyon is a case study in how to read a reconstructed fire timeline. This record separates the published physical explanation from numerical transformations and from information that remains unknown.
What happened at South Canyon?
The Forest Service’s September 1998 report documents lightning ignition on 2 July, 14 firefighter deaths on 6 July, 2,115 acres burned and control on 11 July. Its account places a dry cold front at approximately 15:20 Mountain Daylight Time on 6 July.
The proposed transition combined fire in a narrow canyon, strong upcanyon flow and burning through previously unburned green Gambel oak. Underburned oak was not the blowup’s contributor, but did not provide a safety zone. These are findings attributed to the authors, not new Tecnobosque observations. [1]
Fire DNA: lightning origin; terrain–wind interaction; surface-to-shrub-canopy transition; fatal firefighter entrapment. This descriptive signature is not a severity score. Building vulnerability, household evacuation and drinking-water effects are Unknown in this review.
What do the reported spread rates mean?
The report estimates northward drainage spread at about 3 ft/s and eastward upslope runs at 6–9 ft/s. Its perimeter reconstruction used witness information, not continuous instrument tracking of a single front. The authors describe a probable sequence and acknowledge uncertainty. [1]
| Published estimate | Converted value | Meaning |
|---|---|---|
| 3 ft/s | 0.9144 m/s; 54.864 m/min | Northward drainage direction |
| 6 ft/s | 1.8288 m/s; 109.728 m/min | Lower end of upslope estimate |
| 9 ft/s | 2.7432 m/s; 164.592 m/min | Upper end of upslope estimate |
Derived calculation: multiply feet per second by 0.3048 to obtain metres per second, then by 60 for metres per minute. The decimals above audit the conversion; a public summary should round to about 55 m/min and 110–165 m/min. Exact unit conversion does not make an approximate input exact.
A comparison also needs a direction. Dividing 6–9 by 3 gives a factor of 2–3 between these estimates. It does not demonstrate that one continuously tracked front accelerated by that factor. Different parts of a perimeter can move in different directions during the same interval.
Nor can a fire-spread estimate establish a person’s available escape time by itself. A comparison needs locations, start times, terrain, route length, changing conditions and a defensible account of when exposure occurred. This page does not calculate a safe route or a safe distance.
Worked example: how timing uncertainty changes a rate
Hypothetical teaching example—not South Canyon observations. Imagine two mapped positions of the same front, 300 metres apart along a defined direction. The first time is 16:00 ± 1 minute and the second is 16:05 ± 1 minute. Treat the distance as exact only to isolate timing sensitivity.
| Case | Elapsed time | 300 m ÷ time |
|---|---|---|
| Nominal timestamps | 5 minutes | 60 m/min |
| Shortest permitted interval | 16:04 minus 16:01 = 3 minutes | 100 m/min |
| Longest permitted interval | 16:06 minus 15:59 = 7 minutes | 42.86 m/min |
The timing bounds produce approximately 43–100 m/min, rather than a symmetric range around 60. The upper endpoint is about 1.67 times the nominal value. This is interval arithmetic, not a probability distribution. It assumes the two timestamp errors can reach opposite bounds; a shared clock offset would behave differently.
If the first position were also 20 metres too far forward and the second 20 metres too far back, separation could be 260 metres. Opposite errors could produce 340 metres. Combining those hypothetical distance limits with the time limits gives 260/7 = 37.14 to 340/3 = 113.33 m/min. These deliberately simple bounds illustrate sensitivity; they are not uncertainty estimates for the historical fire.
Before averaging rates, preserve the mapped direction and time interval for each segment. A mean of two rates can conceal a short, fast interval; a rate inferred from total distance and total time answers a different question. Retain the source positions so another reader can reproduce the transformation.
Known At: a later reconstruction is not a field briefing
The 1998 paper is a later reconstruction, explicitly additional to earlier investigations. Its cover date is September 1998; the incident date is not its publication date. [1]
For this record, distinguish three timestamps: when something occurred, when a source recorded or released it, and when this page reviewed it. A reconstructed position may describe 6 July 1994 while being supported here by a 1998 publication. That does not establish who knew that position during the incident.
To assess a specific person’s information at a specific time, this record would require a traceable communication, observation or briefing with identifiable recipient and timing. Without that evidence, the answer remains Unknown. A weather forecast’s existence and its receipt by a particular crew are separate claims.
Material claim register
| ID | Claim | State | Audit trail |
|---|---|---|---|
| C001–C003 | Incident facts; proposed transition; directional rate estimates | Reported | Source 1, executive summary and introduction |
| C004 | Metric conversions and 2–3 rate ratio | Derived | Published inputs × stated conversion factors; directions preserved |
| C005 | Synthetic time and distance sensitivity | Illustrative calculation | Invented inputs; formulas shown above |
| C006 | Individual receipt of a particular warning | Unknown | No recipient-specific finding made here |
Reported identifies a source statement, not a claim that Tecnobosque independently verified it at the scene. Derived identifies arithmetic rather than new empirical evidence. No numerical TB-ECI rating is assigned in this version.
Sources, scope and revision history
- Butler and colleagues, Forest Service, Fire Behavior Associated with the 1994 South Canyon Fire on Storm King Mountain, Colorado, RMRS-RP-9, September 1998. Primary full report, particularly executive summary, preface and introduction. Agency catalogue and DOI.
This focused study relies on one primary reconstruction. The catalogue and reproductions are not additional independent corroboration. It does not assess the complete accident investigation, individual responsibility, legal findings or present-day operational doctrine. A second independent investigation has not been audited for this version.
Revision 1.0, 9 October 2026: initial evidence review, directional conversions, synthetic uncertainty example and explicit information gaps. Cite the record ID and version with the source report. This is an educational historical analysis, not incident command advice.