Carlton Complex Fire 2014: Wind, Fuel Treatments and Post-Fire Flooding

WILDFIRE BLACK BOX · Washington, United States · 2014 · TB-WBB-US-2014-CARLTON · Reviewed 9 October 2026

Carlton Complex is a case of changing conditions, not one uniform fire environment. Wind-driven growth, different treatment histories and later flooding produced different evidence questions. This record follows the 2014 incident and the subsequent landscape research, keeping reported impacts, measured severity and editorial interpretation separate.

What was the Carlton Complex Fire?

Washington DNR’s 2014 Annual Report identifies the July Carlton Complex as lightning-sparked, reporting 256,108 acres burned and 322 residences destroyed. These are the totals attributed to that report, not a new damage inventory. [1]

The Forest Service’s briefing dated 9 September 2014 describes four component fires—Stokes, Gold Hikes, French Creek and Cougar Flat—merging into a larger fire. It reports major runs toward Brewster and Pateros on 17–18 July and approximately 300 homes burned. [2]

Tecnobosque source check: the figures differ numerically by 927 acres. That subtraction does not establish 927 acres of later fire growth: the documents have different labels, and this synthesis does not reconcile their boundaries or counting methods. Retain both observations rather than silently replacing one with the other.

A short timeline: ignition conditions, wind and a second emergency

• July 2014: DNR identifies lightning as the ignition source for the complex. [1]
• 14 July: NWS Spokane’s summer review describes thunderstorms initiating new Cascade wildfires. This is regional weather context, not an individual timestamp for each component ignition. [4]
• 17 July: the same review links a dry cold front and strong winds to extreme growth of the Carlton Complex. [4]
• 17–18 July: the Forest Service briefing records major runs toward Brewster and Pateros. [2]
• 21 August: NWS reports a flood over the burn area destroying ten homes that had survived the fire. [4]
• 9 September: the dated BAER briefing documents post-fire assessment and requested protective actions. [2]

These milestones do not reconstruct a continuous fire front or an evacuation sequence. The ignition cause, the weather associated with rapid growth and the later storm response belong to separate parts of the record.

What did fuel-treatment research find?

Prichard and colleagues’ 2020 study describes the burned landscape as 56% grassland and 37% forestland, with the remainder sparse or unvegetated and developed. Its treatment analysis focused on forested areas, using landscape models that considered treatment history, topography, vegetation and weather by fire-progression interval. [3]

Earlier progressions had more moderate and high severity in treatment areas. Thinning with underburning, underburning alone and previous wildfires performed better than thinning alone or thinning with pile burning. Later progressions under milder weather contained more unburned or low-severity area, with smaller differences among treatments. During wind-driven spread, treatments on leeward slopes tended to have lower severity than those on windward slopes. [3]

Tecnobosque interpretation: a useful comparison identifies which treatment, which part of the landscape and which progression interval it describes. Pooling every treated hectare into one category loses those distinctions. The contrast between early and later intervals also warns against treating exposure conditions as identical.

The measured forest outcome cannot simply be relabelled “homes protected.” A separate building-outcome analysis would need its own locations, construction characteristics, exposure information and comparison design. Nor does a lower severity class establish that fire stopped at a treatment boundary.

For the comparison method, use our fuel-treatment evidence guide and the guide to burn severity maps.

What changed after the flames?

The NWS account of 21 August provides an observed consequence: homes survived the wildfire but were subsequently destroyed by flooding. Those ten flood losses should retain their process and date; this record does not add them to the DNR fire-loss figure as though both sources used one combined inventory. [4]

The September BAER briefing covered the Lower Methow River and Swamp Creek / Columbia River watersheds. It requested measures including road-drainage improvements, storm-warning systems, patrols, signs and gates. These were requested responses in an assessment, not proof here that each action was installed or effective. [2]

Tecnobosque interpretation: surviving the fire did not settle the later watershed question. A fire perimeter, a stream network, an assessment boundary and an exposed property locate different objects. The follow-up task is to identify the relevant drainage and dated hazard product, not infer safety from a point’s position relative to the burned footprint.

Burned area is not a uniform vegetation outcome

DNR describes the Carlton fire burning virtually all of Davis Canyon Natural Area Preserve, removing much of its shrub cover while leaving an intact native grassland. That is a site-specific account, not a recovery measurement for the entire complex. [5]

The distinction matters when summarising a large incident. “Acres burned,” “forest severity,” “shrubs removed” and “native grassland retained” need their original definitions and spatial boundaries. A single acreage cannot substitute for all of them.

Evolution of knowledge: the event date is not the finding date

Known At discipline: these document dates do not establish exactly when every observation first became available to responders. The 2020 conclusions belong to retrospective research; they should not be inserted into a July 2014 operational narrative as knowledge crews already possessed. The annual-report year also is not, by itself, a precise publication date.

What can be transferred to another fire?

Transfer the questions before transferring a conclusion: what outcome was measured, what land was compared, what conditions applied and what period was observed? A treatment label without those fields is too coarse for evaluating performance.

This record does not calculate a new causal treatment effect, resolve every difference between agency totals or reconstruct individual home ignitions. For those tasks, the underlying spatial data and additional incident evidence would need a separate analysis.

FireDNA: lightning ignition · wind-driven progression · grassland–forest mosaic · treatment placement · post-fire flooding · evolving evidence.

Agency records and primary research

1. Washington Department of Natural Resources: 2014 Annual Report, wildfire-protection section. Reported area, residences and lightning cause.
2. Forest Service: BAER Analysis Briefing, SW Carlton Complex, 9 September 2014. Incident overview and post-fire assessment.
3. Prichard, Povak, Kennedy and Peterson (2020): Fuel treatment effectiveness in the context of landform, vegetation, and large, wind-driven wildfires. Ecological Applications 30(5), e02104. DOI: 10.1002/eap.2104.
4. National Weather Service Spokane: The Weather Watcher, 2014 newsletter, page 3, “Summer 2014 in Review.”
5. Washington DNR: Davis Canyon Natural Area Preserve, Carlton Complex vegetation account.

Method: synthesis of agency records and published landscape research. Reported findings and Tecnobosque interpretation are distinguished. The 927-acre difference is arithmetic between two reported figures, not a mapped growth estimate. No new fieldwork, imagery processing, property-level hazard model or forensic ignition reconstruction was performed. Published 9 October 2026.