Wildfire Black Box · Oregon · Historical evidence record
Almeda’s record continues beyond the fire perimeter. To understand the event, keep an issued evacuation notice separate from a resident’s receipt of it, an early loss estimate separate from a later inventory, and a housing project separate from a household’s return.
What happened in the Almeda Fire?
Jackson County’s after-action report dates the Almeda Drive and South Obenchain fires to 8 September 2020. It reviews coordination, communications, evacuation information and sheltering across the two incidents. That combined scope matters: a county-wide finding should not automatically become an Almeda-only statistic. [1]
The report is a review of response, not a complete property-by-property fire reconstruction. It can support questions about how agencies shared information without establishing the first ignition mechanism of every destroyed home. This version makes no original-cause determination or present-day criminal-case claim.
For the housing record, two dated bulletins from Representative Pam Marsh preserve what was being communicated during the early recovery period. A later Oregon Housing and Community Services (OHCS) page describes losses and supported housing development. These sources have different purposes and should retain their own dates and definitions. [2] [3] [4]
What do the evacuation-alert records actually prove?
The county report describes 39 Citizen Alert messages between 8 and 24 September, including 15 during the first 48 hours. These counts span the response to both fires. Its description of the historical system distinguishes stored landline information from opt-in mobile, email and text contacts. [1]
Tecnobosque interpretation: the number of messages is not the number of people warned. Multiple messages can cover different areas, revise instructions or reach the same recipient. An issued message also does not establish delivery, reading, comprehension or departure.
A defensible notification analysis needs separate records for the message, its geographic target, the intended recipients and delivery results. A movement study then needs departure observations with timestamps. Those evidence objects can be linked, but one cannot substitute for another.
The after-action report identifies effective mapping and multiple communication channels, while also describing limited trained access to the alert system and coordination gaps. These are findings about the response organization, not an audited household receipt rate. [1]
To reconstruct what one resident could have known, first establish which version of a warning applied to their location. Preserve the map boundary valid at that time and the message’s issue time. A later map does not necessarily represent the earlier notification area.
Then distinguish availability from receipt. A public notice may have existed before a resident encountered it. Conversely, someone may have received a direct warning that is absent from the particular archive being examined. A missing record in one dataset is not proof that no warning occurred.
How many homes were lost? Read the count with its category
The bulletins preserve two early statements. On 14 September 2020, Marsh described more than 1,500 homes in destroyed manufactured-home and RV parks, with apartments and other houses additional; a complete loss inventory was still pending. On 18 September, the bulletin reported an initial assessment exceeding 2,350 homes across housing types. [2] [3]
| Evidence object | Reported quantity | Boundary |
|---|---|---|
| 14 September bulletin | More than 1,500 | Park-related homes; other housing additional |
| 18 September bulletin | More than 2,350 | Initial assessment across housing types |
| OHCS county snapshot, reviewed 12 October 2026 | 2,508 | Jackson County; Almeda and South Obenchain combined |
OHCS separately describes Almeda as destroying nearly 2,500 homes. Its exact Jackson County figure covers both fires. The approximate Almeda statement and the combined county inventory are different claims. [4]
These entries should not be subtracted to calculate how many additional homes burned after 14 September. Their categories, assessment completeness and geographic scopes differ. A changed assessment can reflect improved accounting; it does not, by itself, document new destruction.
Nor is “more than” an exact number. Preserve that qualifier instead of turning a lower-bound statement into a precise observation. A comparison should first establish whether the same housing units were eligible for inclusion in both inventories.
Why supported rebuilding is not a completed-recovery percentage
The OHCS page displays 1,096 owned homes and 396 rental homes rebuilt or being rebuilt in Jackson County. It explains that its development counts include new housing opportunities and homes supported through OHCS, rather than all privately funded recovery. The figures are a webpage snapshot, not a dated audit of occupancy. [4]
| Quantity | Calculation | Interpretation |
|---|---|---|
| Supported housing pipeline | 1,096 + 396 = 1,492 | Published owned and rental categories combined |
| Pipeline / reported county loss | 1,492 ÷ 2,508 × 100 ≈ 59.5% | Cross-metric ratio; includes differing stages and new units |
The arithmetic is reproducible, but its meaning is narrower than a “recovery rate.” A completed replacement, a home under construction and a newly added unit are not equivalent outcomes. The numerator does not identify how many original lost units were replaced, or how many displaced households returned.
Subtracting 1,492 from 2,508 would likewise not establish the number of households still unhoused. The categories do not form a complete, mutually exclusive household ledger. Recovery outside OHCS is not represented by the numerator, and a supported development can add housing beyond the original stock.
Use the ratio only to describe the relationship between the displayed quantities. A completion rate requires a numerator containing completed outcomes and a denominator representing the population eligible for those outcomes, under a consistent definition.
How to track recovery without losing the people or the units
Suggested analytical method, not a dataset created in this review: maintain one inventory for damaged housing units, another for recovery projects and another for household outcomes. Connect them through documented identifiers and dates instead of assuming that every new unit corresponds to one lost home.
- Housing unit: incident attribution, pre-fire use and documented damage classification.
- Project: replacement or additional supply, funding status, construction stage and completion date.
- Household: temporary accommodation, permanent rehousing and observation date.
- Link: evidence that connects a project outcome to a specific loss or rehousing outcome.
Count each unit once at a stated cutoff. Receiving funding from two programs does not create two recovered homes. A project name is also not a unit count: one development can contain many dwellings, while a rebuilding effort may span several phases.
Specify what success means before calculating it. Restoring the original housing stock, providing additional affordable supply and rehousing displaced residents can all be useful objectives. They require different measures and may progress at different speeds.
For future versions of this record, a reconciled unit inventory would improve loss accounting. A matched project-and-occupancy dataset would improve recovery analysis. Until those are audited, retain the published program figures with their limits.
Known At and evolution of the record
| Checkpoint | Evidence object | Boundary |
|---|---|---|
| 8 September 2020 | Two-fire emergency | Later county response review |
| 14 September 2020 | Legislative bulletin | Incomplete housing accounting |
| 18 September 2020 | Legislative bulletin | Broader initial loss assessment |
| 2021 report version | County after-action review | Retrospective coordination and warning analysis |
| 12 October 2026 | OHCS webpage examined | Access date; the figures’ update date is not established here |
This ordering preserves what the documents say at different stages. It does not turn the later report into information available to residents during the fire, nor turn an access date into the completion date of a housing project.
Fire DNA: what this case can support
- Incident scope: two simultaneous fires share a county response review.
- Warning evidence: issued messages remain distinct from recipient outcomes.
- Loss accounting: qualifiers, housing categories and geography stay attached to counts.
- Recovery: funding, construction, additional supply and rehousing remain distinct.
- Remaining gaps: no reconciled property ledger or matched household-recovery dataset was audited.
No severity ranking or TB-ECI score is assigned. No universal warning success rate or mitigation benefit is estimated. The record describes a historical disaster and subsequent evidence, rather than providing current emergency instructions.
Primary sources, method and citation
- Jackson County Fire Incidents After Action Report. County-hosted 2021 version; introduction, public-information section 2.3 and Appendix B. Listed through the county’s public emergency-plans page. The review covers both Almeda and South Obenchain.
- Representative Pam Marsh: Almeda Fire Response Update for Jackson County. Oregon State Legislature bulletin, 14 September 2020, 1:38 p.m. PDT. Contemporary communication, not an audited final inventory.
- Representative Pam Marsh: Almeda Fire Day 10. Oregon State Legislature bulletin, 18 September 2020, noon PDT. Initial loss assessment across housing types.
- OHCS: Progress on 2020 Labor Day Disaster rehousing and recovery. Jackson County section and explanation of counting scope; webpage reviewed 12 October 2026.
Method: source comparison and separately identified arithmetic. Four primary documents are used; two are bulletins by the same public official and are not independent loss surveys. No raw alert-log audit, household receipt study, unit matching or field investigation was performed. Proposed measurement methods are Tecnobosque’s synthesis.
Cite: Tecnobosque, Almeda Fire 2020, record TB-WBB-US-2020-ALMEDA, version 1.0, evidence review 12 October 2026.