Tecnobosque · Wildfire intelligence
Black Box
Permanent wildfire records that preserve both what happened and how the documented understanding changed—with evidence, uncertainty and revision history kept visible.
The public question
What happened?
Incident facts, fire progression, weather, terrain, exposure, response, impacts and the current technical assessment.
The Black Box difference
How did we come to know?
Known At checkpoints, changing official values, investigation stages, superseded information, evidence strength, remaining gaps and versioned corrections.
The first Reference Grade record
Eaton Fire 2025
The Eaton record tests a demanding evidence problem: extreme nighttime spread, lost aerial intelligence, changing direction, evacuation timing, two reviews with different scopes and a cause record that evolved from Under Investigation to Undetermined and finally Electrical.
It is the visual reference implementation for the Wildfire Black Box system and the model for future record maturity.
Explore the complete record →Six modules make a record auditable
Mechanism and sequence
Reconstructs origin context, documented movement, operational constraints and alternative explanations without turning possibility into fact.
No hindsight contamination
Freezes what a defined observer could document at a particular time. Later findings cannot be projected backward.
Facts can mature
Tracks revised area, impacts, cause and interpretation. Superseded information remains historically visible.
Comparable incident profile
A fixed descriptive framework supports disciplined cross-fire comparison. Unsupported dimensions remain unknown or null.
Trace every material claim
Observed, reported, derived and interpreted statements remain distinguishable and connected to retrievable public sources.
Permanent identity
Every mature record publishes its record ID, version, update state and method so research can cite a stable object.
1,454 wildfire records. Two ways to explore the evidence.
The wider fire archive covers 1,454 public records across seven countries: 308 case narratives and 1,146 preliminary cartographic records. Within that collection, the Black Box comparison currently covers five full pilot records and 52 material claims. Additional case studies extend the reading archive; they do not automatically receive a TB-ECI score or Reference Grade status.
Search the full archive
For a reproducible geometry example, see the QGIS wildfire perimeter comparison guide.
Filter by country, search incident names and topics, or sort the archive by year.
Browse the wildfire archive →Nine reading paths
Explore evacuation timing, home exposure, changing cause findings and post-fire recovery through related cases.
Choose a reading path →Five scored pilot records
Read the evidence matrix alongside its methodology. Evidence confidence describes documentation, not fire severity.
Open the pilot comparison →Seven starting points in the archive
A selection of incident records to begin reading. Their scope and evidence maturity differ; consult each page’s sources and limitations before comparing findings.
Eaton Fire
Nighttime progression, alert chronology, review scope, changing cause states, Forensic Reconstruction, Known At, Evolution of Knowledge and Fire DNA.
Open Eaton →Palisades Fire
Conflicting official structure-loss totals and changing agency cause information.
Open record →Lahaina Fire
Minute-level chronology, two ignition phases and systems-level contributors.
Open record →Marshall Fire
Two documented origins, extreme wind, rapid spread and the boundary between ignition evidence, response and building performance.
Open record →Camp Fire
Ignition findings, rapid spread, evacuation chronology and the limits of retrospective interpretation.
Open record →Fort McMurray / Horse River
Mass evacuation, differing population scopes and dated satellite footprints.
Open record →Black Saturday
A multi-fire disaster in which statewide impacts and fire-specific causes must remain separate.
Open record →Evidence states, not flattened certainty
A material statement cannot enter the public record as an unqualified fact. Its status changes when the evidence changes.
TB-ECI measures evidence strength and auditability—not incident severity and not the probability that a claim is true. Cause is separated from criminal responsibility and civil liability. Repetition of one upstream statement is not treated as independent corroboration.
One canonical record, two controlled layers
Public evidence layer
Wildfire Black Box
Permanent incident pages, readable claims, public sources, evidence states, limitations, visible revisions, selected comparisons and citations.
Protected intelligence layer
Tecnobosque Core
The deeper structured knowledge asset is not published. The public record exposes enough to audit its conclusions without disclosing protected processing, internal relationships or proprietary analytical implementation.
A record built to change
New evidence updates the record. It does not erase its history.
Wildfire Black Box keeps a permanent incident identity while investigations develop, agencies revise totals and better evidence becomes available. Unknown remains a valid result; uncertainty is part of the record, not a defect to hide.