Tecnobosque · Wildfire data methods
A missing search result is a question to investigate. It is not evidence that a wildfire never happened.
Published 19 October 2026 · Research guide · Numerical example is hypothetical.
What the program covers
The official MTBS FAQ describes US coverage beginning in 1984 and size criteria of at least 1,000 acres in its western zone and 500 acres in its eastern zone. Alaska and Hawaii use the western criteria; Puerto Rico uses the eastern criteria. Both wildfires and prescribed fires can be included. Consult the program’s zone map rather than assigning the threshold from an informal definition of “the West.”
MTBS lists several reasons for missing records: inaccurate occurrence information, missing source reports, unsuitable imagery and fire effects too subtle to map. Released products may also be revised. These limitations make it important to preserve the dataset version and retrieval date.
Area coverage is not event-count coverage
A dataset can capture most of the area affected by fire while omitting many small incidents. These are different measures of completeness. Consider this deliberately simplified example, which is not an assessment of MTBS performance:
| Group | Incidents | Area per incident | Combined area |
|---|---|---|---|
| Large fires | 10 | 10,000 acres | 100,000 acres |
| Small fires | 90 | 10 acres | 900 acres |
| All incidents | 100 | Not applicable | 100,900 acres |
If a hypothetical archive contains only the ten large fires, it covers 10% of the incidents but approximately 99.1% of their combined area: 100,000 ÷ 100,900 × 100. Neither percentage can replace the other.
The research implication follows from the arithmetic: counting records in a size-selected archive does not give the total number of ignitions. It also does not directly measure how often people experienced smoke, how many homes were threatened or how many evacuations occurred. Those questions need their own observations and denominators.
A six-step check for a missing incident
- Confirm the incident identity. Record the name, year, location and an agency identifier where available. Names can be reused, and a complex name may differ from its component-fire names. Keep the source that establishes the connection instead of matching names alone.
- Check the program’s scope. Compare the incident’s documented size and location with the applicable mapping criteria. Do not assume that a disaster’s social impact guarantees eligibility in a land-area dataset.
- Inspect the release coverage. Record the years and product covered by the download or map you actually opened. A website being available today does not mean every event through today has a completed assessment. Consult the release information linked from the official MTBS website.
- Clear restrictive filters. Check dates, geography, incident type and search spelling. Repeat the search by location where possible. Document which searches returned no match.
- Compare with incident reporting. Use the responsible agency’s occurrence record or archived incident information to establish that the event happened. A report of occurrence and a mapped severity product answer different questions.
- Assign an honest outcome. Record “outside the stated size criteria,” “not found in the searched release,” or “identity unresolved,” as supported. Do not replace uncertainty with “no fire” or “no damage.”
Choose the dataset for the question
| Question | Evidence to seek | Common mismatch |
|---|---|---|
| How many incidents occurred? | An occurrence dataset with documented reporting coverage | Counting only fires selected for severity mapping |
| What effects were mapped? | The severity product, assessment dates and class definitions | Treating an absent product as zero severity |
| Where was activity detected at a particular time? | A time-stamped active-fire observation | Treating a post-fire map as a live detection |
| How many structures were lost? | An incident-specific damage assessment | Estimating structure loss from mapped acreage alone |
For satellite observations, use our NASA FIRMS evidence guide. For interpreting a downloaded severity raster, continue with how to read burn severity maps. Before calculating hectares, check pixel area, class selection and resampling.
Write a conclusion another reader can check
A useful absence statement identifies its limits: “We did not find a matching record in the named MTBS release, using these dates, locations and filters, on this retrieval date. Incident occurrence is documented separately by the cited agency report.” Replace every placeholder with your actual evidence; do not present this example as a completed search.
For a comparison between regions or years, define the study population before counting. Explain which incident types and size ranges are eligible, which release is used and what records were excluded. A common size cutoff may address one selection difference, but it cannot by itself resolve inconsistent reporting, release coverage or mapping limitations. Keep a small exclusion log so the final totals can be reproduced.