A missing detection is not proof that no launch occurred

You search a public evidence collection for a reported launch and find no matching result. What have you learned? So far, only that your search did not return a matching published record in the place and version you checked.

Several other questions remain open. Was a relevant observation available? Was it usable for the question? Was it examined by the method you are evaluating? Was a result published in the collection you searched? Treating those questions as interchangeable can turn an information gap into a claim about an event.

The decision tree below is an original conceptual reading method. It does not estimate a detector’s sensitivity, diagnose an actual missed launch or establish why any particular record is absent.

Start by defining what is missing

Write a sentence specific enough that another person could repeat the check. For example, “I found no matching public record in collection C using search S at retrieval time T.” Those letters are placeholders for your own notes, not LaunchDetect field names.

That statement is narrower than “the satellite saw nothing.” It is also narrower than “the launch did not happen.” To move between those claims, you would need evidence about the observation and processing chain.

Keep the event interval, the collection checked, the search terms or filters and the retrieval time. If a filter or identifier may be wrong, mark the search unresolved instead of treating the empty result as the endpoint.

The missing-result decision tree

Work through these questions in order. At every step, an unknown remains unknown. It is a reason to narrow the conclusion, not to choose the most convenient explanation.

1. Was a relevant observation available?

Look for a recoverable source file or documented observation that covers the area and time needed for the question. General statements about a satellite’s coverage do not establish a particular observation opportunity.

NOAA’s ABI scan-mode description distinguishes full-disk, regional and smaller mesoscale scans. Its page includes historical GOES-16 and GOES-17 schedules; it should not be silently treated as the operating history of a different satellite or date. For a real case, check the actual file and applicable schedule.

• No matching observation found: report the search boundary and the unavailable evidence. The event question remains separate.
• Observation found: continue to usability.
• Availability unclear: stop the negative inference here and record the gap.

2. Was the observation usable and suitable?

A filename alone does not guarantee that the relevant part of the scene contains a usable measurement. NOAA’s GOES-R archive FAQ describes cases where a mesoscale scan falls within the Solar Avoidance Zone and contains fill values rather than valid observations. That is a documented example of why data presence and usable data differ, not an explanation for an unspecified missing launch result.

Check the instrument, product, band, acquisition interval and relevant quality information. Then ask whether those data suit the feature and time scale you are investigating. If cloud conditions or viewing geometry might matter, seek case-specific support in the source data and instrument documentation. Do not pick clouds as a default explanation simply because no signal was reported.

• Unusable data: record the evidenced reason, without relabeling it as a valid negative observation.
• Usable data, suitability unresolved: keep the interpretation open.
• Usable and suitable under a documented method: continue to processing.

3. Is the processing and reporting outcome known?

Ask what was actually checked. A public search result does not automatically show whether source data were acquired, processed, reviewed, rejected or published. Those are possible stages to investigate, not assertions about LaunchDetect’s internal pipeline.

If the method and its execution are documented, record the version, the input and the outcome. If a public page is your only source, describe what it reports without filling in invisible steps. Our timestamp guide helps separate observation, publication and retrieval when timing evidence is available.

• Processing unknown: “No published result found” remains the defensible description.
• Processing incomplete or failed: report that condition if it is evidenced.
• Documented examination with no identified signature: retain that bounded negative result, including its method and limits.

4. What does an independent event account establish?

Check an appropriate completed-event account separately from the image or detection result. A launch schedule describes a plan. A report of a completed launch, a scrub or a postponement describes a different event state and needs to be matched to the correct attempt.

If the event itself is unresolved, say so. If an appropriate source confirms that it occurred, an absent public detection does not reverse that account. If there is a genuine conflict between sources, investigate the conflict rather than letting an empty search decide it.

Three fictional outcomes, three different conclusions

These cases are deliberately synthetic; they are not detector test results.

Case C is where a false-negative study might begin. It is not a false-negative rate by itself. That rate would require a defined reference set, eligibility rules and consistently classified outcomes, rather than one interesting example.

Preserve the gap instead of explaining it away

The useful end product is a short record of which gate stopped the inference: search, availability, usability, suitability, processing or event confirmation. Sometimes all you can establish is a missing published result. That is still a precise and reproducible finding if you keep its boundaries visible.

For the next step, our launch-claim verification guide explains how to separate the event claim from the imagery interpretation. Our satellite-orbit comparison helps frame observation opportunities. Neither step should turn an unknown into evidence that a launch did not occur.

Sources

Primary pages checked October 3, 2026. No real-event attribution, new sensor analysis or numerical detection-performance claim is made here.

• NOAA NCEI: GOES-R Terrestrial Weather, ABI FAQ
• NOAA NESDIS: ABI Scan Mode Information and Channel Summary — includes historical schedules