Find your next learning step
When delayed reference evidence reopens one stratum, repair should follow the discovered mechanism rather than default to a whole-product rebuild. Reclosure then needs independent evidence that the named gap is controlled. If the gap affected released maps, a historical amendment should identify versions, places, dates and decisions touched while preserving the original product and what was known when it was issued.
Open the chapter contents · Remote Sensing root · Gold Standard of Education · How X Works
Chapter contents
Choose the route closest to the decision, or read from Chapter 1. The chapters remain open; these grouped routes are a navigable index.
Route 1 · Define the decision
Route 2 · Work the ledgers
Route 3 · Transfer the method
Route 4 · Check limits and evidence
Route 5 · Ask and continue
CHAPTER 1 OF 12 · DEFINE THE DECISION
1. Define the reopened stratum and its claim
Describe the stratum using observable attributes: class, season, sensor, geography, terrain, cloud condition or decision context. Attach the exact released claim that lost support. A vague label such as difficult pixels cannot guide repair or communication.
Keep unaffected strata separate. Reopening aquaculture boundaries after tide-mismatched references does not automatically invalidate stable inland forest classes. Narrow scope protects both scientific honesty and useful service.
CHAPTER 2 OF 12 · DEFINE THE DECISION
2. Diagnose before choosing the repair
Trace the error through reference timing, geolocation, class definition, sampling, preprocessing, model inference and post-processing. Replicate the failure on preserved inputs. Then design the smallest repair that addresses the mechanism.
If the problem is tide-phase mismatch, retraining on more mismatched labels can make the model look stable while deepening the error. Repair the reference protocol and temporal matching first.
CHAPTER 3 OF 12 · WORK THE LEDGERS
3. Build a reopened-stratum repair register
A coastal land-cover map reopens three strata after delayed field evidence.
| Stratum | Failure mechanism | Repair | Independent check | Interim claim |
|---|---|---|---|---|
| Aquaculture edge | Boundary displacement | Object-aware boundary model | Reserved farms | Narrowed |
| Tidal mudflat | Tide-phase mismatch | Time-matched reference protocol | Independent tide dates | Suspended |
| Cloud-obscured coast | Selection gap | Targeted probability sample | New clear-view season | Provisional |
| Stable mangrove | No material failure | None | Routine monitoring | Unchanged |
Every repair receives a version and owner. Reference-protocol and sampling repairs are treated as first-class changes, not hidden beneath a new classifier label.
CHAPTER 4 OF 12 · WORK THE LEDGERS
4. Work a reclosure evidence gate
The programme predeclares accuracy, boundary and coverage gates on independent evidence.
| Reclosure check | Threshold | Aquaculture | Mudflat | Cloud coast |
|---|---|---|---|---|
| Stratum accuracy | At least 85% | 89% | 87% | 82% |
| Boundary median error | At most 20 m | 16 m | 18 m | 24 m |
| Reference time match | At least 90% | 94% | 96% | 91% |
| Coverage debt | At most 6 points | 5 | 4 | 9 |
| Decision loss | Below local tolerance | Pass | Pass | Fail |
Reclosure is stratum-specific. The programme restores the two supported claims and keeps the cloud-coast limitation public with a new sample plan.
CHAPTER 5 OF 12 · WORK THE LEDGERS
5. Use independent evidence for reclosure
Evidence used to design or tune the repair belongs to development. Reserve new locations, times, interpreters or sensors for confirmation where feasible. If complete independence is impossible, label the overlap and reduce the claim.
Report uncertainty and sample support alongside a threshold. A nominal pass based on a handful of easy cases is not a defensible closure. Inspect spatial concentration and consequential subgroups even when the aggregate passes.
CHAPTER 6 OF 12 · TRANSFER THE METHOD
6. Amend historical products without rewriting history
Create a versioned amendment that identifies affected product IDs, acquisition dates, geographic scope, strata, error mechanism, corrected layers, metric changes and recommended user action. Keep the original accessible or at least preserve its manifest and checksum.
Distinguish correction from reinterpretation. A geometric correction may justify replacement pixels; a later class-definition change may require a new collection and a crosswalk rather than silently overwriting the old meaning.
CHAPTER 7 OF 12 · TRANSFER THE METHOD
7. Changed-context transfer: rapid flood mapping
Transfer the method to flood maps issued during an emergency. Vegetated floodplains reopen after high-water marks reveal missed water under canopy. The repair combines a canopy-aware feature and targeted field references.
Reclosure tests separate events and independent floodplain reaches. The amendment identifies maps used during response, explains that open-water areas remain supported, and provides corrected floodplain layers without claiming responders had the later evidence at issue time.
CHAPTER 8 OF 12 · TRANSFER THE METHOD
8. Connect reopening, delayed labels and monitoring
Begin with verification-window reopening and rolling debt reassessment. Use delayed-label backfill and window closure to recover the original evidence boundary.
Return to seasonal support renewal and coverage-debt budgets when the repair needs another season before its claim can reclose.
CHAPTER 9 OF 12 · CHECK LIMITS AND EVIDENCE
9. Limits and common misconceptions
Retraining is not always repair. Errors can originate in reference timing, geometry, definitions or sampling. A higher whole-map accuracy does not prove the reopened stratum improved.
An amendment is not an admission that every original decision was unreasonable. It is a dated statement that later evidence changed support. Reclosure also does not guarantee permanence; new sensors, seasons or land-cover processes can reopen the claim again.
CHAPTER 10 OF 12 · CHECK LIMITS AND EVIDENCE
10. Practical evidence and publication checks
Archive the reopening trigger, stratum definition, affected claims, preserved inputs, root-cause analysis, repair code, development data, independent confirmation, thresholds, uncertainty, coverage debt and reclosure decision.
For amendments, record product and collection versions, checksums, spatial footprints, dates, changed metrics, corrected assets, unchanged strata, user action, citation guidance, notification channels and the owner of future support.
CHAPTER 11 OF 12 · ASK AND CONTINUE
11. Frequently asked questions
Should the whole product be reprocessed?
Only when the mechanism or production system affects the whole product. Otherwise a targeted amendment may be clearer and less disruptive.
Can a stratum reclose with a limitation?
Yes, if the remaining limitation is explicit and the supported claim is narrower than the failed one.
What happens to citations of the original map?
Keep the original identifier and add a linked amendment or successor-version notice so readers can recover both the historical record and current recommendation.
CHAPTER 12 OF 12 · ASK AND CONTINUE
12. Sources and next reading
USGS describes Landsat Collection 2 as a major archive reprocessing effort that applied advances in processing and algorithms. Its Collection 1 legacy page explains that preserved sample products support quality assessment, processing-chain testing and cross-collection comparison, while the Collection 2 known-issues page documents discovered artifacts. The CEOS November 2025 land-cover and change-map protocol addresses reference design, temporal correspondence and transparent accuracy reporting. The coastal register and numbers are original teaching examples.
Continue through the Remote Sensing root, the Gold Standard of Education and How X Works.
Top · Contents · Previous · How X Works