Producing coordinates is not enough to accept a result, and a smooth curve does not prove that a site remained still. Acceptance of permanent-site GNSS deformation monitoring must first establish what is measured, which data states support a decision, which conditions were evaluated, and what happens when observations or the reference are unavailable.
An actionable acceptance specification therefore needs more than one accuracy number. It needs a reviewable validation framework so that the owner, consultant and technical team use the same meanings for “accepted,” “accepted with limitations,” “insufficient information” and “nonconforming.”
For site suitability, first read Is My Site Suitable for GNSS Monitoring?. The differences among positioning methods and observation durations are covered in Comparing High-Accuracy GNSS Positioning Methods.
Acceptance requires four layers of validation
1. Turn “what are we measuring?” into a measurable acceptance criterion
Acceptance starts with the measurand, not the receiver model. At minimum, define:
- the monitored object and the ground or structural component represented by each point;
- the horizontal, vertical or three-dimensional components to be interpreted;
- the coordinate frame, reference epoch and sign convention;
- whether the question concerns a sudden, staged or long-term change;
- result frequency, allowed latency and the method for calculating availability;
- the decision rule near an acceptance boundary and who has authority to confirm it.
The same coordinates can lead to opposite conclusions when frame, reference epoch or direction is undefined. JCGM 106 similarly treats specified requirements, measurement uncertainty and the decision rule as connected parts of conformity assessment [1].
2. Separate four states—never draw missing data as stability
A common plotting mistake is to connect every timestamp with one continuous line. Acceptance data should distinguish at least:
| Data state | What it can support | What it must not imply |
|---|---|---|
| Usable result | The observations and quality indicators required by the project are present | Absolute correctness, or that every possible displacement has been excluded |
| Limited result | A result exists, but some evidence is weak or affected by the environment | Equal evidential weight to a complete result |
| Solution failure | No acceptable position result was formed for that period | Zero displacement |
| Missing / not yet received | Observations are incomplete or not yet available | Any one conclusion among site stability, equipment failure or actual movement |
Plots that cross missing periods must preserve missing-state markings and clearly distinguish periods supported by observations from periods without measurement evidence. Visual continuity must not imply that the site was stable while evidence was unavailable.
3. Why no single quality indicator is enough
RST’s repeated reviews of de-identified multi-station measurements show that quality indicators answer different questions. No one field independently proves that a coordinate is correct. A green status is one input to the documented quality decision.
| Common indicator or appearance | What it helps with | Blind spot when used alone |
|---|---|---|
| Solution state | Identifies whether a specific solution type was formed | A state can still contain an incorrect ambiguity decision or input sensitivity |
| Ambiguity-quality statistic | Helps identify some marginal or unstable periods | It is diagnostic evidence, not a direct test of coordinate correctness |
| Dispersion within the observation period | Describes convergence or drift within that result | It is not long-term accuracy and should not rank sites without context |
| Amount of valid observation data | Avoids overconfidence from very little data | More data does not remove systematic effects or reference instability |
| Visual smoothness | Makes trends and changes easier to see | Smoothing may hide gaps, steps or common motion |
In practice, several complementary indicators should support an explainable decision while the original states remain available. Site validation establishes the project-specific combination, thresholds and observation periods and links them to the acceptance records.
4. The reference point must also pass acceptance
A relative result contains the behavior of both the monitoring and reference points. Reference motion, accidental antenna movement or an undocumented equipment change can appear as an opposite-signed displacement at the monitoring point.
Acceptance should therefore confirm that:
- the reference lies outside the expected deformation area and has an independent checking method;
- antenna, monument, receiver and configuration changes have dates and version records;
- earthquakes, construction, maintenance and anomalies trigger renewed verification;
- the project defines which results pause or require reprocessing when the reference is suspect.
IGS and NOAA station guidance treats stable monumentation, equipment information, photographs, station logs and data quality as foundations of long-term GNSS results [2][3]. These records are measurement evidence, not administrative attachments.
5. Use independent evidence to show that the system can see change
An acceptance trial should cover both stable conditions and identifiable change, without asking a processing system to be its own only validator. Depending on project risk, evidence may include:
- comparison with control survey, another sensor or a reference quantity supported by a documented calibration or comparison chain;
- repeat-observation consistency under documented trial settings;
- procedural exercises for equipment change, interruption or a suspect reference;
- documented provenance from raw observations through processing version and quality state to the report;
- a predefined “insufficient evidence” outcome instead of changing the rule after seeing the result.
Independent checks do not require every sensor to agree point by point. Methods may observe different physical objects, spatial footprints and time scales. Define the comparable quantity first; a numerical difference alone does not prove that one method is wrong.
If a document claims metrological traceability for an external reference quantity, it must retain a documented, unbroken calibration chain, reference standards and associated uncertainties. Linking a report back to raw data and processing versions is data provenance; the terms are not interchangeable.
What to put in an acceptance document
| Acceptance item | The document should answer | Evidence to retain |
|---|---|---|
| Measurement objective | Which point and direction, relative to which reference, answering what change? | Point map, frame, reference epoch and sign convention |
| Data completeness | How is availability calculated, and how are planned and unplanned outages separated? | Arrival records, gap list and maintenance records |
| Quality decision | Which complementary evidence is used, and how are limited and failed states labeled? | Quality fields, state dictionary and versioned rules |
| Reference stability | How is it checked independently, and how does interpretation resume after an anomaly? | Control survey, redundant reference and before/after checks |
| Change validation | How does the system demonstrate sensitivity to the change of interest? | Trial case, cross-measurement and repeatability result |
| Delivery provenance | Can a plot be traced to observations, processing and human interpretation? | Raw-data index, processing version, report and approval record |
| Exceptions | Who decides when evidence is insufficient, equipment changes or a result is disputed? | Notification, interpretation hold, reprocessing and review procedure |
This table can support procurement requirements, commissioning plans or an acceptance meeting. Actual values still need joint definition from the site, deformation rate, risk and available external checks.
How to apply the evidence framework
The project trial defines accuracy criteria, thresholds, observation periods, weights, quality-model coefficients and the processing states needed for the site and mission. The acceptance document then records those values by version and links them to the trial results and review procedure.
Customer-visible delivery identifies which states the system provides, what evidence supports each conclusion, which conditions trigger an insufficient-evidence state, and how a result can be reviewed. These states, evidence records and review triggers make the framework executable throughout commissioning and operation.
GNSS monitoring series
This article defines acceptance requirements; the adjacent articles cover site suitability and interpretation of individual displacement results.
Previous: GNSS monitoring site assessment | Series 5 of 7 | Next: GNSS displacement data trust
References
- Joint Committee for Guides in Metrology. JCGM 106:2012, The role of measurement uncertainty in conformity assessment. https://www.bipm.org/en/doi/10.59161/jcgm106-2012
- International GNSS Service. Station Operator Resources and Monumentation Recommendations. https://www.igs.org/station-resources/
- National Geodetic Survey. Guidelines for Establishing and Operating CORS. https://www.ngs.noaa.gov/web/data_imagery/CORS/Establish_Operate_CORS.shtml
Frequently asked questions
Q: Is one achieved accuracy number enough for acceptance?
No. The number still needs a measurand, data state, coordinate frame, evidence period and decision rule. Missing observations, a suspect reference or an unknown processing version cannot be repaired by one accuracy figure.
Q: Does a good solution state guarantee a correct result?
No. It is important evidence, but it should be combined with observation amount, internal dispersion, reference stability, continuity and independent checks. Measurements show that a single state can miss marginal or systematic problems.
Q: How should an outage interval be handled in acceptance?
Treat the interval as missing evidence, apply the predefined insufficient-evidence and outage rules, and resume acceptance interpretation only after the required quality checks pass when monitoring resumes.
Q: Can every site use the same acceptance thresholds?
Not directly. Environment, monitored direction, deformation time scale, reference conditions and risk differ. The evidence structure and data-provenance principles can be common; values should come from the site visit, trial and project requirements.
Need to turn a monitoring requirement into a workable trial and acceptance document? Contact RST to structure the measurand, data states, review records and scope of delivery.