Zero baseline in structural monitoring: how to choose it?

In brief

• The baseline must match a specific engineering decision, not simply "the first record".
• A candidate may be accepted, rejected or deferred until conditions stabilize.
• A project reference, a device reference and a profile version reference have different consequences.
• Changing the zero baseline requires a date, a reason, an audit trail and preservation of the previous context.
• The zero baseline does not replace calibration, coefficients or assessment of the measurement chain health.

What a zero baseline is, and what it is not

Zero baseline, also called a reference measurement or baseline, is the recorded state against which later changes are calculated.

If the current value is 12.8 mm and the reference value is 10.1 mm, the reported change may be 2.7 mm. This is a simple mathematical operation, but choosing the number 10.1 mm is a decision that affects the whole history. If the reference state was recorded during grout settlement, improper installation or structural movement, every later result will inherit that interpretation error.

A zero baseline is not calibration. Calibration establishes the relationship between indication and a reference value with known metrological quality, together with defined conditions and uncertainty. A reference selects the point from which you calculate change in a specific project. Setting a new zero will not fix wrong sensitivity, a wrong unit, a shifted clock or a damaged sensor.

This boundary is key during system acceptance. A supplier may set the baseline correctly, while the chain still needs calibration verification. The opposite can also happen: a sensor may be correctly calibrated, but the chosen reference moment may be useless.

Why "the first valid record" is often not enough

The first record has one technical advantage: it exists. It does not automatically have engineering value. After installation, the following may still be taking place:

• curing or stabilization of the material around the sensor,
• settlement of the fixing element,
• thermal equalization between the sensor and the structure,
• bleeding of a hydraulic system,
• stress relaxation after drilling or installation,
• changes in construction loads,
• correction of orientation or element length,
• clock synchronization and communication repair.

The State Geological Institute describes reference measurements for inclinometer casings as being taken after the initial conditions have stabilized. This is not a universal term for every device and structure, but it shows the principle well: baseline timing follows the method and installation conditions, not software convenience.

In the GEOKON interpretation guide, collecting several independent early series and assessing their repeatability before selecting a baseline for a manual inclinometer probe is recommended. For an automatic string, the frequency and procedure are different, but the logic remains the same: one sample does not show stability.

First define what state zero should represent

There is no single "natural zero" for a structure. The team should answer which moment it wants to use for decisions:

• the state after installation and stabilization,
• the state before excavation or test loading begins,
• the moment the structure is handed over for use,
• the start of a specific work stage,
• a new, explicitly separated phase after a configuration change.

These purposes are not interchangeable. The designer may need displacement from the pre-excavation state, while the contractor may need the increment from the start of the current stage. Rather than overwriting one reference for both questions, it is better to keep the main reference and build an additional reporting comparison for the stage.

The most risky case is an "operational reset" whose only purpose is to zero a growing value. Such a reset does not remove movement. It removes visibility of its accumulated part.

Criteria: accept, reject or defer

Accept when all conditions are met

A candidate is suitable for a zero baseline when:

1. the device and configuration are clearly identified,
2. time is reliable and recorded in UTC,
3. required components have a valid quality status,
4. there is no unexplained gap or duplicate,
5. installation and stabilization have been completed according to the project method,
6. the work stage matches the reference purpose,
7. early series are sufficiently repeatable,
8. temperature and load are known within the range needed for interpretation,
9. the responsible person has approved the selection,
10. there is a procedure for preserving the previous baseline when it changes.

Reject when you know the specific reason

Reject a candidate if, during measurement, there was a device error, a missing required axis, a wrong unit, an impossible time, a change in orientation, incomplete installation or an event that makes the state incomparable with later measurements.

Rejection does not mean deleting the record. Raw data may be needed for failure audit or reconstruction of the sequence. It only means the record should not drive change calculations.

Defer when there is not enough data

A lack of grounds for acceptance is not always proof of an error. If the sensor is still reaching thermal equilibrium, grout is curing or two series differ without explanation, the decision must be deferred. Set a condition: for example, three matching cycles, completion of the work stage or a site inspection.

The "defer" option protects the team from pressure to launch the dashboard at the expense of reliability. The system may already be collecting data, but the differential result remains marked as unapproved.

The seven-step procedure for selecting a baseline

1. Define the quantity and the decision purpose

Record whether the baseline concerns displacement, strain, pressure, temperature or a profile. State which decision will rely on the change: safety alarm, trend observation, stage settlement, or assessment of the effectiveness of a measure.

2. Lock the input configuration

Make sure identifiers, units, segment lengths, orientations, coefficients and equations match the physical state. A reference selected for a wrong configuration will not become correct after a later change in the description.

3. Collect a candidate window

Do not limit the assessment to a single sample. Gather a series before and after the planned moment, show temperature, statuses, events and adjacent sensors. The window length depends on the dynamics of the asset and the measurement method.

4. Assess repeatability and context

Compare the candidates, but do not apply an arbitrary "closest to the mean" rule. A stable but wrongly installed sensor still produces repeatable data. Technical evidence and site knowledge are both needed.

5. Choose the reference scope

Decide whether the change concerns all project devices at one point in time, one device, or a new profile version. This choice affects the comparability of charts and alarms.

6. Approve and describe

The minimum note should include the reason, stage, conditions, data source, author, approval date and known limitations. For a critical asset, a four-eyes rule is worth requiring.

7. Observe the post-activation period

After setting the baseline, review the first differential results. An unusual jump on all channels may indicate that different times, units or configuration versions were chosen. Activating the reference does not end quality control.

Project, device and profile references are three different decisions

Project reference

A common moment for multiple devices allows the response of the whole asset to be compared against the same stage. This makes sense when all required devices have reliable data for the selected moment. Missing data should not be filled with the "nearest record" without an explicit rule, because individual channels would start their history under different conditions.

Device reference

An individual reference is needed after later installation of a new device or when the method requires a separate stabilization moment. It should not silently change the meaning of project aggregates. The dashboard must let the recipient identify different starting points.

Profile version reference

In a chain inclinometer, the result depends not only on the zero reading, but also on the order, length and orientation of the segments and on the anchor point. After a physical change to the chain, a new calculation context is created. The earlier configuration and its references must remain tied to the earlier period; the details are described in the article on versioning a chain inclinometer configuration.

Illustrative example: baseline measurement before excavation

Illustrative example. Six automatic devices are operating on the retaining wall of an excavation. Installation was completed on Monday, but the grout was still stabilizing and wiring work continued until Wednesday. The first series are complete, but two devices show a downward trend and a difference between day and night.

The team does not select Monday's "first record". On Thursday it confirms orientation and lengths, compares the series with geodetic control and waits for two full thermal cycles. On Friday at 06:00 UTC all devices have a valid status, earthworks have not started yet and the results are repeatable. This moment is approved as the common project reference.

Three weeks later, one device is replaced. The new one receives its own reference after stabilization and a clear note; the team does not reset the other five. The project report shows different starting points, so the accumulated movement from the start of excavation does not disappear.

The example does not set a minimum waiting time. The required period depends on the device, installation, asset and monitoring plan.

How the zero baseline affects alarms and responsibility

A threshold based on change relative to the baseline inherits every reference decision. If you shift the baseline by 4 mm, the alarm value may drop by the same 4 mm even though the physical state has not changed. Therefore, changing the reference during active monitoring should trigger at least:

• an assessment of the impact on current alarm states,
• preservation of the previous values and history,
• approval by a person with the proper role,
• notification to report recipients,
• a decision on whether the thresholds remain valid,
• control of the first results after the change.

A new baseline should not be used to "close" an alarm. An alarm is resolved by assessing the cause and making an operational decision. A new reference point may be justified after a permanent state change or reconstruction, but in that case the report should show continuity with the previous period.

USACE in the manual Instrumentation of Embankment Dams and Levees emphasizes the need to control initial observations, because later results are referred to them. The same responsibility logic applies: an error at the start does not remain local.

What this looks like in Inclify

In Inclify, an administrator can select a reference from an existing measurement instead of entering the value manually. For a project, a common date and time can be assigned to devices, while for a single device the reference can be set separately. An optional note makes it possible to record the work stage, conditions and reason for the decision.

The change leaves an audit trail. This helps answer not only "what is the reference?" but also "who changed it, when and why?". The platform does not treat this operation as calibration and does not confirm the metrological correctness of the selected record. Responsibility for acceptance criteria remains with the project team.

For automatic chain inclinometers, the reference is tied to the versioned profile configuration. Changing the active configuration creates a new version instead of editing the historical meaning of the data. The result value can be analyzed as segment increment and accumulated profile; the difference is explained in the guide incremental vs accumulated displacement.

The platform records the decision, but it does not choose the correct structural state for the project team. Stabilization criteria, acceptable statuses and required approvals should be defined in the monitoring plan before alarms are activated.

Acceptance checklist for a zero baseline

• [ ] The purpose of the reference is recorded in one sentence.
• [ ] The exact time and UTC zone are specified.
• [ ] The work and load state has been described.
• [ ] Installation and the required stabilization period are complete.
• [ ] All required components have valid status.
• [ ] Units, orientations and coefficients are approved.
• [ ] More than one series has been collected to assess repeatability.
• [ ] The data were compared with temperature and site events.
• [ ] Deviations between adjacent sensors were explained.
• [ ] The scope was defined: project, device or profile version.
• [ ] An authorized person approved the selection.
• [ ] The note includes the reason and known limitations.
• [ ] The impact on thresholds and current alarms has been determined.
• [ ] The previous reference remains reproducible.
• [ ] The first results after activation were reviewed.

Limitations and cases where a separate procedure is needed

There is no universal time from installation to baseline. Recommendations for a manual probe, an automatic chain, a piezometer and a vibrating wire sensor are not interchangeable. The monitoring design, the manufacturer's instructions, the installation method and the decision of the responsible engineer take priority.

The baseline does not remove uncertainty. Two stable records may both be affected by a unit or orientation error. The reference also does not confirm that the inclinometer toe is in stable ground. If the point treated as the anchor moves, the accumulated profile will receive a systematic shift.

ISO 18674-3 addresses measurement of displacements along a line using inclinometers and places the method in the context of geotechnical applications. ISO 18674-1 defines the general rules for field monitoring. The standards do not replace a project-specific specification or the device manual.

If the system is to calculate the result from the raw value, the reference is only one element of the full chain. The formula, coefficients, unit and version must be described separately in the calculation sheet. When choosing between an automatic chain and a manual probe, start by comparing inclinometer methods.

Sources and further reading

1. ISO 18674-1:2015 - Geotechnical monitoring by field instrumentation: General rules, ISO.
2. ISO 18674-3:2017 - Measurement of displacements across a line: Inclinometers, ISO.
3. Model 6140 Vertical MEMS In-Place Inclinometer String - Instruction Manual, GEOKON.
4. Manual Inclinometer Probe - Data Interpretation Guide, GEOKON.
5. Long-Term Bridge Performance Program - Baseline Data, Federal Highway Administration.
6. Instrumentation of Embankment Dams and Levees, EM 1110-2-1911, U.S. Army Corps of Engineers.
7. Inclinometer measurements - deep monitoring of landslides, State Geological Institute - PIB.

What next: approve the baseline measurement with full context

If you are launching a new asset or cleaning up an existing history, prepare several candidates, the work log and the current configuration. We will help turn the criteria into an auditable workflow in the platform. Book a call about references and data acceptance.