How to Interpret Deformation Monitoring Data

TL;DR: Interpreting deformation monitoring data means separating genuine structural movement from measurement noise, thermal effects and instrument drift, then comparing the residual movement against pre-agreed trigger levels. The skill is not reading a single number — it is reading the trend, the rate of change and the velocity, and knowing when a 2 mm shift is meaningless and when it demands you stop work and evacuate.

Key takeaways

• Always compare displacement against your measurement uncertainty first. With a Leica TM60 monitoring total station you can resolve sub-millimetre movement, but a single epoch still carries roughly ±0.5–1 mm of real-world noise once atmospherics and prism centring are included — anything below that is signal you cannot trust.
• Set Alert, Action and Alarm trigger levels in writing before monitoring starts, expressed in both absolute displacement (mm) and velocity (mm/day or mm/week). A wall that has moved 8 mm slowly over six months is a different problem from one that moved 8 mm overnight.
• Strip out the thermal and seasonal cycle before you call a trend. Steel and concrete structures breathe daily and annually; a tank or bridge can show ±3–5 mm of reversible thermal movement (per AS 5100 design assumptions) that is not deformation at all.
• Velocity and acceleration matter more than absolute position for slope and excavation monitoring. A constant creep rate is manageable; an accelerating rate (progressive failure) is the single strongest predictor of collapse.
• Tie every result to a stable datum in GDA2020 / MGA2020 horizontal and AHD vertical, and re-check your reference marks every epoch — most "structure moving" alarms are actually a reference prism that has moved.

Why interpretation is harder than measurement

Capturing deformation data is the easy part. A modern automated total station such as a Leica TM60 or Trimble S9 HP, paired with prisms and software like Leica GeoMoS or Trimble 4D Control, will hammer out an epoch every 15 minutes around the clock and email you a graph. The hard part — the part that protects people and assets — is deciding what those graphs actually mean.

The cost of getting it wrong runs both ways. Call a false alarm and you stop a longwall, halt a tunnel boring machine or evacuate a process plant unnecessarily, at tens of thousands of dollars per hour of lost production. Miss a real one and you are explaining to a coroner why the retaining wall, tailings embankment or heritage façade failed while it was being watched. Across Australian mining, civil and structural projects, deformation monitoring exists precisely because the consequences sit at both extremes.

This guide walks through how Industrial Spatial Solutions reads deformation data on real Australian sites — from process plants in the Pilbara to rail underpins in the Hunter Valley and excavations alongside heritage structures in Sydney and Melbourne. It assumes you already have data coming in and need to know what to do with it.

Step 1: Establish whether the movement is real

Before you interpret a single trend, you must know your measurement uncertainty, because nothing below that floor is interpretable.

A geodetic monitoring total station resolves angles to 0.5″ and distances to roughly 0.6 mm + 1 ppm. That sounds like you can see any movement, but the real-world repeatability of a monitored point — including prism centring, atmospheric refraction over the line of sight, and tripod or pillar stability — is closer to ±0.5 mm to ±1.5 mm per epoch depending on range and conditions. Over a 300 m sight line across a hot iron-ore stockpile yard at Port Hedland in February, refraction alone can swamp a millimetre of structural movement.

So the first interpretive question is never "how much has it moved?" It is "has it moved by more than my noise band?" Practical rules:

• Plot the data with error bars or a confidence band, not as a bare line. Movement is only real once the trend clears two to three times the single-epoch standard deviation.
• Use a moving average (a 6- or 12-epoch mean) to see the trend through the scatter. A single spike epoch is almost always atmospheric or an obstruction, not a structure failing.
• Check reference marks first. If three independent reference prisms all "move" together by the same amount, your instrument or datum shifted — the structure did not.

Tip: For high-precision work over short ranges, a Leica Nova MS60 or a static FARO Focus Premium scan can drop sub-millimetre detail, but scanning gives you whole-of-surface change detection rather than discrete-point precision. Match the tool to whether you need point precision or area coverage.

Step 2: Remove the thermal and environmental signal

The most common misread in deformation monitoring is mistaking reversible thermal movement for permanent deformation.

Every steel and concrete structure expands and contracts. A 50 m steel conveyor gantry can move several millimetres end to end between a 5°C Mount Isa night and a 45°C afternoon. Bridges are designed to it — AS 5100.2 sets the effective temperature range structures must accommodate. Tanks, silos, kiln supports and crane rails all show the same daily and seasonal breathing.

The signature is unmistakable once you look for it: the displacement trace tracks temperature, peaking each afternoon and recovering each night, with a larger annual sine wave underneath. If your "settlement" graph goes up in summer and back down in winter, it is thermal, not structural.

To interpret correctly:

• Overlay temperature on the displacement plot. Most monitoring software (GeoMoS, Trimble 4D Control) ingests a temperature sensor feed for exactly this.
• Look for the residual after the cyclic component is removed. The deformation you care about is the slow drift the structure does not recover from between cycles.
• Account for groundwater and rainfall on slopes and excavations. A batter in the Bowen Basin can accelerate after rain and stabilise after — that correlation is itself useful interpretation, not noise to discard.

Step 3: Read the trend, the rate and the acceleration

A single displacement number is nearly useless. Interpretation lives in the time series. There are three quantities to read, in order of increasing importance for safety-critical work:

1. Displacement (mm) — where the point is now relative to its baseline. Compare against absolute trigger levels.
2. Velocity (mm/day or mm/week) — how fast it is moving. This is what distinguishes slow consolidation from active failure.
3. Acceleration — whether the velocity is increasing. A point creeping at a steady 0.2 mm/day is stable behaviour; the same point accelerating from 0.2 to 1.0 to 4.0 mm/day over successive days is the classic progressive-failure curve that precedes slope and embankment collapse.

For a building settling under a new adjacent excavation in Sydney, you expect an initial movement that decays toward zero — the rate slows as the ground consolidates. That is healthy. The danger sign is the opposite shape: movement that holds a constant rate, or worse, an upward-curving rate. On tailings dams and pit walls, an inverse-velocity trend (where 1/velocity heads toward zero) is the established method for forecasting time-to-failure and is exactly the kind of reading monitoring data exists to surface.

Tip: Always plot velocity alongside displacement. Two points can sit at the same 10 mm of total movement, but the one decelerating is safe and the one accelerating may need the area cleared today.

Step 4: Compare against trigger levels (Alert, Action, Alarm)

Interpretation only has meaning against thresholds agreed before monitoring began. The standard three-tier scheme used on Australian projects:

These must be set by the responsible engineer for the specific structure — there is no universal millimetre figure. A heritage masonry wall in Melbourne may have an Alarm at 5 mm of differential movement because masonry cracks easily; a flexible earth embankment may tolerate 50 mm before it matters. Express each level in both displacement and velocity, because a structure can breach an Action level slowly (manageable) or rocket through it overnight (not).

When data crosses a trigger, the first interpretive act is verification, not panic: confirm the reference network is stable, check for an obstruction or weather event, and look at neighbouring points. Real structural movement is almost always spatially coherent — adjacent points move in a sensible pattern. A single lone point jumping while its neighbours sit still is far more likely to be a knocked prism than a failing structure.

Step 5: Read the spatial pattern, not just one point

Deformation rarely happens at a single point in isolation. Interpreting the whole network — or, with laser scanning, the whole surface — tells you the mechanism.

• Uniform settlement across a footprint usually means consolidation under load — often acceptable and predictable.
• Differential settlement (one corner dropping faster than another) is what cracks buildings and tilts equipment; it is far more serious than uniform movement of the same magnitude.
• Tilt and rotation on a tank, silo or wall point to foundation or bearing-capacity problems.
• Bulging or convergence in a tunnel or shaft, captured cleanly by a FARO or Leica RTC360 scan comparison between epochs, indicates ground load redistributing onto the lining.

Scan-to-scan comparison (a coloured deviation map between two epochs) is the most intuitive way to read spatial deformation, because the eye spots a pattern instantly where a table of coordinates hides it. For discrete-point networks, plot displacement vectors on a site plan so direction and magnitude are visible together.

Common mistakes to avoid

Mistake 1: Trusting a moving reference

The most frequent false alarm is a reference prism or pillar that has itself moved — knocked by plant, thermally affected, or founded on ground that is also deforming. If the whole structure appears to shift uniformly overnight, suspect the datum before the asset. Always carry redundant reference marks and check them every epoch.

Mistake 2: Reacting to single epochs

One bad epoch is not a trend. Atmospherics, a momentary obstruction (a truck parked in the sight line), or a target with condensation on it will all throw a single reading. Wait for the moving average to confirm before acting — unless the magnitude is so large that safety overrides patience.

Mistake 3: Ignoring measurement uncertainty

Reporting "3 mm of movement" when your single-epoch noise is ±2 mm is not interpretation, it is wishful precision. Every result should be stated against its uncertainty.

⚠️ Watch out: The most dangerous error is dismissing accelerating movement as "thermal" or "noise" because previous alarms were false. Alarm fatigue kills. When velocity is genuinely increasing across multiple coherent points and clearing the noise band, treat it as real until proven otherwise — verification takes minutes, a collapse does not wait.

Cost considerations

The interpretation effort — and therefore cost — scales with how critical and how fast-moving the structure is. Indicative Australian ranges:

A well-designed monitoring and interpretation programme typically costs a fraction of one hour of unplanned downtime on a major asset — the value is entirely in catching the problem early enough to act on it.

Frequently asked questions

How much movement is too much?

There is no universal figure — it depends on the structure and is set by the responsible engineer before monitoring starts. Heritage masonry may alarm at a few millimetres of differential movement; a flexible earth structure may tolerate tens of millimetres. What matters more than the absolute number is the rate: steady, decelerating movement is usually safe, while accelerating movement is the warning sign regardless of magnitude.

How do I tell real deformation from measurement noise?

Compare every result against your single-epoch measurement uncertainty (typically ±0.5–1.5 mm for a geodetic total station, sub-millimetre for short-range precision work). Use a moving average to see the trend through the scatter, plot a confidence band, and require movement to clear two to three times the noise before calling it real. Check reference marks first — coordinated "movement" across all references means the datum shifted, not the structure.

Why does my data go up and down every day?

That is almost always thermal cycling. Steel and concrete expand in the heat and contract overnight, and the daily peak in your displacement trace will line up with the daily temperature peak. Overlay temperature on the plot to confirm. The deformation that matters is the residual drift left after this reversible cyclic movement is removed.

What is velocity and why does it matter more than position?

Velocity is the rate of movement (mm/day or mm/week). It distinguishes safe behaviour from dangerous behaviour better than total displacement does: a point can sit at 20 mm of movement and be completely stable if it is no longer moving, while a point at 5 mm that is accelerating may be days from failure. For slopes, embankments and pit walls, an accelerating (or inverse-velocity) trend is the primary failure indicator.

What datum should deformation monitoring be tied to?

Horizontal results should be referenced to GDA2020 / MGA2020 and vertical results to AHD, tied through stable reference marks well outside the zone of influence. The reference network must be checked every epoch — a monitoring result is only as trustworthy as the datum it sits on.

Can laser scanning replace total station monitoring?

They answer different questions. A monitoring total station with prisms (Leica TM60, Trimble S9) gives high-precision, automated, continuous movement at discrete points. Terrestrial laser scanning (FARO Focus, Leica RTC360) gives whole-surface change detection between epochs — ideal for spotting bulging, convergence or unexpected movement you did not have a prism on. Many programmes use both: prisms for continuous critical points, scans for periodic whole-of-structure checks.

Talk to ISS about your monitoring data

Reading deformation data correctly is the difference between a costly false alarm and a missed warning — and both are avoidable with the right interpretation framework, trigger levels and datum control behind your numbers. Industrial Spatial Solutions designs, runs and interprets deformation monitoring programmes across mining, civil and structural projects Australia-wide, from automated total-station networks to scan-based change detection, with engineer-reviewed reporting you can act on. If you have data you are not sure how to read, or a structure you need watched properly, call ISS on 0407 057 015 for a quote and a straight answer.