Before asking how far the wall moved, ask what stayed still.
A total station reports that a wall moved 6 millimetres. What if it was the reference prism that moved? Learn how a change in the reference can enter the result, how different sight lines help locate the change, and how to report a finding without confusing a relative measurement with an absolute one.
We will start with one horizontal line, then move to a real survey problem: a station that has to be set up again, reference marks with different sight lines, and one apparent movement that refuses to agree across setups. You do not need to operate a total station to work through the examples. Draw the points, choose a sign, and keep the reference visible in every sentence.
The numerical examples are fictional. Real monitoring schemes need a qualified survey team, site-specific uncertainties and an agreed response plan. In particular, uncertainty in the survey is never permission to ignore a safety alert.
Geodesy subject guide · Learning that transfers into life · How X Works hub
Choose your reading route
Route 1 · Chapters 1–3: Make the measurement explicit
Start with a difference, not a coordinate
Start with one signed difference, then see what direction and orientation do to it.
Route 2 · Chapters 4–6: Find the changed part
Give the network a way to disagree with itself
Use independent control, process a disputed epoch, and compare uncertainty with the decision.
Route 3 · Chapters 7–9: Decide what to report
Separate three changes that can look alike
Separate instrument effects, scope the alert, and transfer the check to another site.
Full chapter index · Independent worked case · Questions and practical checks
Chapter index
Make the measurement explicit
Start with a difference, not a coordinate
What the total station actually observes
Work a small orientation error before looking at a whole network
CHAPTER 1 OF 9 · Make the measurement explicit
1. Start with a difference, not a coordinate
Put a reference point R and a wall target W on a straight east-west line. Choose east as positive. At the first visit R is at 20.000 metres and W is at 35.000 metres in an agreed coordinate system. The signed difference W minus R is +15.000 metres. At the second visit the difference is +14.994 metres. What did the observation actually say? W ended up 6 millimetres less far east of R than it was before.
The simplest explanation may be that W moved 6 millimetres west and R stayed still. But if R moved 6 millimetres east and W stayed still, the same difference results. So does a 3 millimetre westward motion of W combined with a 3 millimetre eastward motion of R. Repeating that difference many times can make its estimate better. It cannot, on its own, tell us which combination happened.
Write it as a bookkeeping rule: change in the difference equals change in W minus change in R. This is true even when both move east. If W moves east by 2 millimetres and R east by 8, the observed difference still shrinks by 6. A minus sign in a relative result does not by itself establish westward wall motion. It establishes westward motion relative to the reference in this one-dimensional case.
The useful habit is to name the datum before interpreting the number. A datum is the reference used to express positions. If a sheet says “W: −6 mm,” rewrite it as “W moved −6 mm relative to the assumed position of R in the east component.” Once the assumption is visible, another person can ask whether R really was checked or merely held fixed by the software.
CHAPTER 2 OF 9 · Make the measurement explicit
2. What the total station actually observes
A total station measures angles and distance along a line of sight to a target. The prism reflects the instrument’s signal. The machine does not directly measure a global easting, a wall’s stability or the identity of the bracket behind the prism. Coordinates are derived using the instrument setup, orientation, observations and a model that includes relevant corrections. This is why a stable-looking coordinate can still be tied to a changing reference.
A range change is chiefly sensitive to the part of a small movement along the sight line. Imagine standing to the west of a target while it moves north. In the simplest sketch, that move is almost across your sight line. The range may scarcely change even though the horizontal direction does. Stand south of the same target and the range becomes much more responsive. A second view adds evidence only if its geometry helps distinguish the possibilities you care about.
An angle observation can locate a target across a sight line, but it also needs a sound orientation. If a single distant reference establishes the direction and that reference moves sideways, the new orientation can rotate all targets together in the computed solution. The farther the object point from the station, the larger the transverse coordinate error from the same small angle. Seeing several targets shift in a pattern does not immediately identify the cause. A common error can produce a coherent pattern.
A useful desk check is to draw an arrow at every target for the proposed movement. Put your pencil along each sight line and ask which part of each arrow lies along it and which part crosses it. If two hypotheses create almost the same pattern at every observed target, a more precise instrument in the same positions may not resolve the question. You need a different view, a trustworthy reference or an independent instrument that observes a helpful component.
CHAPTER 3 OF 9 · Make the measurement explicit
3. Work a small orientation error before looking at a whole network
Imagine the backsight prism is 40 metres east of the station and shifts north by 4 millimetres between epochs. In this deliberately simplified horizontal sketch the station itself is stable, and a single backsight sets the orientation. The true backsight direction rotates by roughly 0.004 divided by 40, or 0.0001 radians. If the software still assigns the backsight its old direction, the computed directions inherit a rotation of approximately the opposite sign.
Now consider a wall target due east of the station, 60 metres away, that did not move. An orientation error of 0.0001 radians produces a transverse coordinate error of about 60 times 0.0001 metres, or 6 millimetres. A 20-metre target along the same line would have about 2 millimetres of transverse error. This is an approximate small-angle calculation. It illustrates a pattern to investigate, not a substitute for a network adjustment.
A changed prism position perpendicular to the sight line can therefore affect computed wall coordinates even where the observed backsight distance hardly changes. Checking only the backsight distance misses the most informative observation in this example. Conversely, a change predominantly along the line might have little effect on its direction but show clearly in the distance. Record what you checked; do not write “reference good” if only one component was tested.
Can we reverse the example and conclude that any displacement increasing with distance is an orientation error? No. Real rotational motion of a structure can also produce a distance-dependent pattern. The clue is what else shares the apparent rotation. If independently stable control points show it as well, an orientation explanation becomes stronger. If the stable points are clean and only the structure shows the pattern, investigate structural motion with suitable geometry and uncertainty.
CHAPTER 4 OF 9 · Find the changed part
4. Give the network a way to disagree with itself
Reference points are selected for stability, but stability is an evidence claim with a time interval. A target on an apparently solid wall may share a foundation with the structure being monitored. Several prisms along the same retaining wall may all follow one ground movement. Three targets are not three independent checks when the failure that concerns you could move all three together.
A simple design question is: if the suspected reference moves, what observation will tell us so without using that same reference? Two surviving references can reveal a change in their separation, but they may not identify which one moved. With a better-connected network of references on plausibly independent supports, you can examine multiple lines and use an outside tie where the project requires one. The additional observation should distinguish hypotheses, not merely increase the number of readings.
Suppose A, B and C form the local reference group. If A–B and A–C both change in a pattern consistent with A moving while B–C stays consistent, A is a sensible suspect. But B and C might have moved together relative to A. If their monuments share the same ground, that alternative is not far-fetched. A tie to stable control outside the suspected movement area is the kind of evidence that can help. If no such tie exists, report the limitation as part of the result.
In practice, compare the observations in a solution that does not automatically hold every supposed reference rigid. A constrained adjustment can redistribute the disagreement across coordinates and residuals. Use an appropriate reference-stability analysis and keep enough redundancy to test it. Trying solutions with and without a suspect point is useful sensitivity evidence, but removing points until a warning disappears is not an independent test. Document which observations support the exclusion and which movements the surviving geometry can detect.
CHAPTER 5 OF 9 · Find the changed part
5. Worked case: a wall target appears to move twice
Use a simple one-dimensional model for the east component. The following numbers are changes in signed coordinate differences between visits, already obtained from appropriately oriented observations in this fictional case. W minus A changes by −6 mm, W minus B by +1 mm, and W minus C by 0 mm. In the same reduction, A minus B changes by +7 mm, A minus C by +6 mm, and B minus C by −1 mm. Write each pair in that order, because reversing the pair reverses the sign.
Start with the identity W minus B equals W minus A plus A minus B. The changes give −6 + 7 = +1 mm, matching the reported W minus B. W minus C similarly gives −6 + 6 = 0 mm. This is an arithmetic consistency check. If those values had all been computed from the same coordinates, agreement would be automatic and would add no independent evidence. In a real audit ask which ranges, directions and setups supply independent checks rather than counting derived differences as new observations.
The pattern is consistent with A moving east relative to B and C, while W has little detectable east motion relative to B and C. The pattern is also consistent with A staying put while B, C and W move together west by roughly the same amount. To decide which account is plausible, inspect the point supports and use additional reference evidence. In this fictional case an independently tied check finds A east of its earlier position while B and C remain consistent within uncertainty.
A physically independent visit also finds that A’s mounting bracket was disturbed. That does not by itself explain the whole numerical change: the bracket finding and the directional evidence should agree. With these checks, we can say the original −6 mm from the A-based reduction does not establish wall displacement. We can give the revised result relative to verified B and C and state the components and sensitivity available. We cannot say no part of the wall moved in any direction.
CHAPTER 6 OF 9 · Find the changed part
6. Compare uncertainty before replacing an alert with a conclusion
An apparent change should be interpreted alongside the uncertainty of that change. Suppose the W relative to B component has a standard uncertainty of 1.4 mm in one epoch and 1.8 mm in another, and the relevant errors are independent across visits. For their difference the standard uncertainty is the square root of 1.4 squared plus 1.8 squared, about 2.3 mm. An observed +1 mm is small against that uncertainty. It does not establish a 1 mm wall movement.
If the errors share a common reference, calibration or observation model, the independent calculation may not apply. The variance of a difference is the sum of the two variances minus twice their covariance. Positive covariance for the same component and definition can make the difference more precise than the independent calculation. A new unmodelled error, such as a moved bracket, is not cured by a small formal variance. Keep covariance and possible unmodelled changes separate in the record.
An alert threshold and a measurement uncertainty answer different questions. A threshold belongs to a monitoring and response plan. Uncertainty describes what the measurement can resolve under stated conditions. A report can say the first solution triggered review, that a reference fault was found, and that the revised result does not confirm the apparent displacement. It must not silently replace the alert or assert safety from the revised coordinate alone.
If the project needs to detect a particular magnitude of movement, check whether the actual arrangement can do it in the required direction. A setup might resolve east motion well but be weak for up-down movement at the same target. Report the component and the comparison interval. If performance was not established at the threshold, say so and use the agreed escalation process. Giving more decimal places is not a way to improve detection.
CHAPTER 7 OF 9 · Decide what to report
7. Separate three changes that can look alike
First, the object or ground may move. Look for a pattern across object targets and times that is physically plausible and is seen against verified control. Independently tied observations can help, and the project may have other instruments or inspections to compare. Different instruments can legitimately disagree if they measure different components or cover different spans of time. Put both results on the same question before treating one as wrong.
Second, the reference or station setup may change. Compare stable-control observations, station centring and orientation, and the definition of the coordinate frame. A station set up at a new point is not itself a failure if it is tied correctly into the same monitoring frame. A station placed on a mark whose coordinates were assumed unchanged can spread the mark’s error into every derived object position. Preserve the setup record when making the comparison.
Third, the target assembly or observation can change. A replacement prism or mount can shift the point being measured even if the wall behind it is unchanged. An incorrect prism constant can affect range. A new sight line near a heated surface can change atmospheric effects. These are different hypotheses with different signatures. A single jump associated with a target replacement is a reason to investigate and document the discontinuity, not a reason to erase the earlier series.
A practical worksheet gives every hypothesis three entries: what observation it predicts, which evidence is independent of the suspected fault, and what result would make the explanation less plausible. If all three hypotheses are tested only by repeating the same station setup, the worksheet exposes that weakness. Add a check that would actually change your mind.
CHAPTER 8 OF 9 · Decide what to report
8. Write a report that preserves the original evidence
A useful note includes time, coordinate frame, point identifiers, targets and mounting details, relevant atmospheric observations, processing choices, control checks and uncertainties. Keep raw observations and the original adjustment alongside any revised result. Make it possible to tell what was measured, what was assumed fixed and what changed in processing. A later reviewer should be able to reproduce the decision without having to trust the conclusion first.
For the worked example, a short report might say: “The A-based comparison indicated −6 mm east-component change at W. Independent control checks showed A had changed relative to the reference network. A reduction tied to checked B and C did not resolve an east-component change at W across these visits.” Add the supported uncertainty, note any component not resolved, and identify what has been sent for engineering review.
Compare that with “The wall did not move; false alarm.” The shorter sentence hides the frame, treats uncertainty as proof of zero and closes a safety question a survey alone cannot close. It also loses the reason to inspect A before the next epoch. The accurate sentence is not evasive. It makes three useful decisions possible: what to trust now, what to fix and what is still open.
Where possible, add one planned check to the next visit. For example, reobserve the independent tie, confirm the replacement bracket’s measured point and verify the solution in the weak component. Record what would count as agreement before you see the measurements. If a repeated value is close to the original but arises from the same untested assumption, it is not the extra assurance the plan needed.
CHAPTER 9 OF 9 · Decide what to report
9. Transfer: the reference group moves together
A footbridge monitoring team has two reference prisms on separate brackets attached to one retaining wall. Over three visits, distances and angles between the two references agree closely. The bridge appears to shift north relative to them. A student says the two matching references prove that the bridge moved. What is the missing distinction? The two brackets agree with each other; they need not agree with a frame outside their shared support.
Choose an outside check that can see the disputed north component. An independently verified monument and a differently oriented setup could be useful, subject to design and access. In a simplified one-dimensional sketch, imagine the outside check shows both references moving south by 4 mm while the bridge remains consistent with its earlier position. The local result of +4 mm bridge-minus-reference is then explained by reference movement, without changing any local difference.
Now change one fact. The outside check sees height very well but has poor north sensitivity. Does it settle the dispute? No. Independence in ownership or instrumentation is not enough. The check must actually measure the disputed component with sufficient sensitivity. Identify another line of sight or a suitable positioning method rather than drawing a conclusion from irrelevant precision.
Finally, suppose the bridge and references are all stable relative to each other, but an outside network detects the entire area moving together. A local survey can still be doing its job if its task was to detect movement of the bridge relative to its local surroundings. It is inadequate if the task was to detect motion relative to regional control. The question determines the frame; the frame determines the scope of the answer.
Independent worked case and evidence checks
A second fictional case uses an inspection gantry with two monitoring targets. A new prism was fitted at M between visits; the target N was untouched. The original setup reports a 5 mm range increase at M only. A side-on setup sees no detectable cross-sight change. Do the checks prove the structure moved along the first sight line? No. Both a change in the measured point along that direction and a range offset in that setup can fit those observations.
Ask for the target and instrument records first. Was the replacement prism type configured? Was the bracket left in place? Can the measurement be repeated from a second direction that has meaningful sensitivity to the proposed motion? If the mounting point can be measured independently, use that check to connect the prism centre to the physical point the project wants to monitor. If those checks cannot discriminate the explanations, record the result as unresolved.
Try the sign check with two numbers: a reference shifts east 5 mm and its object target shifts east 2 mm in a common independent frame. What does the object-minus-reference comparison report? It reports −3 mm, not +2 mm. The right calculation takes less than a minute. The useful achievement is noticing which of those answers the question asked for.
For a peer review, ask another reader to circle the independent evidence in the case. Do not count a second plot of the same data as a separate observation. If the independent evidence cannot see the relevant direction, ask them to propose a check that can.
Suppose instead that two checks both see a comparable eastward jump at M and at a witness mark on the bracket, while a mark in the surrounding masonry remains consistent. That localises the change more narrowly: the bracket moved relative to the masonry, at least in the measured component. If the monitoring requirement refers to the masonry, the prism result cannot be used unchanged. If the requirement refers to the bracket itself, the same observation may be the result the team needed. Record the physical point before deciding whether a displacement is real for this purpose.
Worked comparison
| Pattern | Investigate | What you still need |
|---|---|---|
| Several targets share a rotation-like change | Orientation, backsight stability and possible object rotation | Independent control and a second useful geometry |
| One target changes after replacement | Prism configuration, mount and point definition | A check of the same physical point |
| Local references agree with each other | Common support or common-mode motion | Outside evidence if the decision needs a wider frame |
| A small revised change remains | Uncertainty, covariance and the required sensitivity | A component-specific result and the agreed response |
Questions and practical checks
If a prism is called a control point, is it fixed?
Fixed may mean constrained in an adjustment. Its physical stability must still be established for the comparison.
Can a single reference ever be useful?
Yes, for a relative measurement with an explicitly scoped question. It cannot independently distinguish its own movement from object movement.
Does passing closure prove stability?
No. Closure can support consistency of observations under the model. A coherent reference shift or a common error can still evade the check.
Can I use a second sensor as independent confirmation?
Yes if its reference and failure modes are sufficiently independent and it measures the relevant direction and time interval.
Does a quiet plot mean the site is safe?
No. A survey reports measurements within its sensitivity and frame. Site safety decisions require the applicable professional assessment and response plan.
How do I label the revised result?
Give the target, direction, reference, interval and uncertainty, and explain why the original and revised solutions differ.
Continue with the connected subjects
Reference frames and the mathematics of change
Free networks, datum defects and what relative precision can establish
Use error ellipses to read uncertainty by direction
Separate local strain from rigid rotation and network effects
Nearby observation problems
When a common signal is real motion rather than noise
Reciprocal levelling: target heights, closure and hidden common errors
Geomatics: control points and total-station observations
Sources and scope
USACE Survey Engineering and Mapping: manuals and survey control resources
The numerical cases in this guide are original fictional teaching examples, not measurements from a real site. This is an educational explanation, not a field safety instruction, engineering approval or an official certification. For a real project, use qualified practitioners, current product documentation and site-specific evidence. Sources were checked on 5 November 2026.