Reciprocal leveling across a channel: keep the height difference in one direction
Reduce paired observations from opposite banks, average the two signed estimates and test what happens when the sight conditions differ.
On this page
ProblemFix the direction on the worksheetExplain the cancellationTest an asymmetric changeOne setup on the west bank says the east mark is 0.157 m higher. The other says it is 0.189 m higher. Averaging gives a sensible elevation transfer only if both numbers represent the same direction of height change. A sign swap on the second setup can make a real rise almost disappear.
Problem
Use a simplified reciprocal-leveling exercise. Fixed marks W and E are on opposite banks. W has an accepted elevation of 8.425 m. Each setup reads upright staffs on both marks. Model the equivalent extra long-sight reading error as equal for the two directions during the observation interval; neglect the short-sight error in this illustrative model. Real river crossings require the observing design, repetitions and controls designated for the work.
| Setup position | Staff on W | Staff on E |
|---|---|---|
| Near W | 1.418 m | 1.261 m |
| Near E | 1.734 m | 1.545 m |
For each setup calculate E minus W. Average those two differences and find the transferred elevation of E. Determine the magnitude and sign of the long-sight error implied by this model. Finally, test how the result changes if the long sight from the east setup alone increases by 0.012 m.
Fix the direction on the worksheet
From either setup, an ideal horizontal line of sight gives elevation E minus elevation W equal to the rod reading on W minus the rod reading on E. This follows because the higher ground mark needs a shorter reading to reach the same line of sight. The physical position of the instrument changes between setups. The desired direction of the elevation difference does not.
Near W, the difference is 1.418 − 1.261 = +0.157 m. Near E, use the same order to obtain 1.734 − 1.545 = +0.189 m. Their average is +0.173 m. Add that to the known W elevation for an E elevation of 8.598 m. If you write the second difference as −0.189 m, label it W minus E and reverse it before averaging.
Explain the cancellation
Suppose each long sight reads too large by an unknown e, while each short sight is treated as unbiased. Near W, the long reading is on E and enters the subtraction with a minus sign. The derived E-minus-W difference will be low by e. Near E, the long reading is on W and enters with a plus sign. This estimate will be high by e.
The estimates differ by 0.032 m, twice the implied common excess. The inferred long-sight reading error is +0.016 m under this model. Check by reducing the near-W long E reading from 1.261 to 1.245 m and the near-E long W reading from 1.734 to 1.718 m. Both corrected differences then equal +0.173 m. This is a model-based diagnostic. The observations alone do not isolate the physical shares due to refraction, instrument collimation or other effects.
Test an asymmetric change
If only the near-E setup’s long W reading rises from 1.734 to 1.746 m, its estimate becomes +0.201 m. The other estimate remains +0.157 m. Their average is +0.179 m, shifting the transfer by +0.006 m. Reversing observation direction suppresses a shared effect that enters with opposite signs. A change unique to one observation contributes half its size to the simple mean.
A complete answer reports the two signed estimates as well as the average. Reporting only 8.598 m hides a 32 mm spread that a reviewer needs to know. An instructor can ask students to sketch the four individual sights and identify which are long. If they can place each reading on the sketch, the signs in the equations become easier to defend. For a real transfer, use the applicable survey procedure to judge whether the conditions, repeatability and additional checks support acceptance.