Differential Leveling

Add the backsight, subtract the foresight, and repeat until you arrive.

Before this Lesson

A brass disk set in a bridge abutment is stamped with its elevation: 512.36 ft. The manhole you care about is several hundred feet away, well down the hill, and you need its rim elevation to the hundredth of a foot. Nothing you own reads elevation. You have a telescope that looks along a perfectly flat line and a very long ruler, and that turns out to be enough.

One flat line

A level is a telescope on a tripod that does one thing: it looks along a horizontal line. Spin it around and that line sweeps out a flat plane at a single height, like the surface of a still pond hanging in the air. A rod is a long ruler with its zero at the foot. Stand the rod on any point, look at it through the level, and the number you read is how far that point lies below the line.

That's the whole instrument. To get elevations out of it, you need one point you already know, usually a benchmark (BM), a permanent mark with a recorded elevation. The reading on a known point is a backsight (BS). It tells you how far the benchmark sits below the line, so adding it gives the height of the line itself, called the height of instrument (HI):

HI=Elevknown+BS\text{HI} = \text{Elev}_{\text{known}} + \text{BS}

Now swing to a point you want. That reading is a foresight (FS), and it drops you from the line down to the point:

Elev=HI−FS\text{Elev} = \text{HI} - \text{FS}
BS = 4.82 ft FS = 7.35 ft HI = 517.18 ft line of sight BM Elev = 512.36 ft A Elev = 509.83 ft
A backsight finds the line, and a foresight drops from it

Notice what this does to your instincts: a big rod reading means a low point. The Handbook has no section on leveling, so these two equations are yours to carry. They don't weigh much.

One setup, many points

Nothing limits you to one foresight. Once the HI is known, you can swing to anything in sight, and each reading drops from the same line. With the line at 517.18 ft, a reading of 2.40 ft on a floor slab puts the slab at 514.78 ft, and a reading of 11.65 ft on a pipe invert (the inside bottom of a pipe) puts the invert at 505.53 ft.

Setting a grade stake runs the same equation backward. Now you know the elevation you want and you need the reading that goes with it:

rod reading=HI−Elevdesign\text{rod reading} = \text{HI} - \text{Elev}_{\text{design}}

If the design calls for 510.00 ft and the line is at 517.18 ft, the foot of the rod belongs 7.18 ft below the line. The person holding the rod slides it up or down against the stake until the level reads 7.18, then marks the stake at the rod's foot.

Checkpoint
1 of 2

From one setup, a level reads a backsight of 5.27 ft on a benchmark at elevation 318.44 ft. It then reads 9.63 ft on a pipe invert and 2.08 ft on a floor slab. What are their elevations?

A.
Invert 321.63 ft, slab 314.08 ft
B.
Invert 333.34 ft, slab 325.79 ft
C.
Invert 314.08 ft, slab 321.63 ft
D.
Invert 303.54 ft, slab 311.09 ft

A point above the line

Every reading so far has been a drop, because every point has been below the line. A ceiling or the underside of a beam sits above it. So the rod goes upside down, or inverted, with its zero pressed against the beam, and you read it the usual way. That reading is still the gap between the line and the point, but now the point is on the other side of the line, so the gap gets added:

Elev=HI+inverted reading\text{Elev} = \text{HI} + \text{inverted reading}
0 0 BS = 4.75 ft inverted reading = 7.20 ft HI = 241.25 ft beam underside Elev = 248.45 ft floor mark Elev = 236.50 ft
An inverted rod measures up from the line of sight

Nothing in the number itself says the rod was upside down, so a problem has to tell you. Read for the word inverted.

Checkpoint

A level reads a backsight of 5.18 ft on a floor mark at elevation 412.75 ft. A rod held inverted against the underside of a beam reads 6.44 ft. What is the elevation of the beam's underside?

ft

Turning points carry the elevation

The disk and the manhole can't both be seen from one setup. Sights get too long to read, and on a hill the rod runs out too: far enough downhill, the line of sight passes clean over the top of it.

So you take the elevation for a walk. Put the rod on any firm spot ahead, such as a spike or a rock, and read it as a foresight. Now you know that spot's elevation, and it's called a turning point (TP). Leave the rod exactly where it is, carry the level past it, set up again, and turn back to read that same rod as a backsight. The spot hasn't moved, so its elevation still holds, and it finds the height of the new line.

A turning point exists only to hand the elevation across. Nobody cares about the spot afterward. A whole level run is those two moves, repeated.

BS 4.82 FS 7.35 BS 3.16 FS 9.04 BS 2.71 FS 8.40 HI = 517.18 HI = 512.99 HI = 506.66 BM 512.36 TP 1 509.83 TP 2 503.95 Rim 498.26
A run down the hill, with each turning point read from both sides

Crews write this in a table called level notes, with one row per point:

Level notes for the run in the figure (ft)
StationBSHIFSElevation
BM4.82517.18512.36
TP 13.16512.997.35509.83
TP 22.71506.669.04503.95
Rim8.40498.26

Watch how a row reads. Its foresight comes off the HI on the row above, because that's the setup it was read from. Its backsight is then added to its own elevation to make the HI on its own row.

Worked Example Up a hill through two turning points
1 of 2

A benchmark is at elevation 638.42 ft. In order, the rod readings are BS 5.16, FS 2.87 on TP 1, BS 6.33, FS 1.94 on TP 2, BS 4.71, and FS 3.08 on point C. Find the elevation of C.

The check that keeps the notes honest

Each setup changes the elevation by BS − FS: up by the backsight, then down by the foresight. A run is a stack of those changes, so they have to add up to the whole trip:

∑BS−∑FS=Elevend−Elevstart\sum \text{BS} - \sum \text{FS} = \text{Elev}_{\text{end}} - \text{Elev}_{\text{start}}

For the run down to the manhole, ΣBS is 10.69 and ΣFS is 24.79, a difference of −14.10 ft, and 498.26−512.36498.26 - 512.36 is −14.10 ft too. A mismatch means an arithmetic slip somewhere in the notes. A match proves only that the arithmetic is right, since a misread rod flows through both sides just the same.

One kind of reading stays out of the sums. A foresight to a side point, like the slab and the invert earlier, is an intermediate foresight, or side shot. No backsight follows it, so it's never part of the chain from start to end, and it has no HI of its own, so the rows after it still read off the last HI above. The check uses every backsight, and only the foresights on turning points and the final point.

Partly filled notes use the same two equations. In a setup, HI = Elev + BS and Elev = HI − FS, so any two of the numbers give the third.

Worked Example Filling the gaps, then checking
1 of 3

Find the missing backsight and the elevation of P, then check the notes. The row marked Inv is an intermediate foresight on a pipe invert.

Partly completed level notes (ft)
StationBSHIFSElevation
BM 56.12810.37804.25
TP 13.47809.794.05806.32
Inv9.66800.13
TP 2?812.932.38807.41
P7.90?

Working back from the far end

Sometimes the known elevation is at the far end: the run started on a point nobody had an elevation for and closed on a benchmark. The readings were still taken in the usual order, so to travel backward, you undo each move. The last thing done was subtracting a foresight, so add it back to recover the HI. The thing before that was adding a backsight, so subtract it to reach the previous point:

HI=Elev+FSElevprevious=HI−BS\text{HI} = \text{Elev} + \text{FS} \qquad \text{Elev}_{\text{previous}} = \text{HI} - \text{BS}

Say a run ends on a benchmark at 300.00 ft, with a last foresight of 6.55 ft and a backsight of 4.20 ft on the turning point before it. The HI was 306.55 ft, and the turning point is at 302.35 ft. To jump straight to the start, rearrange the check:

Elevstart=Elevend−(∑BS−∑FS)\text{Elev}_{\text{start}} = \text{Elev}_{\text{end}} - \left(\sum \text{BS} - \sum \text{FS}\right)
Checkpoint
1 of 2

A run from a manhole rim closes on BM 9, at elevation 476.18 ft. What is the elevation of TP 1?

Rod readings (ft)
StationBSFS
Rim3.62
TP 12.947.15
TP 25.078.20
BM 91.33
ft

That's all of differential leveling: two moves and some bookkeeping. Before tilting the line, make sure the vocabulary is solid.

Checkpoint

Which statements about level notes are true?

A.
A turning point has to be a benchmark with a recorded elevation
B.
A backsight is added to a known elevation to get the HI
C.
The arithmetic check sums the intermediate foresights with the others
D.
A turning point is read as a foresight, then as a backsight from the next setup
E.
When ΣBS − ΣFS matches the net change in elevation, the rod readings were all correct
F.
A foresight is subtracted from the HI to get an elevation

Tilting the line: trigonometric leveling

A level insists on a flat line, which makes a steep hillside a long string of setups. A total station can tilt. It measures a slope distance and an angle along its line of sight, and one setup can cover a rise that would take a level several.

Set the total station over a known point and tape the height of its telescope above that point, hih_i. Put a reflector on a pole of known height hrh_r over the point you want. Then climb from one to the other: up the tripod, along the sight, and down the pole.

ElevP=ElevBM+hi+V−hr\text{Elev}_{\text{P}} = \text{Elev}_{\text{BM}} + h_i + V - h_r
hi V hr H S α BM ElevBM P ElevP
Up the tripod, along the sight, down the pole

It's the level again in disguise. ElevBM+hi\text{Elev}_{\text{BM}} + h_i is the HI, with the tape doing the backsight's job, and the pole is a rod that always reads hrh_r. Set α\alpha to zero, and VV vanishes, leaving Elev = HI − FS.

A sight below horizontal makes α\alpha negative, so VV comes out negative and the equation handles the rest. With a zenith angle, use V=Scos⁡ZV = S\cos Z and H=Ssin⁡ZH = S\sin Z instead, and past 90° the cosine goes negative on its own. Like leveling, none of this is in the Handbook.

Worked Example The elevation under a reflector
1 of 3

A total station stands over a benchmark at elevation 684.20 ft, with its telescope 5.30 ft above the mark. It reads a slope distance of 742.60 ft and a vertical angle of +3°48′ to a reflector on a 6.50 ft pole. Find the elevation of the point under the pole.

The height of a tower

Nobody is climbing a water tower to hang a reflector for you. But if you can get the horizontal distance HH to it, the vertical angles to its top and base are enough. Each sight makes a right triangle with HH as its flat side, and the Handbook's Trigonometry section defines tan⁡θ=y/x\tan\theta = y/x, which here reads V=Htan⁡αV = H\tan\alpha.

Vtop Vbase H αtop αbase
Two sights share one horizontal distance

Both values of VV are measured from the same flat line through the telescope, so the tower's height is the gap between them, signs included:

height=Vtop−Vbase=H(tan⁡αtop−tan⁡αbase)\text{height} = V_{\text{top}} - V_{\text{base}} = H\left(\tan\alpha_{\text{top}} - \tan\alpha_{\text{base}}\right)

Say H=150.0H = 150.0 ft, with the top 32° above horizontal and the base 5° below. Then Vtop=93.73V_{\text{top}} = 93.73 ft and Vbase=−13.12V_{\text{base}} = -13.12 ft, so the height is 93.73−(−13.12)=106.8593.73 - (-13.12) = 106.85 ft. For the elevation of the top rather than the height, start at the telescope: Elevtop=HI+Vtop\text{Elev}_{\text{top}} = \text{HI} + V_{\text{top}}.

Checkpoint
1 of 2

From the bottom of a hill, an instrument is 220.0 ft horizontally from a building on the slope. The vertical angles are 24°10′ above horizontal to its top and 3°05′ above horizontal to its base. How tall is the building?

ft

Why careful crews balance their sights

Back to the level. That flat plane, the still pond in the air, was a simplification, and the arithmetic on the exam keeps it unless a problem says otherwise. Reality has two complaints.

First, the earth curves. A level surface, the kind a real lake follows, falls away from a straight horizontal line. So the farther a sight goes, the higher it runs above the level surface, and the rod reads too high. The air also bends the sight back toward the ground, which takes back only a small part of that. What's left is the gap between a circle and its tangent, which grows with the square of the distance: twice as far, four times the error. It's about 0.002 ft at 300 ft and about 0.6 ft at a mile.

e e 4e line of sight level surface D D 2D
The sight line climbs away from the level surface
The curve is hugely exaggerated.

Second, no instrument is perfect. The line of sight may tilt a hair off horizontal, which is called collimation error. That tilt travels with the telescope, so it tips the same way whichever direction you look, and its error grows with distance too.

Now the trick. Set the level midway between the rods. Both readings are then off by the same amount ee, and the elevation change is

(BS+e)−(FS+e)=BS−FS(\text{BS} + e) - (\text{FS} + e) = \text{BS} - \text{FS}

The error cancels without anyone ever computing it, for curvature, for refraction, and for collimation alike. A little pacing beats a lot of arithmetic.

Checkpoint

A crew sets up a level 100 ft from the backsight rod and 300 ft from the foresight rod. Which statements are true?

A.
The combined curvature and refraction effect on the foresight is about 9 times that on the backsight
B.
Curvature and refraction make both rod readings too low
C.
Moving the level to 200 ft from each rod would cancel the curvature, refraction, and collimation error in this setup
D.
Because the sights are unequal, collimation error doesn't cancel between the two readings

Recap

  • A rod reading is how far a point lies below one flat line: HI = Elev + BS, and Elev = HI − FS. An inverted reading adds to the HI, and a grade stake's reading is HI − design elevation.
  • A turning point is read twice, as a foresight and then a backsight, to carry the elevation ahead. Working back undoes each move: add the foresight, subtract the backsight.
  • ΣBS − ΣFS equals the net change in elevation, leaving out intermediate foresights. It checks the arithmetic, not the rod readings.
  • With a tilted line, ElevP=ElevBM+hi+V−hr\text{Elev}_{\text{P}} = \text{Elev}_{\text{BM}} + h_i + V - h_r, where V=Ssin⁡αV = S\sin\alpha or Scos⁡ZS\cos Z. For a tower, V=Htan⁡αV = H\tan\alpha for each sight, and the height is Vtop−VbaseV_{\text{top}} - V_{\text{base}}, signs included.
  • Curvature and refraction grow with the square of the distance and make a distant rod read too high. Balanced sights cancel them and collimation error.