Auto Level Survey: How Levelling Is Done on Site

What is an auto level survey?

An auto level survey measures height differences between points by reading a graduated staff through an automatic level. Its built-in compensator keeps the line of sight horizontal once the instrument is roughly levelled. From a point of known height, every other level follows by addition and subtraction.

The older dumpy level needed its long bubble re-centred before every reading. In an automatic level, a pendulum-mounted prism keeps the sight line level once the circular bubble sits near centre, so the surveyor sets up, focuses and reads. That speed makes automatic level surveying the everyday method for floor levels, drains and road profiles on Indian sites.

Every run starts from a known height: a Survey of India benchmark nationally, or a temporary project mark on a building site. Our Survey Benchmark (BM) glossary entry explains the difference and why the starting mark is checked first.

How is a levelling run done on site?

Set the level between two points and read the staff on the known point (backsight) and on the new point (foresight). The level then moves forward on change points until the run closes on a known mark.

1. Recover the benchmark and confirm its value from the project register.
2. Set up roughly midway between backsight and foresight on firm ground, with the bubble centred.
3. Backsight on the benchmark and add it to the benchmark level to get the height of instrument, then take intermediate sights on floor corners or drain inverts.
4. Foresight on a firm change point, move forward and backsight the same point again.
5. Close the run on a known benchmark and compare the result with its value.

Reading the staff correctly

Most site staffs carry 10 mm blocks in an "E" pattern. The surveyor reads metres, decimetres and centimetres and estimates millimetres, so 1.525 means 1 metre, 5 decimetres, 2 centimetres and about 5 millimetres. Under a ceiling or soffit, the staff is held upside down and the reading booked as negative. The booker repeats each reading aloud so the observer catches a slip.

How are readings reduced? A worked field book

Readings are reduced by the height-of-instrument method or by rise and fall, and both carry an arithmetic check. The table is a real-format field book for two set-ups.

Check: ΣBS − ΣFS = 3.390 − 2.385 = 1.005; Σrise − Σfall = 1.480 − 0.475 = 1.005; last RL − first RL = 101.005 − 100.000 = 1.005. On a township, the same routine is repeated over a grid of hundreds of points; see our contour survey for township team for how that scales.

With the height-of-instrument method, each new HI is the change point RL plus the new backsight, and each RL is HI minus the staff reading. Rise and fall compares each reading with the one before it, where a smaller reading means the ground rose. NPTEL's surveying course works both methods with the same checks¹. The arithmetic check proves the sums only, so the run must also close on a known mark.

How accurate is levelling, and how is error kept down?

A well-adjusted auto level closes to about ±2.5 mm per kilometre. Most errors come from collimation, unequal sight lengths, a tilted or sinking staff and heat shimmer, and field routine controls them all.

• Two-peg test at the start of a job and after any knock; ISO 17123-2 gives the international field procedure².
• Balanced, short sights of 30–50 m, which cancel most collimation and curvature error.
• Vertical staff and firm change points, using a staff bubble and a change plate or solid kerb.
• Closed loops, with the misclosure set against the project allowance, a constant times the square root of the run length, and then distributed.

Which levelling mistakes are most common?

Booking errors, a wrong benchmark value, an unbalanced set-up, a staff on loose soil and a skipped closing check spoil most level sheets. Habit fixes them, not better equipment.

• Wrong starting value copied from an old drawing, so check it against a second known mark.
• Wrong hair read: the stadia hairs give plausible-looking readings.
• Disturbed change point or loose staff section: a moved plate or unlocked telescopic joint puts a fixed error into every later reading.
• No return run, so none of these errors can show.

Where is levelling used on projects?

A levelling survey serves wherever height matters more than position: floors, plinths, drains, sewers, roads, embankments, contour grids and settlement checks.

• Building work: plinth and floor levels for construction layouts, slab checks and lift pit depths.
• Drainage and roads: inverts and gradients where millimetres decide flow, plus sections for earthwork.
• Land development: spot level grids for contours; our explainer gives more on Contour Survey and how intervals are chosen.
• Monitoring: repeated levels near deep excavations or old structures to detect settlement.

Auto level, total station or DGPS for heights?

For pure heights over short distances the auto level is cheapest and most precise, and a total station survey adds positions. Satellite equipment covers large open areas quickly with centimetre heights. A GPS receiver working off a base station reads GPS signals from the global positioning system and gives heights above the ellipsoid, which must be converted to mean sea level. A level run from a benchmark gives levels directly, so many surveys combine satellite control, total station detail and an auto level run for critical heights.

For a drain that must fall 1 in 500, only the auto level is safe on its own. For a 50-acre contour plan, satellite based positioning from a reference station is far quicker, with an auto level run checking a few critical points.

What does a levelling or contour survey cost?

With Mezux Touch, a contour survey has a ₹15,000 minimum covering up to 1 acre, 1–10 acres cost ₹3,500 per acre, and each acre beyond the first 10 is ₹2,000. Site levelling and layout work is ₹8,000 per crew-day, GST extra. Full detail of what a contour survey costs is on the service page, and the written quote fixes the final figure.

The company keeps six auto levels alongside its total stations and satellite sets. NCR levelling starts on the day you call or the next working day. Outside NCR, a leveller who lives in your area begins at once or within one working day, and the written quote shows any mobilisation or travel cost that applies. Field books, an Excel level sheet and DWG and PDF drawings follow within three working days. Share the site and the levels you need, and get help with your site for your project.

Sources

1. NPTEL, Surveying, Module 4: “Reduction of Level” and “Rise and Fall Method”. archive.nptel.ac.in
2. ISO 17123-2:2001, Optics and optical instruments — Field procedures for testing geodetic and surveying instruments — Part 2: Levels. iso.org

Frequently asked questions

What is the difference between an auto level and a dumpy level?

A dumpy level must be fine-levelled with its long bubble before each reading. An auto level has a compensator that keeps the sight horizontal automatically after rough levelling.

What is HI in levelling?

Height of instrument: the reduced level of the line of sight, equal to the benchmark level plus the backsight.

How often should an auto level be checked?

Run a two-peg test at the start of each job and after any knock or long journey.

Can an auto level measure distances?

Roughly, through the stadia hairs, but it is not a distance instrument, and positions need a total station or satellite survey. There is more on total station work in our guide.

What is a change point?

A firm point where the staff stays while the instrument moves; it takes a foresight from one set-up and a backsight from the next.

Which method is better, HI or rise and fall?

HI is quicker with many intermediate sights; rise and fall gives a fuller check on intermediate readings. Both give the same levels.

Written by Saniya Parveen · Technically reviewed by Rahul Rai · Updated 8 November 2026