3.1 Control Surveys: Purpose and Procedures

3.1 Control Surveys: Purpose and Procedures

Quick Answer: A control survey builds a stable geometric skeleton—plan positions and elevations—so every later topographic shot, design coordinate, and construction stake refers to the same origin, orientation, and datum. Distinguish primary, secondary, and working control; recover published monuments when you need a named datum; use a documented assumed system when the project is local; and never promote construction stakes into the role of primary control.

Why Control Exists on Engineered Projects

Business and Professions Code (BPC) 6731.1 authorizes a California civil engineer to locate, relocate, establish, reestablish, or retrace alignment or elevation for the fixed works listed in BPC 6731, and to determine the configuration of the earth's surface. None of that work is trustworthy unless a control survey first answers three questions:

1. Where is the origin?
2. Which way is north, or what azimuth does a project axis hold?
3. What is elevation zero, and on which vertical datum?

Without those answers, two crews can occupy the same site and produce maps that do not overlay. A grading contractor can cut to a plan elevation that is 1.2 ft off a storm-drain invert because the design used the North American Vertical Datum of 1988 (NAVD 88) while the field crew held an assumed benchmark on a fire-hydrant flange. Control is not a convenience item on a fee proposal. It is the only way design and construction share a geometry.

On a typical California civil site—a street widening, a detention basin, a building pad, a channel improvement—the control survey is the first field effort after reconnaissance. Everything that follows in this chapter (route geometry in Section 3.2, topography in Section 3.3) and later construction staking inherits that skeleton. If the skeleton is wrong, every subsequent measurement is systematically wrong, and the error will not average out in a pile of random shots.

Horizontal Control vs Vertical Control

Horizontal control locates points in a plane: northing and easting, or station and offset from an alignment. It is established by traversing with a total station, by static or real-time kinematic Global Navigation Satellite System (GNSS) observations, or by tying to recovered monuments whose published coordinates are held.

Vertical control locates points in elevation. Differential leveling remains the workhorse for construction-grade elevations. Trigonometric leveling and GNSS ellipsoid heights converted through a geoid model are common on larger sites, but they are still vertical control. They do not automatically fix horizontal position.

Treat the two systems as independent unless you have deliberately built a three-dimensional network. A crew can set excellent horizontal control on an assumed origin while leveling to an NAVD 88 benchmark—or the reverse. Mixing them without documentation is a classic field and exam trap: the map looks fine in plan, then storm inverts do not match the hydraulic profile.

Named datums (NAD 83, CCS83, NAVD 88) are developed in Chapter 5. The planning point here is simpler: decide what you are holding, write it on the control diagram, and do not silently switch mid-job.

Primary, Secondary, and Working Control

Think of control as a hierarchy, not a pile of hubs.

Primary control is the fewest, strongest points. It either holds published coordinates and elevations or defines a well-documented assumed origin, rotation, and scale. Observations are redundant: closed traverses, reciprocal measurements, closed level loops. Primary points are monumented to last for the life of design and construction—often months to years on a California public-works job.

Secondary control densifies the primary network so that topography and layout can be run without long sights. It is still redundant. A street job might have primary points at each end of the corridor and secondary points every few hundred feet where a total station can occupy with a short backsight.

Working control (sometimes called construction control) is set for daily use: a hub near a catch basin, a nail in asphalt for a day's curb layout. Working points may be lower order, but they must be checked back into secondary or primary control. They are disposable. They are not the project's geometric truth.

Exam trap: construction stakes as primary control

Offset stakes, grade stakes, and blue tops exist to communicate a designed surface to a contractor. They are products of control, not sources of control. If a primary monument is destroyed, you recover or re-establish from other primary or published points—not by measuring from a chewed-up curb stake whose elevation was set to a finished-grade call. A stake that was correct on Tuesday can be knocked, paved over, or reset to a revised grade by Wednesday. Promoting it to primary control launders that movement into every later layout.

Recovering Published Monuments vs Establishing Project (Assumed) Control

Published monuments come from the National Geodetic Survey (NGS), Caltrans, counties, cities, and water agencies. Recovering them means find the physical mark, confirm the stamping and description, record the published coordinates or elevation and the datum/epoch, and check into a second mark whenever possible. One recovered disk with no check is a spur, not a network.

Project (assumed) control is a local origin the engineer chooses. A common California civil pattern is:

1. Reconnaissance finds durable on-site features (existing building corners used as site geometry, not as a boundary determination; existing street monuments used as construction reference).
2. The crew occupies two or more points, assigns assumed coordinates to the origin hub — in the worked example below, N = 5,000.00 ft and E = 10,000.00 ft on Hub A — plus an assumed elevation and an assumed bearing on line A–B.
3. A closed traverse and a closed level loop—or a check into a second site benchmark—prove the network.
4. Hold that system for design, mapping, and construction. Do not shift to State Plane in the middle of the job unless the entire dataset is rigorously transformed and the change is documented on the plans.

Assumed control is appropriate when the project does not need to match a regional GIS layer to the tenth of a foot, when published marks are distant or destroyed, or when the design is entirely local (a private parking lot, a small detention basin). The moment you must overlay city utilities in GIS, match an adjacent Caltrans contract, or report elevations to a floodplain administrator, you need a named datum—or a documented transformation from assumed to that datum.

Horizontal and vertical decisions can split: hold an assumed horizontal origin while tying vertical control to NAVD 88. That split is professional only if the plans state it. Ignoring the vertical datum while you set horizontal control is how a job gets two elevations for the same hub.

Network vs Spur

A network has redundant paths. A four-sided closed traverse can compute latitudes, departures, and a linear misclosure. A level loop that returns to the starting benchmark, or that closes on a second published benchmark, can compute a vertical misclosure. Redundancy is how you know a blunder occurred.

A spur is an open-ended traverse or a one-way level run that never checks. Spurs are sometimes unavoidable (a side shot to a tree, a temporary point in a trench). They are not primary control. If a procedure never closes and never occupies a second known point, treat the result as unchecked.

Typical California Civil Workflow

For many site and municipal jobs the sequence is:

1. Reconnaissance — walk the site, hunt for published marks and durable site features, note traffic, vegetation, and intervisibility (can Station 1 see Station 2?).
2. Design the network — primary points with long, clear sights; secondary points to cover topography; avoid occupying a future excavation or a pavement grind.
3. Establish or recover control — assumed origin tied to site control, or ties to published monuments.
4. Observe redundantly — close the traverse; double-run or loop the levels; occupy a known point as a check.
5. Hold the system — design drawings, quantity takeoffs, and later construction staking all use the same coordinates and datum.

Intervisibility still matters with total stations. GNSS reduces line-of-sight needs but does not remove the need for a vertical-datum decision or for checking into known elevations. A GNSS-derived ellipsoid height that is never converted, or never checked to a leveled benchmark, is not vertical control you can stake a sewer invert from.

Worked Example: Closed Primary Traverse on an Assumed Origin

Hub A is assigned N = 5,000.00 ft, E = 10,000.00 ft on an assumed horizontal datum. A four-sided primary traverse A–B–C–D–A yields these latitudes (change in northing) and departures (change in easting):

Lengths are the hypotenuses of each latitude–departure pair. For A–B: √(412.18² + 185.40²) = √204,265.51 = 451.96 ft. The other three courses follow the same square-root step.

Sum of latitudes: 412.18 − 96.55 − 418.30 + 102.55 = −0.12 ft
Sum of departures: 185.40 + 528.72 − 102.15 − 611.85 = +0.12 ft
Perimeter: 451.96 + 537.46 + 430.59 + 620.38 = 2,040.39 ft

Linear misclosure = √[(−0.12)² + (0.12)²] = √0.0288 = 0.170 ft
Relative precision = 0.170 / 2,040.39 ≈ 1:12,000

The network closes, so you have something to report. Whether 1:12,000 is good enough is a project-specification question (mapping accuracy, construction tolerance), not a universal cutoff the Board publishes for the Civil Engineering Surveying exam. The exam point is the method: you cannot claim primary control from an unclosed spur with no misclosure to report.

Coordinates of B from A: N = 5,000.00 + 412.18 = 5,412.18 ft, E = 10,000.00 + 185.40 = 10,185.40 ft. After adjustment of the −0.12 ft and +0.12 ft misclosures (Chapter 8), those values will shift by a few hundredths. Unadjusted, they are still far more defensible than a single side shot from a curb nail.

Worked Example: Vertical Check into a Published Benchmark

A level circuit starts on BM-12, published NAVD 88 elevation 247.32 ft. Sum of backsights = 12.46 ft; sum of foresights = 8.58 ft. The computed elevation of BM-18 (published 251.18 ft) is 247.32 + 12.46 − 8.58 = 251.20 ft. Misclosure = 251.20 − 251.18 = +0.02 ft. The loop provides a check. If the same crew had stopped at a construction hub with no return run, that +0.02 ft—or a 0.20 ft blunder—would be invisible. Closing on a second published mark is the vertical counterpart of closing a traverse.

Procedures Checklist

• Reconnaissance: existing marks, access, hazards, and whether photo control will be needed later.
• Intervisibility: plan occupy/backsight pairs; clear lines or add intermediate stations.
• Redundant observations: closed figures, reciprocal zenith angles if using trigonometric heights, double-tied GNSS bases.
• Checks into known points: a second published benchmark, a second recovered monument, or a previously adjusted primary hub.

Exam Traps to Memorize

1. Mixing control types — treating a secondary hub's coordinates as if they were published NAD 83 values, or combining assumed elevations with NAVD 88 inverts on the same profile without a note.
2. Using construction stakes as primary control — restaking a destroyed primary from an offset lath.
3. Ignoring the vertical datum when setting horizontal control — a perfect assumed northing/easting system with an unlabeled elevation origin.

If an item describes a beautiful GNSS localization that never mentions a benchmark, ask what elevation the contractor will actually build to.