5.2 Vertical Datums and Benchmarks

5.2 Vertical Datums and Benchmarks

Quick Answer: A vertical datum is the zero and the definition of elevation on the job. Civil plans use orthometric heights—typically NAVD 88, a local city datum, legacy NGVD 29, or an assumed benchmark. Recover a published benchmark with its description and a check, or set a project benchmark through a closed level loop. A GNSS ellipsoid height is not an NAVD 88 elevation.

Orthometric Height vs Ellipsoid Height

Orthometric height (H) is height above the geoid—the quantity that civil drawings call elevation when they name NAVD 88 or a local city vertical datum. Ellipsoid height (h) is height above the reference ellipsoid (GRS 80 for NAD 83). They differ by the geoid height (N) in the relation h = H + N. In California that difference is tens of feet, not hundredths.

A rover that reports h = 412.88 ft next to a leveled NAVD 88 benchmark of 498.20 ft is not 85 ft off in the rod. It is reporting a different kind of height. Do not print 412.88 ft as a finished-floor elevation. Converting ellipsoid heights through a geoid model (GEOID12B, GEOID18, and their successors) is how GNSS can support orthometric elevations. That conversion is not a 2022 CES core skill, and it does not replace a check into a leveled benchmark. The exam point: know they are different, and check.

Differential leveling measures orthometric differences (with curvature, refraction, and collimation treated in later field chapters). That is why construction-grade inverts still come from a level loop, not from an unchecked ellipsoid height. Trigonometric leveling from zenith angles can transfer orthometric height too, but it is still not an ellipsoid height, and it still needs a datum origin.

NAVD 88, NGVD 29, and Assumed Local Benchmarks

NAVD 88 is the current national orthometric datum most California public-works and FEMA-related jobs name. NGVD 29 (National Geodetic Vertical Datum of 1929) appears on older as-builts, older city benchmark lists, and some legacy flood documents. They are not the same zero.

The difference is location-specific. NGS tools (historically VERTCON; now the NGS Coordinate Conversion and Transformation Tool, NCAT) interpolate a model. There is no California-wide "add 2.5 feet" that is safe to memorize as a constant. A city control diagram or a project-specific conversion note is what you use, and you write the source on the plans.

Assumed local benchmarks assign a convenient elevation—often 100.00 ft or 500.00 ft—to a durable on-site mark. They are appropriate on the same small, isolated jobs that use assumed horizontal coordinates, when no published vertical datum is required. They fail when a storm drain must match a city invert on NAVD 88 and nobody recorded the difference between 100.00 assumed and the published system.

NAPGD2022, mentioned in Section 5.1, is the planned NGS replacement for NAVD 88. It is not what most 2026 California civil plan sets still name. Until a job's title sheet says otherwise, treat NAVD 88, NGVD 29, city, and assumed as the practical menu.

Worked example: NGVD 29 invert onto an NAVD 88 plan

A city's project note (not a statewide constant) states: NAVD 88 = NGVD 29 + 2.47 ft at this site. An as-built 18-inch storm invert is 318.20 ft NGVD 29.

NAVD 88 elevation = 318.20 + 2.47 = 320.67 ft

If the new profile is labeled NAVD 88 but the designer copies 318.20 ft as the match-in invert, the contractor is 2.47 ft low at the connection. That is larger than most pipe-grade tolerances on a 0.40% sewer (0.40 ft per 100 ft). Subtracting 2.47 ft instead of adding is the equal-and-opposite trap: take the sign from the local note. In much of California the NAVD 88 number is higher than the NGVD 29 number, but you still do not invent 2.47 ft for a different city.

A second numeric check: a first-floor elevation of 412.00 ft NGVD 29 becomes 412.00 + 2.47 = 414.47 ft NAVD 88. A floodplain administrator reading 412.00 ft on an NAVD 88 certificate has the building 2.47 ft lower than the paper says.

Recovering a Published Benchmark vs Setting a Project Benchmark

Recovering a published benchmark means finding the physical mark, matching it to the description, and confirming you are on the published datum.

A typical description looks like: BM CITY 12 — bronze disk in concrete headwall, stamped CITY 12 1988, 2.4 ft south of the north face of the headwall, 18.0 ft west of the east wingwall. Published 247.32 ft NAVD 88.

Recovery checklist:

1. Description — directions, distances, and stamped text match what you see. If the datasheet says a disk in a headwall and you found a punch mark in a curb, you found a different mark.
2. Condition — not leaning, not in a shattered headwall, not buried under a new overlay. A disk under 0.3 ft of asphalt is not recovered; it is buried. Do not level to a "maybe" and hope the overlay is uniform.
3. Published metadata — designation, PID if it is an NGS mark, datum, and any epoch or supersession note. A 1974 NGVD 29 elevation restamped on a disk that a 2008 list republished as NAVD 88 is two different numbers on one piece of brass.
4. Check — level to a second published mark, or close a loop back to the first. One mark with no check is a spur.

Setting a project benchmark means monumenting a durable on-site point (disk, punch mark, or well-described structure), leveling from the chosen origin (published or assumed), closing the loop, and putting the name, elevation, and datum on the control diagram. Project BMs exist because the published disk may be a mile away, in traffic, or destined to be demolished by the same grading you are supporting.

Do not call a hub-and-tack from Tuesday's curb layout a project BM. That is working control (Section 5.3). If the curb is ground off on Wednesday, your vertical datum disappeared with it.

Worked example: transferring a BM with a closed level loop

BM-City is recovered and accepted: 247.32 ft NAVD 88. The crew runs a five-setup loop around a one-block site to set BM-P1 on a concrete wall that will survive grading, then closes back on BM-City.

Height-of-instrument steps:

• 247.32 + 5.26 = 252.58; 252.58 − 2.87 = 249.71 (TP1)
• 249.71 + 6.41 = 256.12; 256.12 − 3.12 = 253.00 (TP2)
• 253.00 + 4.18 = 257.18; 257.18 − 3.07 = 254.11 (BM-P1)
• 254.11 + 1.94 = 256.05; 256.05 − 3.12 = 252.93 (TP3)
• 252.93 + 3.55 = 256.48; 256.48 − 9.13 = 247.35 (close)

Independent check using sums:

ΣBS = 5.26 + 6.41 + 4.18 + 1.94 + 3.55 = 21.34 ft
ΣFS = 2.87 + 3.12 + 3.07 + 3.12 + 9.13 = 21.31 ft
Computed close = 247.32 + 21.34 − 21.31 = 247.35 ft
Misclosure = 247.35 − 247.32 = +0.03 ft

The table and the ΣBS − ΣFS check agree, so the arithmetic is not the bust. The loop closes. Whether +0.03 ft is acceptable is a project specification question. The 2022 CES materials do not publish a Board-mandated third-order formula as a cut score. Many job specs still compare misclosure with a form of k√L (L in miles or kilometers). The exam skill is to compute the misclosure and to refuse an unclosed spur, not to invent a Board number.

If the specification requires adjustment, a simple distribution by setup count puts 3/5 of the +0.03 ft on BM-P1 (three setups from the start): correction = −0.03 × 3/5 = −0.018 ft, adjusted BM-P1 = 254.09 ft. If the specification holds published marks and accepts 0.03 ft, you may keep 254.11 ft unadjusted. Either way, report the misclosure on the control notes. An open run that "sets" BM-P1 at 254.11 ft with no return to BM-City is not a project benchmark—it is a one-way opinion.

A second published mark, BM-School at 251.88 ft NAVD 88, is a stronger check than closing only on the start. If the same methods computed 251.94 ft at BM-School, the +0.06 ft misclosure would send you back to descriptions, disturbed marks, and the possibility that one published elevation is still on NGVD 29. Two marks that disagree by 0.06 ft after a short loop are a datum or recovery problem, not a reason to average 247.32 and 251.88 and invent a new zero.

Exam traps

• Labeling ellipsoid heights as NAVD 88.
• Mixing NGVD 29 as-builts into NAVD 88 profiles without a local conversion.
• Adopting a buried or disturbed disk because the stamping is still legible in a photo.
• Setting a project BM on an open level run.
• Assuming a statewide NGVD 29 ↔ NAVD 88 constant.
• Using an assumed 100.00 ft BM on a job that must match city NAVD 88 inverts, with no recorded difference.