Quality Control

Overview#

Quality control in surveying ensures that measurements and deliverables meet required standards of precision, accuracy, and completeness. It is not a single step performed at the end of a project but a continuous process woven through every phase of survey work -- from instrument preparation and field observation to data processing, adjustment, and final deliverable production.

QC encompasses field procedures (closure checks, redundant measurements), data validation (statistical testing, outlier detection), adjustment analysis (residual examination, positional uncertainty), and documentation (field notes, calibration records, adjustment reports). Together, these elements form the framework that allows a surveyor to state, with evidence, that the work meets the standards required.

Professional surveyors are ethically and legally responsible for the quality of their work. A survey that fails to meet applicable standards can result in boundary disputes, construction errors, regulatory rejection, professional liability claims, and loss of licensure. Quality control is not optional -- it is the mechanism by which a licensed professional demonstrates that the seal on a plat or map is warranted.

"The surveyor must exercise care to see that a survey meets the accuracy requirements of the client and the standards of the profession." -- Ghilani & Wolf, Elementary Surveying (13th Ed.), Ch. 2, p. 27

Quality Assurance vs Quality Control#

Quality assurance (QA) and quality control (QC) are related but distinct concepts. Both are essential; neither substitutes for the other.

Quality Assurance is proactive. It consists of the planned, systematic activities implemented before and during a survey to provide confidence that the work will meet requirements. QA prevents problems. It includes:

• Written survey procedures and standards of practice
• Instrument calibration schedules and maintenance programs
• Personnel training and competency verification
• Project planning (control network design, observation schemes, redundancy requirements)
• Standard operating procedures for field and office operations
• Checklists for deliverable completeness

Quality Control is reactive. It consists of the inspection, testing, and verification activities performed on completed work to determine whether it meets requirements. QC detects problems. It includes:

• Closure checks and tolerance comparisons
• Statistical testing of adjustment results
• Blunder detection through residual analysis
• Independent check measurements
• Peer review of computations and deliverables
• Comparison of results against independent data sources

A well-run survey operation invests in both. QA reduces the frequency and severity of problems that QC must catch. QC provides the evidence that QA procedures are working -- and catches the problems that slip through.

Field Checks and Procedures#

The most effective quality control occurs in the field, where problems can be identified and resolved while the crew is still on site. Leaving the field without verifying the work means that any undetected blunder will require a costly return trip.

Closing the Horizon

When measuring all angles around a station, the sum must equal $360°$. Any misclosure reveals the presence of errors:

$\text{Angular misclosure} = \sum \theta_i - 360°$

If the misclosure exceeds the tolerance for the survey class, the angles must be remeasured.

Level Loop Closures

A level circuit that returns to its starting benchmark provides a direct check. The misclosure is the difference between the observed elevation of the closing benchmark and its known elevation. For a loop that starts and ends at the same point:

$\text{Misclosure} = \sum(\text{BS}) - \sum(\text{FS}) - 0$

The misclosure is compared against the allowable tolerance for the survey class. If it exceeds the tolerance, the leveling must be repeated.

Traverse Closures

A closed traverse provides both angular and linear closure checks. The angular misclosure is computed first, and if it falls within tolerance, it is distributed. The linear misclosure is then computed and expressed as a ratio:

$\text{Closure ratio} = \frac{1}{\frac{L}{\sqrt{(\Delta E)^2 + (\Delta N)^2}}}$

where $L$ is the total traverse length and $\Delta E$, $\Delta N$ are the departures and latitudes misclosure. A closure ratio of $1{:}10{,}000$ means the total positional error is one ten-thousandth of the traverse length.

Additional Field Checks

• Check shots to known points. After establishing new stations, observe distances and angles to existing control points. The computed position should agree with the known position within tolerance.
• Direct and reverse observations (face left / face right). Measuring angles in both telescope positions cancels systematic instrumental errors including collimation, trunnion axis tilt, and vertical circle index error. The mean of FL and FR is free of these biases.
• Double-rodded leveling. Running two simultaneous level lines using two rods (at different heights or on different turning points) provides an independent check on every setup.
• Independent measurements for verification. Measuring the same distance by EDM and tape, or checking a GPS-derived position against a total station traverse, provides external validation that no method-dependent systematic error is present.

Tolerances and Closure Standards#

Tolerances define the maximum allowable error for a given survey type and accuracy class. They are the benchmarks against which field results are evaluated.

Angular Closure

For a closed traverse or polygon with $n$ measured angles, the allowable angular misclosure is:

$c = K\sqrt{n}$

where $K$ is the allowable error per angle and $n$ is the number of angles. For example, if the standard allows $10''$ per angle and the traverse has 12 angles:

$c = 10''\sqrt{12} = 34.6''$

Any angular misclosure exceeding $34.6''$ requires remeasurement.

Level Loop Closure

The allowable misclosure for differential leveling is:

$c = K\sqrt{D}$

where $D$ is the total loop distance (in km or miles, depending on the standard) and $K$ is a constant that depends on the accuracy order. For example, second-order Class II leveling allows:

$c = 8\text{ mm}\sqrt{D_{\text{km}}}$

Traverse Closure Ratios

Minimum closure ratios vary by survey purpose:

These are minimum standards. Many jurisdictions and agencies impose stricter requirements, and prudent practice often exceeds minimums.

Blunder Detection#

Blunders -- gross errors caused by human mistakes or equipment malfunction -- are the most dangerous threat to survey quality because they can be large enough to invalidate an entire survey while small enough to escape casual notice. Statistical methods provide a systematic framework for identifying them.

Residual Analysis

After a least squares adjustment, every observation has a residual $v_i$. If the functional and stochastic models are correct and only random errors are present, residuals should follow a normal distribution with a standard deviation close to the a priori estimate $\sigma$.

A common screening criterion: if the absolute value of a standardized residual exceeds a threshold, the observation is flagged as a likely blunder:

$\text{If } \frac{|v_i|}{\sigma_{v_i}} > k, \text{ flag observation } i$

The threshold $k = 3$ corresponds to a 0.3% probability under the normal distribution. An observation whose residual exceeds $3\sigma$ is almost certainly contaminated by a blunder, not merely affected by random error.

Baarda's Data Snooping

Baarda's method formalizes blunder detection using the w-test. For each observation, a test statistic is computed:

$w_i = \frac{v_i}{\sigma_{v_i}}$

where $\sigma_{v_i}$ is the standard deviation of the residual (not the observation). The observation with the largest $|w_i|$ is tested against a critical value determined by the chosen significance level. If it exceeds the critical value, the observation is rejected, the adjustment is repeated, and the process continues until no more outliers are found.

The Tau Test

The tau test is similar in concept but uses the estimated (rather than a priori) variance factor, making it more appropriate when the a priori variance is uncertain. The test statistic follows a tau distribution rather than a normal distribution.

The Role of Redundancy

Statistical blunder detection requires redundancy -- observations in excess of the minimum needed to solve the problem. Without redundancy, there are no residuals, and no statistical test can identify a blunder. A triangle with only three measured angles has no redundancy for the angular observations; add a fourth measurement and a blunder in any one can be detected.

"Redundant observations are essential for detecting mistakes and evaluating the precision of measurements." -- Ghilani & Wolf, Elementary Surveying (13th Ed.), Ch. 3, p. 46

The minimum redundancy for blunder detection is one extra observation. For reliable localization of a blunder (determining which observation is flawed), substantially greater redundancy is needed -- typically at least two or three redundant observations beyond the minimum.

ALTA/NSPS Standards#

The ALTA/NSPS Minimum Standard Detail Requirements for ALTA/NSPS Land Title Surveys represent the most widely referenced standard for boundary survey accuracy in the United States.

Relative Positional Precision

The current standard defines survey quality in terms of Relative Positional Precision (RPP):

The Relative Positional Precision of the survey shall not exceed 0.07 feet + 50 ppm (approximately 2.1 cm + 50 ppm) at the 95 percent confidence level.

This means that the positional uncertainty of any point on the survey, relative to any other point on the same survey, must not exceed 0.07 ft at the 95% confidence level. This is expressed mathematically as:

$\text{RPP} = 2\sigma_r \leq 0.07 \text{ ft} + 50 \text{ ppm}$

where $\sigma_r$ is the standard deviation of the relative position between two points.

What This Means in Practice

A 95% confidence level corresponds to approximately $2\sigma$ for a two-dimensional positional quantity (more precisely, $2.447\sigma$ for a circular error probable at 95%). To achieve an RPP of 0.07 ft + 50 ppm, the underlying measurement precision must support a relative positional standard deviation of approximately:

$\sigma_r = \frac{0.07}{2} = 0.035 \text{ ft}$

This requires high-quality instruments (1" or 2" total stations, survey-grade GNSS), careful field procedures, and properly designed observation networks.

Demonstrating Compliance

Surveyors demonstrate compliance with the ALTA/NSPS positional precision standard through:

• Least squares adjustment of the survey network with positional uncertainty computed for all points
• Verification that the largest relative positional uncertainty between any two surveyed points does not exceed the 0.07 ft threshold
• Documentation of the methodology, instruments, and adjustment results in the survey deliverables

FGCS Standards#

The Federal Geodetic Control Subcommittee (FGCS) established a classification system for geodetic control surveys that defines orders and classes based on accuracy requirements. These standards apply to the National Spatial Reference System (NSRS) and are widely used as benchmarks for control survey quality.

Horizontal Control Standards

Leveling Section Misclosure Tolerances (FGCS Sec. 3.5)

"The FGCS classification system provides a uniform framework for specifying, achieving, and evaluating the accuracy of geodetic control surveys." -- Federal Geodetic Control Subcommittee, Standards and Specifications for Geodetic Control Networks (1984)

Practical Implications

Higher-order control requires better instruments, longer observation sessions, more redundancy, and more rigorous adjustment procedures. The cost and time increase significantly with each order. Survey projects should specify the order that matches the actual accuracy requirements -- overspecifying wastes resources, while underspecifying compromises the work that depends on the control.

Instrument Calibration#

Instruments that are out of calibration introduce systematic errors into every measurement they make. Regular calibration is not merely good practice -- it is a professional obligation and, for many survey types, a regulatory requirement.

Total Station Calibration

• Collimation error (line-of-sight): Checked by observing a target in face left and face right. The difference divided by two gives the collimation error. Should be within manufacturer specifications (typically $< 5''$ after adjustment).
• Trunnion axis error: Also detected through FL/FR observations on steeply inclined targets.
• Vertical circle index error: Checked by observing the same target in both faces and comparing the zenith angle sum to $360°$.
• EDM calibration: Measured distances are compared against known baseline distances. The NGS maintains a network of EDM calibration baselines throughout the United States. Testing determines scale error (proportional) and zero error (constant offset, including the prism constant).

Level Calibration: The Two-Peg Test

The two-peg test (or peg test) detects collimation error in a differential level. The procedure:

1. Set two points A and B approximately 30 m (100 ft) apart
2. Set up the level at the midpoint and read rods on A and B -- the true elevation difference is obtained because any collimation error cancels when BS and FS distances are equal
3. Move the level close to one point (say, within 3 m of A) and read both rods again
4. If the computed elevation difference from the unbalanced setup disagrees with the midpoint setup, collimation error is present

The allowable collimation error depends on the level class and survey order. If the error is excessive, the instrument must be adjusted.

GNSS Receiver Calibration

• Antenna calibration: GNSS antenna phase center offsets and variations are determined through calibration. The NGS publishes antenna calibration data for most survey-grade antennas. Using uncalibrated antenna models introduces systematic position errors.
• Baseline testing: Receivers are periodically tested by observing known baselines and comparing results to published values.

Documentation

Every calibration event should be documented with:

• Date of calibration
• Instrument serial number and model
• Calibration baseline or reference used
• Results (measured vs. known values, computed corrections)
• Pass/fail determination against applicable standards
• Name of person who performed the calibration

Calibration records should be maintained for the life of the instrument and be available for audit.

Documentation and Records#

The work product of a survey is not just the plat or the coordinates -- it is the entire body of evidence that supports those results. Documentation is the medium through which a surveyor demonstrates that the work was done correctly, that the results meet standards, and that the conclusions are defensible.

Essential Records

• Field notes. The original record of observations made in the field. Whether in a traditional field book or a digital data collector, field notes should be complete, legible, and unambiguous. They should record what was measured, how it was measured, what conditions prevailed, and who made the measurements.
• Metadata. Instrument models and serial numbers, calibration dates, software versions, coordinate system and datum, units, observation dates and times, weather conditions.
• Adjustment reports. The complete output of the least squares adjustment: observations, residuals, adjusted coordinates, standard deviations, error ellipses, relative positional accuracies, variance factor, chi-square test results.
• Calibration certificates. Documentation that instruments were calibrated and met specifications at the time of use.
• Computation files. All intermediate calculations, coordinate transformations, and data reductions performed between raw observations and final coordinates.

Chain of Evidence

Every deliverable should be traceable back to the original field observations through a clear chain of evidence:

$\text{Field Observations} \rightarrow \text{Data Reduction} \rightarrow \text{Adjustment} \rightarrow \text{Final Coordinates} \rightarrow \text{Map/Plat}$

If any link in this chain is missing or unclear, the defensibility of the survey is compromised.

Legal Defensibility

Survey work frequently becomes evidence in legal proceedings -- boundary disputes, construction claims, eminent domain cases, title insurance claims. Documentation that is incomplete, disorganized, or missing altogether undermines the surveyor's testimony and may expose the surveyor to professional liability. A survey that cannot be independently verified from its own records is, for practical purposes, unverifiable.

"Good field notes are the most important part of surveying. No matter how carefully measurements are made, if they are not properly recorded, the work is of little value." -- Ghilani & Wolf, Elementary Surveying (13th Ed.), Ch. 1, p. 15

Courts give significant weight to contemporaneous records -- documents created at the time the work was performed. Field notes made on the day of the survey carry far more evidentiary weight than a reconstruction made months later from memory.

Key Takeaways#

• QA is proactive; QC is reactive. Quality assurance prevents problems through planning, training, and calibration. Quality control detects problems through checking, testing, and statistical analysis. Both are essential.
• Field checks are the first line of defense. Closing the horizon, level loop closures, traverse closures, check shots, and FL/FR observations catch errors while the crew is still on site.
• Tolerances are quantitative thresholds. Angular closure ($K\sqrt{n}$), level closure ($K\sqrt{D}$), and traverse closure ratios provide objective criteria for accepting or rejecting field work.
• Blunder detection requires redundancy. Without measurements in excess of the minimum, blunders cannot be detected statistically. Baarda's data snooping and the tau test formalize outlier detection.
• ALTA/NSPS requires 0.07 ft + 50 ppm RPP at 95% confidence. This relative positional precision standard demands high-quality instruments, careful procedures, and rigorous adjustment.
• FGCS standards classify control surveys by order and class. Higher orders demand greater accuracy, more redundancy, and more sophisticated procedures.
• Calibrate instruments regularly and document it. EDM baselines, two-peg tests, collimation checks, and GNSS antenna calibrations prevent systematic errors from contaminating every measurement.
• Documentation is the foundation of defensibility. Field notes, adjustment reports, calibration records, and computation files form the chain of evidence that supports the survey. If it is not documented, it did not happen.

References#

1. Ghilani, C.D. & Wolf, P.R. Elementary Surveying: An Introduction to Geomatics (13th Ed.). Pearson, 2012. Chapters 2--3, 9--10.
2. Ghilani, C.D. Adjustment Computations: Spatial Data Analysis (6th Ed.). Wiley, 2017. Chapters 2, 21.
3. ALTA/NSPS. Minimum Standard Detail Requirements for ALTA/NSPS Land Title Surveys. 2021.
4. Federal Geodetic Control Subcommittee. Standards and Specifications for Geodetic Control Networks. NOAA, 1984.
5. Federal Geodetic Control Subcommittee. Geometric Geodetic Accuracy Standards and Specifications for Using GPS Relative Positioning Techniques. NOAA, 1989.
6. Mikhail, E.M. & Gracie, G. Analysis and Adjustment of Survey Measurements. Van Nostrand Reinhold, 1981.
7. National Geodetic Survey. EDM Calibration Baselines. https://www.ngs.noaa.gov/CBLINES/.