How to check a roughness survey on a TMH 13 validation section

A validation section is a short, marked length of road whose profile a reference device measured before the survey, giving a reference international roughness index (IRI) for every 100 m. To check a roughness survey on one, run the survey device over it and compare each 100 m segment it reports with the reference IRI. The South African road manual TMH 13, Automated Road Condition Assessments, Part C: Roughness (Committee Draft Final, May 2016), calls this test calibration for response-type devices, which sense suspension movement, and validation for profilers, which measure the road profile; the two are different tests. (COTO, 2016a)

Key facts

• In the South African road manual TMH 13 Part C (Committee Draft Final, May 2016), a response-type roughness device is calibrated and a profiler is validated.
• TMH 13 Part A, section A.4.4, says validation sections should be profiled before any survey, ideally by the network agency or a contractor other than the one surveying.
• TMH 13 Part C, section C.4, says profiler validation does not adjust the output to match a benchmark, and a profiler that is not valid is fixed by the manufacturer.
• In TMH 13 Part C, profiler validation is carried out for a specific operator and measurement protocol (section C.4), so the validation file should name the vehicle, the operator and the speeds used.
• TMH 13 Part C gives no maximum age for a profiler validation, and COTO Standard Specifications Chapter 20 (Draft Standard, October 2020), clause A20.1.5.5, sets one unless the contract says otherwise.
• In TMH 18 road-data files, neither Draft Version 4 nor Committee Draft 5.0 has a column for device, class or validation date in the combined instrument data record.

TMH (Technical Methods for Highways) is a series of South African road manuals compiled under the Committee of Transport Officials (COTO). The IRI is a roughness measure in m/km, computed from a longitudinal road profile; the insight on interpreting IRI in South Africa explains what one value does and does not say.

What is a roughness validation section?

On a validation section a reference device has measured, before the survey starts, the profile of each wheelpath (the line a vehicle wheel follows along the lane), and from it the IRI for every 100 m. Part C calls these sites calibration sections for both kinds of test and applies the same selection and reference-profiling requirements to both (sections C.3.1 and C.4.1). (COTO, 2016a)

TMH 13 Part A: General (Committee Draft Final, May 2016), section A.4.4, says the sections should represent the network, avoid rapid deterioration and carry GPS (global positioning system) coordinates and markers so the test can be repeated. They should be profiled before any survey, ideally by the network agency or a contractor other than the one surveying. (COTO, 2016d) Table C.5 in Part C lists the roughness ranges the set of sections should cover; it is not a scale of road condition.

Is a roughness test a calibration or a validation?

The name depends on the device: a response-type device is calibrated and a profiler is validated. A response-type device, also known as a road meter system, records suspension movement rather than the profile, so a regression (a straight-line fit) against reference IRI values converts its output to IRI; that regression is the calibration, made as part of the survey. A profiler measures the profile with lasers and accelerometers and has factory-calibrated components, so in the field it is validated: the test confirms that the whole system, as operated, is accurate and repeatable (sections C.2.2, C.2.3 and C.4). (COTO, 2016a)

Part C adds a prohibition: a profiler is never calibrated by comparing its output with the reference IRI, and one that is not valid is repaired by the manufacturer, not adjusted. (COTO, 2016a)

Question	Response-type device (C.3)	Profiler (C.4)
What the test establishes	A regression that converts output to IRI	That the system is accurate and repeatable
Holds for	The vehicle configuration used	The operator and protocol used

Narrated film · 1:27

Checking a roughness validation section

A step rail of the checks beside five synthetic repeat runs set against a reference line on one section, closing on the difference between calibration and validation.

Read the film transcript

How do you check a roughness survey against a validation section? A validation section is a marked length of road that a reference device profiled before the survey. It gives a reference international roughness index, the IRI, for every one hundred metres. In the South African road manual TMH 13 Part C, a response-type device is calibrated and a profiler is validated. A response-type device senses suspension movement, so a regression against the reference calibrates it. A profiler measures the road profile, so it is validated, never adjusted to match the reference. We take the checks in a working order, beginning with the device, class and certificates. Then come the reference profile, positioning, vehicle and speeds, a plot, and judgement against the contract. In this synthetic example the reference is five point two metres per kilometre. The five runs read five point four or five point five, an average of zero point two four above it. Their spread is only zero point one, so bias, not repeatability, is the finding. Keep the device, section, reference, runs, certificates and decision together in one pack. The reliability level, acceptance criteria and validity period are left to the specification or client. Calibration and validation are different tests, and tight repeats alone never prove accuracy. Next comes TMH 18, whose road data files have no column for validation date, so the pack travels with the data.

How do you check a roughness survey on a validation section?

Work through eleven checks in a working order drawn from Parts A to C, not one the manual prescribes; early checks decide whether later calculations mean anything.

1. Identify the device, its class and the test. Record make, model, serial number and the class under Table C.1; a response-type device follows section C.3, a profiler section C.4. Draft TMH 22, Road Asset Management Manual (March 2013), section D.4.3, ties the permitted class to the road class. (COTO, 2016a) (COTO, 2013)
2. Collect the pre-validation paperwork. Ask for current calibration certificates for the system components, including the reference device. Part A, section A.4.3.3, separates component calibration from system validation. (COTO, 2016d)
3. Check the reference profile. Confirm that a Class 1 device, such as a precision rod and level, profiled each wheelpath four times, as two closed loops returning along the same wheelpath, and that the reference IRI is the mean of the runs, computed for every 100 m by the ASTM International E1926-98 practice (section C.3.1.2). (COTO, 2016a)
4. Check positioning first. Confirm that the distance-measuring instrument (DMI) and GPS were validated, because a distance error shifts every 100 m segment against its reference. TMH 13 Part B: Positioning (Committee Draft Final, May 2016), section B.3.3, describes the checks. (COTO, 2016c)
5. Match vehicle, operator, speeds and run counts. Use the survey's vehicle, operator and speeds, because validation is carried out for a specific operator and protocol and a response-type calibration holds only for the vehicle configuration used (sections C.3.4 and C.4). Compare the sections and repeat runs with the contract; Tables C.6 and C.8 give the manual's recommended numbers. (COTO, 2016a)
6. Plot before calculating. Plot measured against reference values, with a line of equality for a profiler. Appendix C-2 says data lying consistently above or below that line indicate a systematic error. (COTO, 2016a)
7. Run the manual's checks. For a response-type device: linearity, coefficient of determination and standard error (Table C.7), and repeatability on each section (section C.3.2). For a profiler: repeat-run variability, the regression between measured and reference values (per section and overall), the error on each 100 m segment and agreement between days (Table C.9). (COTO, 2016a)
8. Judge against the contract's criteria. Tables C.7 and C.9 are guidelines that should not be enforced too rigidly, and the network manager may relax them for conditions such as localised distress. Record any relaxation and who agreed it. (COTO, 2016a)
9. Check the age of the validation. Compare its date with the survey dates. Part C gives no maximum age for a profiler validation, although section C.2.2.3 sets a minimum recalibration frequency for response-type devices; COTO Standard Specifications for Road and Bridge Works for South African Road Authorities, Chapter 20: Quality Assurance (Draft Standard, October 2020), clause A20.1.5.5, sets one unless the contract says otherwise. (COTO, 2020a)
10. Plan the control tests. Re-measure control or validation sections at intervals through the survey. If a profiler fails one, discard and re-measure the data since its last successful control test; check a failing response-type device for defects first, then recalibrate and re-measure if none is found. Never use control data to adjust a device (sections C.3.4 and C.4.4). (COTO, 2016a)
11. Assemble the file. Put the evidence listed below into one pack a reviewer can follow.

What does a worked roughness check look like?

Synthetic example; every value is invented. A profiler is validated on Synthetic Road S1, validation section V1, left wheelpath, over the 100 m segment from chainage (distance along the road from its start) 1,200 m to 1,300 m.

Four reference runs give 5.15, 5.25, 5.20 and 5.20 m/km, so the reference IRI is (5.15 + 5.25 + 5.20 + 5.20) / 4 = 5.20 m/km. Five repeat runs of the survey device at one speed give 5.4, 5.5, 5.4, 5.5 and 5.4 m/km, with a mean of 27.2 / 5 = 5.44 m/km; a real validation follows the run and speed pattern in Table C.8. The differences from the reference are +0.2, +0.3, +0.2, +0.3 and +0.2 m/km, averaging +0.24 m/km (5.44 − 5.20).

The runs span only 0.1 m/km, yet all five lie above the reference and the offset exceeds the whole spread. Repeatability is good; bias is the finding, and a check of scatter alone would miss it, as the insight on repeatability and uncertainty explains. Table C.9 and Appendix C-2 express each segment error as a percentage of the reference, so a checker divides each difference by 5.20 m/km before applying the contract's criterion. (COTO, 2016a)

One segment neither validates nor rejects a device; Table C.9 applies its checks across all segments, sections, runs and speeds. On a response-type device the calibration regression, fitted to the reference values, absorbs an offset like this.

What records should a roughness validation file keep?

The file should hold:

• The device: make, model, serial number, class and IRI software.
• The section: road, identifier, start and end chainage, GPS coordinates, direction, wheelpath, and photographs of the markers and surface.
• The reference: device, date profiled, and each reference run's 100 m IRI.
• The vehicle and operator: vehicle, operator, speeds and, for a response-type device, occupants, loading and tyres (sections C.3.2 and C.5.1).
• The runs: date, time, speed, run number and 100 m IRI for each wheelpath, with raw data.
• The report: regression summary, other check results and the reliability level the contract chose.
• The certificates: component calibration certificates, including the reference device. (COTO, 2016d)
• The decision: who accepted the result, the criteria applied, any relaxation, and the date the validation ages out under the contract. (COTO, 2020a)
• The control tests carried out during the survey.

The delivery file does not carry this evidence. TMH 18, Road Asset Data Electronic Exchange Formats (Draft Version 4, October 2014; Committee Draft 5.0, February 2016), defines the combined instrument data record in section 5.6 with road, lane, start and end kilometre, date and IRI by wheelpath. Neither version has a column for device, class or validation date. (COTO, 2014) (COTO, 2016i) Send the pack with the data, naming the TMH 18 version the receiver expects.

What does TMH 13 leave to the specification or the client?

TMH 13 is a Committee Draft Final, and its own Document Versions note gives committee drafts no legal standing; the contract documents set which parts apply, as COTO Standard Specifications Chapter 20 does by calling up the draft TMH 13 protocols. (COTO, 2016a) (COTO, 2020a) TMH 13 leaves six decisions to the specification or the client:

• The reliability level. Tables C.6 and C.8 offer lower and higher reliability levels, chosen from the survey objectives and the TMH 22 collection frequency.
• The acceptance criteria. The contract should let the client override requirements where special circumstances need it (section C.3.4).
• The validity period. Part A says the specification should state how long a validation is valid. (COTO, 2016d)
• Who validates. Clause A20.1.5.5 of COTO Standard Specifications Chapter 20 has high-speed profilers validated by the Employer or an approved agent, following the draft TMH 13 protocols. (COTO, 2020a)
• A new device or contractor. Section C.4.1 says a higher-level validation with spectral analysis can be considered and is not covered. (COTO, 2016a)
• What the IRI means for treatment. Part C concerns measurement. The reporting format is in TMH 18 (section C.5.3).

What are the common misreadings of TMH 13 Part C?

"Calibrated" and "validated" mean the same. Not so: response-type devices are calibrated, profilers validated (sections C.2.3 and C.4). (COTO, 2016a)

A manufacturer's certificate validates the system. It covers components. TMH 13 Part B says calibration does not confirm that the parameter is appropriate when measured from a moving vehicle on the network (section B.3). (COTO, 2016c)

Tight repeats show accuracy. Not so: scatter measures repeatability and cannot reveal a common offset.

The procedure, step by step

1. Identify the device, its class and the test

   Record make, model, serial number and the class under Table C.1 of TMH 13 Part C. A response-type device follows section C.3 and a profiler follows section C.4. TMH 22 section D.4.3 ties the permitted class to the road class.

2. Collect the pre-validation paperwork

   Ask for current calibration certificates for the system components, including the reference device. TMH 13 Part A, section A.4.3.3, separates component calibration from system validation.

3. Check the reference profile

   Confirm that a Class 1 device such as a precision rod and level profiled each wheelpath four times, as two closed loops, and that the reference IRI is the mean of the runs, computed for every 100 m (section C.3.1.2).

4. Check positioning first

   Confirm that the distance-measuring instrument and the global positioning system (GPS) were validated, because a distance error shifts every 100 m segment against its reference value (TMH 13 Part B, section B.3.3).

5. Match vehicle, operator, speeds and run counts

   Check that the vehicle, operator and speeds are those of the survey, then compare the sections and repeat runs delivered with the contract. Tables C.6 and C.8 give the manual's recommended numbers.

6. Plot before calculating

   Plot measured against reference values for the 100 m segments, with a line of equality for a profiler. Data lying consistently above or below that line indicate a systematic error (Appendix C-2).

7. Run the manual's checks

   For a response-type device, check linearity, the coefficient of determination, the standard error and repeatability on each section. For a profiler, check repeat-run variability, the regression, the error on each 100 m segment and agreement between days.

8. Judge against the contract's criteria

   Treat Tables C.7 and C.9 as guidelines that should not be enforced too rigidly. The network manager may relax them for conditions such as localised distress; record any relaxation and who agreed it.

9. Check the age of the validation

   Compare validation and survey dates. TMH 13 Part C gives no maximum age for a profiler validation (section C.2.2.3: minimum recalibration frequency for response-type devices); COTO Standard Specifications Chapter 20, clause A20.1.5.5, sets one unless the contract says otherwise.

10. Plan the control tests

    Re-measure control or validation sections periodically. After a failed test, discard and re-measure profiler data since the last successful test; for a response-type device, look for defects, then recalibrate if none is found. Never adjust a device with control data.

11. Assemble the file

    Put the device, section, reference, vehicle, run, certificate, decision and control-test records into one pack that a reviewer can follow from device to section to runs.

Frequently asked

What is the difference between calibrating and validating a roughness device?

In Part C of the South African road manual TMH 13 (also written TMH13), a response-type device senses suspension movement and does not measure the road profile, so a regression (a straight-line fit) against reference international roughness index (IRI) values converts its output to IRI (section C.3). A profiler has factory-calibrated components, so the field test, validation, confirms that the whole system is accurate and repeatable and does not adjust the output (section C.4).

How old can a roughness validation be before it is out of date?

The South African road manual TMH 13 Part C gives no maximum age for a profiler validation, and Part A says the specification should state how long a validation is valid. Clause A20.1.5.5 of the Committee of Transport Officials (COTO) Standard Specifications, Chapter 20 (Draft Standard, October 2020), sets a maximum age for the validation certificate of a high-speed profiler unless the contract documentation says otherwise. Read that clause and the contract for the figure that applies.

What does a contractor submit to show calibration of a response-type roughness device?

In the South African road manual TMH 13 Part C, section C.3.2 lists the calibration report: details of the calibration sections, device readings for each run against the reference international roughness index (IRI), an evaluation summary of repeatability, a regression summary, and the raw data in electronic and hard-copy form. The vehicle configuration, including driver, occupants and tyre details, is recorded with it.

Can a profiler be adjusted to match the reference section?

No. In the South African road manual TMH 13 Part C, section C.4 says validation of a profiler, a device that measures the road profile, is not meant to adjust the output to match a benchmark. A profiler is valid or it is not; if it is not, the manufacturer fixes it. Calibrating a profiler by comparing its processed international roughness index (IRI) with the reference should not be attempted under any circumstances.

Who selects and profiles a roughness validation section?

In the South African road manual TMH 13 Part A, section A.4.4 says validation sections should be identified and profiled before any survey starts, ideally by the network agency or by a contractor other than the one doing the survey. The sections should represent the network, avoid rapid deterioration and carry global positioning system (GPS) coordinates and clear markers so that the validation can be repeated.

Does a TMH 18 road-data file record the validation date of a roughness survey?

No. In TMH 18, the South African road asset data exchange manual, the combined instrument data record (section 5.6) holds road, lane, start and end kilometre, date and international roughness index (IRI) by wheelpath. Neither Draft Version 4 (October 2014) nor Committee Draft 5.0 (February 2016) has a column for device, class or validation date, so the evidence travels as a separate pack and the sender names the TMH 18 version the receiver expects.

Where this connects

The next manual is TMH 18 (Draft Version 4 or Committee Draft 5.0), which packages the checked IRI values for submission; follow that step in the guide on how a profiler run reaches the register. After it comes Draft TMH 22 Part D, whose section D.5.3 sets out how often condition data are collected. Survey-wide quality control is mapped in road condition survey quality control, and the TMH 13 explainer covers all seven parts. Last reviewed on 21 October 2026 against the editions named above.

References

1. COTO (2016a). TMH13: Automated Road Condition Assessments. Part C: Roughness. Committee Draft Final, May 2016. Original source.Sections C.2–C.5; Table C.5 is for calibration-section selection
2. COTO (2016c). TMH13: Automated Road Condition Assessments. Part B: Positioning. Committee Draft Final, May 2016. Original source.Positioning and distance measurement guidance
3. COTO (2016d). TMH 13: Automated Road Condition Assessments. Part A: General. Committee Draft Final, May 2016. Original source.Sections, tables and clauses are cited inline in each guide
4. COTO (2013). Draft TMH 22: Road Asset Management Manual. Draft, March 2013. Original source.Sections, tables and clauses are cited inline in each guide
5. COTO (2020a). Standard Specifications for Road and Bridge Works for South African Road Authorities. Draft Standard, October 2020. Original source.Sections, tables and clauses are cited inline in each guide
6. COTO (2014). TMH 18: Road Asset Data Electronic Exchange Formats. Draft Version 4, October 2014. Original source.Sections, tables and clauses are cited inline in each guide
7. COTO (2016i). TMH 18: Road Asset Data Electronic Exchange Formats. Committee Draft 5.0, February 2016. Original source.Sections, tables and clauses are cited inline in each guide

Use the manual edition and project requirements that apply to your work. Worked examples use synthetic values and are not site-specific design advice.