Watch the overview
55 seconds. A narrated explainer; the transcript is below. The written guide includes additional detail.
Read the video text
- 0:00A survey file can be perfectly
- 0:02formatted and still contain engineering inconsistencies.
- 0:06Match your survey to the network,
- 0:08classification and lane registers.
- 0:10Cross-check the references, too.
- 0:13Compare coordinates with your road shapefile.
- 0:15Identify positions away from the stated
- 0:18road or on another section.
- 0:21Check kilometre ranges, direction and position
- 0:24along the route.
- 0:26Flag conflicting condition ratings and assessment
- 0:29lengths that need review.
- 0:31Detect repeated instrument readings across different
- 0:33locations.
- 0:36Trace findings to source rows and
- 0:38references.
- 0:39See what was checked, what needs
- 0:41review and what could not be
- 0:42checked.
- 0:44Start your seven-day free trial, with
- 0:46up to one hundred thousand analysis
- 0:48rows per organisation, at TMH18.co.za.
A well-formed field can still be in the wrong context
A number can fit a column’s declared type while describing the wrong road, lane, date or section. Field validation establishes useful structure, but relationships supply a different kind of evidence. Keep the original survey export, chosen definition and project references available throughout review.
A project’s reference set can include geometry, network definitions, classification and lane data. These inputs are not interchangeable. Their availability, quality and active version determine which alignment comparisons can be made and how their findings should be interpreted.
Resolve identity before interpreting location
Begin with the submitted road identifier and the available reference index. Treat ambiguity as a limitation to resolve rather than silently selecting a plausible candidate. Once the identity is established, compare kilometre extents and lane context with the relevant source.
A section ending beyond its reference range needs investigation even if its start and end fields are numeric. A nearby road line is similarly evidence to consider, not permission to overwrite the source identifier or drag a point into place.
Read the specific engineering signal
Different survey families need different explanations. A combined-instrument pattern repeated across ranges, a deflection drop attached to the wrong context, or a conflicting pair of visual ratings each points to a different investigation. Do not treat every warning as the same generic data-quality score.
Read the rule’s applicability, example rows and stated basis. Operational tolerances and implementation baselines are not universal guarantees of measurement accuracy. Preserve missing observations honestly and keep field evidence with any justified correction.
Report what ran and what remains unresolved
Coverage describes the evidence available to a run. Missing geometry, absent fields or incomplete references can leave a comparison not checked. That is different from aligned or passed. Project totals also need their stated inclusion limits and reference context.
The reviewer should retain the current findings, reference version and decision notes, then follow the permitted approval workflow. Certification records a review decision and file fingerprint. It does not prove all engineering requirements, calibrate instruments or establish government endorsement.
Sources and scope
This guide describes the platform workflow; it does not establish a requirement of a TMH manual. Platform review and certification are described in the product documentation.