GIS and Survey Data Management
This domain tests data stewardship: whether you can move survey observations into a spatial database without losing the things that make them usable - datum, epoch, accuracy, lineage and version. Exam questions often present a data-merging failure and ask you to diagnose it.
Coordinate reference systems
Every layer must carry a defined coordinate reference system. In Hong Kong, HK80 Grid is commonly identified as EPSG:2326, while GNSS and web services frequently use WGS84 or an ITRF-aligned frame. Two failure modes dominate: treating grid coordinates as geographic coordinates, which displaces features by kilometres, and re-projecting with a different datum transformation from the one used for the original survey, which shifts features by metres or tens of metres. Transformation parameters must be documented, not chosen at the keyboard.
Data quality and metadata
Assess data against the recognised dimensions: positional and attribute accuracy, completeness, logical consistency, temporal currency and lineage. Metadata should follow a recognised standard structure and record purpose, owner, capture method, datum, accuracy statement, date and access constraints. Survey data that may be relied on evidentially needs a chain of custody: raw observations, reduction sheets, adjustment output, final plans and the software versions used.
Practical controls
- Standard feature codes and naming conventions agreed before fieldwork, not after.
- Validation at capture: closed traverses, redundant observations, automatic range and code checks.
- Version control with a single authoritative dataset and an audit trail of edits.
- Backup, access control and retention schedules.
- Interoperability: DWG, DXF, LandXML, shapefile, GeoPackage and CityGML each lose different information, so the exchange format must suit the receiving system.
Worked example
An as-built DWG survey on HK80 Grid is loaded into a GIS that assumes WGS84 geographic coordinates. The features land off West Africa. The diagnosis is a missing CRS definition, not a bad survey. The fix: define the layer as HK80 Grid, transform with the documented parameters to the target frame, then compare twenty known control points before and after to confirm residuals. Any residual above the stated accuracy means the transformation parameters are wrong.
Traps that cost marks
Assuming that anything in a GIS is accurate. Publishing layers with no metadata or CRS. Losing raw observations after producing a drawing. Overwriting the only copy of a survey dataset with an edited version. Ignoring personal data obligations when the dataset includes identifiable individuals.