Skip to main content

Why GIS is different

GIS can reveal more than the information originally supplied. Location, geometry, imagery, overlays and derived analysis can turn apparently ordinary records into information about people, whenua, cultural sites, taonga species or relationships. That makes spatial governance different from ordinary file or database management.

Location can identify​

Removing a person's name does not necessarily make a spatial record anonymous. A point may fall on a single dwelling. A GPS track can reveal home, work or repeated visits. A photograph may contain coordinates or visual clues. A coarse cultural-site polygon may become identifiable when combined with cadastral parcels and aerial imagery.

The practical question is therefore not only does this dataset contain personal information? It is also what could somebody work out from the location, geometry, attributes, metadata and other datasets?

A map makes a claim​

A point, line or polygon is not the place itself. It is a representation chosen for a purpose.

That representation can still acquire authority. A polygon labelled rohe may be read by an external user as if it were a surveyed legal boundary. A line drawn to centimetre precision can look certain even when the underlying knowledge describes a broad, relational or contested area.

For Māori GIS, data quality therefore means more than coordinate accuracy. It can also include:

  • who supplied the information
  • who had authority to approve its use
  • the intended purpose
  • whether interests overlap
  • the level of uncertainty
  • whether the geometry is indicative or authoritative
  • what should accompany the geometry so it is not misunderstood

See Maps are not neutral.

GIS creates new information​

GIS is a derivation engine. Common operations can create information that did not exist as an explicit field in any source dataset.

Examples include:

  • a spatial join producing a list of properties near a cultural feature
  • a buffer revealing which proposed developments intersect a sensitive area
  • terrain and imagery helping narrow an approximate site location
  • LiDAR revealing archaeological forms beneath vegetation
  • remote sensing detecting environmental change without field access
  • machine learning identifying likely features from imagery or historical sources

The derived result may require its own authority and sensitivity assessment. Permission to use two source datasets does not automatically mean every possible output is appropriate to publish.

Several harmless layers can become one sensitive answer​

A site may not be directly mapped at all, yet a knowledgeable user may infer it from combinations such as:

This is why inference risk deserves its own assessment.

GIS makes copies easily​

A database can appear controlled while copies spread through:

  • desktop exports
  • GeoPackages and shapefiles
  • web feature services
  • mobile offline areas
  • screenshots
  • test and development databases
  • cloud backups
  • consultant laptops
  • email attachments
  • cached map tiles
  • archival exports

A useful sovereignty question is therefore where can this knowledge now exist?, not just who can log into the main database?

See Backups are data too.

Metadata can disclose what the map hides​

A restricted feature may be hidden while its item title, thumbnail, geographic extent, owner, tags or service URL remain visible. A dataset called Unrecorded urupā - upper valley can reveal a great deal before anybody sees a point.

See Metadata can give you away.

Precision is not the same as certainty​

GIS software encourages precise coordinates and crisp polygons. Māori knowledge may not always fit that model.

A place can be important through whakapapa, movement, seasonal use, story, relationship or overlapping interests. Converting all of that into a single polygon can make the database easier to query while making the knowledge less accurate in another sense.

A strong GIS may therefore preserve narrative, source, authority and uncertainty alongside geometry, or decide not to create geometry at all.

AI increases the inference problem​

Removing coordinates from a photograph or document does not necessarily remove geographic sensitivity. Computer vision and multimodal AI can infer places from terrain, vegetation, buildings, signs and text. Remote-sensing models can identify features that no community supplied as a GIS layer.

This creates a new distinction between:

  • data explicitly supplied
  • data derived by ordinary GIS analysis
  • data inferred by automated models

Each may require a different governance decision. See Artificial intelligence and GIS.

What to do​

Before publishing or sharing a spatial dataset, ask:

  1. What does the location itself reveal?
  2. Does the geometry imply more certainty than the evidence supports?
  3. What can be inferred when this is combined with common public layers?
  4. What derived products could users create?
  5. What metadata remains visible?
  6. Can users export, cache, screenshot or download it?
  7. Is a generalised publication derivative enough for the purpose?
  8. Should some information remain outside GIS entirely?

These questions should be answered before choosing the publishing platform.

Sources​

Last reviewed: 16 August 2026

Last updated on