Reference · Living document

Glossary

The vocabulary of land administration and its GIS products. Terms here are used consistently in every lesson — if a lesson contradicts this file, the file wins and the lesson gets fixed.

Tags: INST institutional · LADM ISO 19152 class or concept · TECH software/data · DOCT donor doctrine

The core distinction

Cadastre INST

The parcel-based inventory of land: where the units of land are, what their boundaries and area are, and how they are identified. Primarily spatial information. Answers what and where.

Land registration INST

The recording of rights and interests in land, and of the transactions that change them. Primarily legal information. Answers who and on what basis. A cadastre without registration is a map of parcels nobody owns; registration without a cadastre is a list of rights over land nobody can locate.

Land administration INST

The umbrella: the processes of determining, recording and disseminating information about tenure, value and use of land. Cadastre and registration are two of its functions; valuation and land-use planning are others.

Land administration system (LAS) INST

The institutional + technical machinery that performs land administration. In this workspace, the thing being built.

Tenure INST

The relationship, legal or customary, between people and land. Deliberately broader than "ownership" — it includes leases, use rights, customary and communal claims. Donor projects care about tenure precisely because ownership is too narrow a frame.

Title system vs deeds system INST

Two registration philosophies. A deeds system registers the documents of transaction; proving a right means tracing the chain. A title system registers the right itself, with the state warranting it (Torrens is the classic example). Which one applies changes what your system must guarantee.

LADM — ISO 19152

LADM LADM

Land Administration Domain Model. The ISO conceptual model for land administration. Edition I was ISO 19152:2012, a single document; Edition II is multi-part — Part 1 (overview/fundamentals), Part 2 (land registration), Part 3 (marine), Part 4 (valuation), Part 5 (spatial plan). Its purpose is not to replace national systems but to give a formal language for describing them.

LA_Party LADM

A person or organisation that plays a role in a land transaction. Specialised by LA_GroupParty (a group holding jointly), linked via the association class LA_PartyMember. The who.

LA_RRR LADM

Abstract class with three concrete subclasses: LA_Right, LA_Restriction, LA_Responsibility. The what kind of claim. A mortgage is a restriction; a maintenance obligation is a responsibility; ownership is a right. Modelling all three as siblings is one of LADM's better ideas.

LA_BAUnit LADM

Basic Administrative Unit — the thing that RRRs actually apply to. Holds together one or more spatial units under a homogeneous set of rights. The hinge between the legal and spatial halves.

LA_SpatialUnit LADM

A spatial unit: parcel, building, network, or a point/text description. The where. Note that LADM permits a spatial unit to be a point or even a text description — which is what makes fit-for-purpose approaches representable.

Source (LA_AdministrativeSource / LA_SpatialSource) LADM

The evidence behind a record — a deed, a court order, a survey plan, a photo of a boundary. LADM insists evidence is first-class, not an attachment. Build this in from day one; retrofitting provenance is agony.

VersionedObject LADM

LADM's base abstract class: records carry beginLifespanVersion / endLifespanVersion rather than being overwritten, so the database can be reconstructed as it stood at any past moment. Land administration is inherently temporal — "who owned this in 2019" is a routine question, not an audit-log afterthought.

Bi-temporal LADM

Two independent clocks on every record: system time (when the database believed it) and real-world time (when it was actually true). LADM Edition II adds beginRealWorldLifespanVersion / endRealWorldLifespanVersion / acceptance alongside the system-time pair. A deed signed in March and lodged in September is true from March, known from September — and one timestamp cannot say that.

Donor doctrine

Fit-For-Purpose Land Administration (FFP) DOCT

The prevailing donor approach (Enemark, McLaren & Lemmen; GLTN / UN-Habitat / Kadaster, 2016): build the spatial, legal and institutional frameworks to the accuracy the purpose requires, achievable now and improvable later — rather than surveying everything to cadastral precision first. Fast, affordable, incremental, participatory.

General boundaries vs fixed boundaries DOCT

Fixed boundaries are surveyed to a defined precision and legally definitive. General boundaries are approximate — often traced off imagery — and the physical feature (the hedge, the wall) governs. FFP leans hard on general boundaries; this single choice changes your accuracy model, your capture cost, and your dispute workflow.

Systematic vs sporadic registration DOCT

Systematic: the state sweeps an area and records every parcel at once, usually with public display and objection periods. Sporadic: records are created one at a time, on demand, triggered by a transaction. Systematic-first is the donor default because coverage is the goal. It implies batch workflows, adjudication and objection states in your data model.

Adjudication INST

The process of authoritatively determining existing rights over a parcel — including hearing and resolving competing claims — before they are recorded. An adjudication record does not change rights; it establishes what they already are. In systematic registration this is the hard part, not the mapping.

Public display INST

The mandated period in a systematic campaign when claims and maps are posted publicly so anyone may object or correct. Typical windows: 15–30 days, 30–45 days, or up to six months. Systematic-only — sporadic registration has no equivalent stage, which is why campaign workflows cannot be retrofitted onto a transaction system.

Continuum of tenure DOCT

The FFP position that land rights run along a spectrum — from informal occupation through customary and group claims to registered freehold — and that a system should record what people actually hold rather than forcing everything into ownership. In schema terms: group parties and a wide right-type vocabulary from day one.

Products & technology

SOLA / Open Tenure TECH

FAO's open-source land administration suite. SOLA Registry (registration + cadastre), Systematic Registration, State Land; plus Open Tenure (mobile field capture) and Community Server. Stack: Payara (Java EE), PostgreSQL/PostGIS, GeoServer, JasperReports. Data structures based on LADM. The reference implementation to read before designing your own.

PostGIS TECH

Spatial extension for PostgreSQL: geometry/geography types, spatial indexes, topology, SQL analysis. In the open stack it is both the store and the processing engine.

GeoServer TECH

Open-source server that publishes spatial data as standard web services (WMS, WFS and successors). The "make the database visible to maps" tier.

QGIS TECH

Open-source desktop GIS. The analyst/editor client in the open stack. QGIS Server publishes map and feature services but ships no web-processing service — processing typically pushes down into PostGIS SQL or GDAL.

ArcGIS Enterprise TECH

Esri's self-hosted stack (Portal, Server, enterprise geodatabase, plus GeoAnalytics and Image Server tiers). Buys governed publishing, server-side geoprocessing services, and vendor support — at licence cost, which is a sustainability question in donor projects where funding ends.

WMS / WFS TECH

OGC web service standards. WMS returns rendered map images; WFS returns the features themselves, editable (writes need transactional WFS, "WFS-T"). If a requirement says "the map must be viewable" that is WMS; "the parcels must be downloadable/queryable" is WFS.

OGC API — Features TECH

The REST/JSON re-expression of WFS: resource-oriented, GeoJSON, OpenAPI 3.0. Parts: Core, CRS by Reference, Filtering. OGC encourages implementers to use it instead of WFS, and it is an INSPIRE Good Practice. Siblings modernise the rest of the family — OGC API — Maps for WMS, Tiles for WMTS.

CRS / EPSG code TECH

Coordinate Reference System — how coordinates map to the earth. Identified by EPSG codes (e.g. 4326 for WGS 84 lat/long). Cadastral data usually lives in a national projected CRS; getting this wrong shifts parcels by metres, silently.

Parcel TECH

The atomic unit of the cadastre: a contiguous area under homogeneous rights, with an identifier. Note the distinction from LA_BAUnit — one BAUnit may bind several parcels.

Datum TECH

An ellipsoid plus the way it is tied to the actual earth. The same physical point has different coordinates under different datums — which is why moving data between them needs an explicit geographic (datum) transformation, an empirical, region-specific, approximate calculation. Reprojecting within one datum is exact maths; crossing datums without a transformation produces a systematic shift that renders perfectly and is wrong.

Georeferencing TECH

Assigning real-world coordinates to something that had none — a scanned sheet, a drone frame — using ground control points and a transformation (1 point for a shift, 3 for affine, 6 for second order, 10 for third). Distinct from orthorectification, which additionally removes terrain and tilt displacement and resamples the pixels. A low RMS proves control points agree with the chosen transformation, not that the image is correctly placed.

Geodatabase TECH

Esri's data model plus behaviour over a store: domains, subtypes, relationship classes, topology, networks, parcel fabric. Three types — file (folder, single editor, no versioning), mobile (one SQLite .geodatabase file), enterprise (inside a DBMS, multiuser, supports versioning and archiving). Contrast a spatial extension like PostGIS, which adds spatial types and functions to a normal database and enforces rules for every client, not one stack.

Topology TECH

Explicit encoding of spatial relationships. A true topological store (PostGIS Topology) keeps nodes, edges and faces once, with TopoGeometry objects referencing them, so a shared boundary is one object. Geodatabase topology instead validates a rule set over ordinary simple features.

Topology rules TECH

Named constraints validated over feature classes: Must Not Overlap, Must Not Have Gaps, Must Be Covered By, Boundary Must Be Covered By, Contains Point, and others. For a cadastre, overlap is a competing claim (never auto-fix it); gaps under general boundaries are findings, not errors.

Cluster tolerance TECH

The minimum distance between vertices treated as coincident. During validation a gap larger than the tolerance is flagged as an error, while a gap smaller than it is auto-adjusted — cracked and clustered into coincidence. Validation therefore moves coordinates silently: keep the tolerance well below your accuracy class, decide it deliberately, and document it.

COGO TECH

Coordinate geometry — a boundary described as measurements rather than coordinates, as deeds actually define land (metes and bounds). A COGO-enabled line class carries Direction (north azimuth, decimal degrees, the chord for curves), Distance (straight lines only), Radius/Radius2 and Arc Length. The dimensions are the legal record; coordinates are a computed interpretation of them.

Closure error TECH

The gap left when a traverse's bearings and distances fail to return to the start point, usually a ratio (1:5,000). The cheapest quality check on transcribed deed dimensions — run it at ingest.

Parcel fabric TECH

Esri's connected parcel model: parcels share points and lines, lines carry COGO dimensions, and every feature is created or retired by a record (the legal document), making lineage traversable both ways. Coordinates are improved by least-squares adjustment from dimensions plus control — free network to evaluate internal consistency, weighted/constrained to improve absolute accuracy. Needs dimensions and control to be worth anything, so it fits deed-based and surveyed data rather than first-pass imagery capture.

Spatial network TECH

A graph of edges (travelled), junctions (connect edges) and turns (restricted movements), with costs. Esri splits it into the network dataset (transportation; connectivity discovered on demand) and the geometric / utility network (directed flow; connectivity maintained during edits). Connectivity comes from geometric coincidence plus connectivity groups and rules — crossing lines without a junction correctly do not connect. Open equivalent: pgRouting over an edge table.

LA_Level LADM

A set of spatial units sharing kind, dimension and registration type. Lets formal parcels, customary claims, informal settlements and utility networks coexist as parallel levels over the same ground. Most "our system doesn't fit LADM" claims dissolve once levels are understood.

LA_SpatialUnitGroup LADM

An aggregation of spatial units — for example all parcels in a municipality — nesting hierarchically to express district, block and adjudication area without a parallel invented tree.

LA_BoundaryFaceString LADM

A 2D boundary as a first-class object (LA_BoundaryFace is the 3D form, LA_Point the point form). Making the boundary an object is what lets evidence, accuracy and adjoiner agreement attach to the shared line instead of being duplicated inconsistently on both parcels.

Legal space LADM

A rights-bearing volume, defined by its documented boundaries rather than by physical fabric — LA_LegalSpaceBuildingUnit for apartments and units, LA_LegalSpaceUtilityNetwork for pipes and cables. A building or part of it is not equivalent to the physical separation: the wall may sit inside, outside or across the legal volume. Bounded by LA_BoundaryFace (3D), against LA_BoundaryFaceString for 2D units. 3D lives in the LADM core (Part 1), not in Parts 3–5.

Least-squares adjustment TECH

Computing best-fit coordinates from redundant observations by minimising the sum of weighted squared residuals, where weight = 1/σ². Redundancy = observations minus the minimum needed; it is what makes blunder detection possible at all. Free / minimally constrained tests whether measurements agree with each other; constrained ties them to control and tests absolute fit. Free clean plus constrained poor means the control, not the survey, is suspect.

Variance factor TECH

Also reference variance. Whether residuals came out the size your stated precisions predicted. ≈1 means honest error estimates; ≫1 means a blunder or over-claimed precision; ≪1 means you were pessimistic, which quietly inflates every error ellipse downstream.

Data snooping TECH

Baarda's method for finding blunders: normalise each residual by the uncertainty of that residual (not of the observation), test the largest, reject one, re-adjust, repeat. The tau criterion is the variant used when only the a-posteriori variance estimate is available. Rejecting several at once usually discards good data, because one blunder distorts every other residual.

Redundancy number TECH

Per observation, how much of its own error the network can absorb and reveal. Near zero means that observation is effectively unchecked and a blunder in it will pass into the coordinates undetected — a network design signal, best read before going to the field.

Error ellipse TECH

An adjusted point's uncertainty expressed as a shape rather than a single number, because uncertainty has direction: a point may be tight across a sight line and loose along it. Reporting one "accuracy" figure discards the part a surveyor needs.

DSM / DTM TECH

DSM — digital surface model, elevation including objects such as trees and buildings. DTM — digital terrain model, bare ground with objects removed. Orthorectify against the DTM for cadastral tracing: against a DSM, building tops land correctly while their bases shift.

GSD TECH

Ground sampling distance — how much ground one pixel covers. Resolution, not accuracy. Photogrammetric accuracy is the product of GSD, image overlap (typically 70–90% forward, 60–80% side) and ground control points; strong GSD with no GCPs yields an internally excellent model in the wrong place.

COG TECH

Cloud Optimized GeoTIFF — an ordinary GeoTIFF whose internal organisation (tiling plus overview pyramids) lets HTTP range requests fetch only the tiles a client needs. Removes bulk downloads, duplicate caches and the "which copy is current" problem, while remaining readable as a plain GeoTIFF by legacy software.

STAC TECH

SpatioTemporal Asset Catalog. An Item is a GeoJSON feature carrying metadata and links to Assets (the files, usually COGs); Catalogs list items; Collections add extent, licence and provider metadata. Turns "what imagery covers this parcel, from when, under what licence" into a query. Link each traced parcel to its STAC item so re-verification after a bad flight is a work list, not a crisis.

Mutation INST

A change to the cadastral fabric — subdivision, merger, boundary correction. Mutations are where naive data models break: the parcel is not a stable entity, it has a lifespan, and its successors must be traceable.