Course · ← 0013 Parcel Fabric · 0015 LADM In Depth → · engine layer

Lesson 0014 · ~13 minutes

Spatial Networks

Topology asks whether things touch. A network asks whether you can get from one to the other, and at what cost. Land systems need this the moment access, utilities or rights-of-way enter the picture — which is sooner than most designs assume.

Networks are topology with direction and cost

A network is built from three element types. Edges connect to other elements and are the links over which agents travel; junctions connect edges and let navigation pass from one edge to another; turns are optional elements storing a particular turning movement — for instance a restricted left turn from one edge to another.¹

Add attributes — length, travel time, capacity, one-way restriction — and the graph becomes answerable: shortest route, service area, nearest facility, what is upstream of this valve.

The two Esri models

The geodatabase has two core network models: the network dataset, which models transportation networks, and the geometric network, which models directed-flow systems such as river networks and utility lines — the latter now superseded by the utility network.²

The connectivity difference is the one that matters when you choose. A geometric network is built from simple line and point features that become junction, simple-edge and complex-edge features, and network connectivity is then continuously updated during edits; a network dataset instead has network sources from which a connectivity model is discovered on demand.²

How connectivity is actually decided

Not by wishful thinking. In a network dataset, connectivity is based on geometric coincidence of line endpoints, line vertices and points, and on the connectivity rules set as properties of the dataset.³ It begins with connectivity groups: each edge source is assigned to one group, and each junction source may belong to one or more.³

That grouping is how a road network and a footpath network can cross without connecting, while a car park entrance node joins both. Overpasses are the classic case: two lines crossing on screen, no junction, no connection — correctly.

Utility networks take the same idea further with a rule set determining which feature types may connect to which⁴ — a transformer connects to a line, not to another transformer.

In the open stack

pgRouting extends PostGIS with routing over a plain edge table: build source and target node columns, add a cost column, then call Dijkstra, A*, driving distance and similar. There is no separate network object — the graph is your table plus its topology columns, and the algorithms are SQL functions.

-- edges: id, geom, source, target, cost, reverse_cost
SELECT * FROM pgr_dijkstra(
  'SELECT id, source, target, cost, reverse_cost FROM road_edges',
  start_node, end_node, directed := true);

Cheaper and more transparent than a proprietary network model; correspondingly more of the modelling — turn restrictions, connectivity groups, flow direction — is yours to encode rather than configure.

Why a land project meets this

Networks look like someone else's problem until the requirements arrive:

Access and landlocking. Does every parcel have legal access to a public road? Subdivision approval frequently turns on this, and it is a reachability query, not a geometry query.

Rights of way and easements. An easement is a linear restriction — an LA_Restriction (lesson 0001) whose spatial unit is a corridor across someone else's parcel. Utility corridors, irrigation channels and access tracks all take this shape.

Utility service and valuation. Distance along the network to water, power or road frontage drives land value, and valuation is a full LADM package in Edition II.

Field logistics. Routing the systematic-registration teams of lesson 0003 across blocks is a vehicle-routing problem, and at 40,000 parcels it is worth solving properly.

Drill 1 — Which model?

Drill 2 — Recall

Your primary source

Esri, Understand network elements — edges, junctions and turns, defined precisely and briefly:
pro.arcgis.com — Understand network elements

Then Connectivity for connectivity groups — the concept that most often explains why a network "does not work". For the open equivalent, pgRouting's own documentation is the reference.