5.1 Spatial Topology: Planar Enforcement, Adjacency, Connectivity & DE-9IM

5.1 Spatial Topology: Planar Enforcement, Adjacency, Connectivity & DE-9IM

Core Principle: Topology transforms disconnected geometric coordinates into an intelligent spatial graph. While coordinate geometry defines where features are located in metric space, topology defines how spatial features relate to one another independently of coordinate systems. Understanding adjacency, connectivity, containment, planar enforcement, and formal intersection models like DE-9IM is essential for maintaining data integrity and executing advanced spatial queries.

---

1. Mathematical Graph Theory and Topological Foundations

In classical Euclidean geometry, entities are defined strictly by absolute metric coordinates—lengths, angles, areas, and Cartesian ($x, y, z$) positions. If a geometric figure is stretched, rotated, translated, or projected, its metric properties fluctuate. Topology, by contrast, is the branch of mathematics concerned with the geometric properties that remain invariant under continuous deformations—such as stretching, twisting, crumpling, and bending—without tearing, cutting, puncturing, or gluing surfaces together (often referred to in GIS as "rubber-sheeting").

Historically rooted in Leonhard Euler's 1736 resolution of the Seven Bridges of Königsberg, spatial topology in GIS models real-world geography as a planar graph composed of discrete topological primitives:

• 0-Simplex (Node / Vertex): A point entity representing discrete locations, intersections, or line endpoints.
• 1-Simplex (Edge / Arc / Link): A directed line segment bounded by an initial node and a terminal node.
• 2-Simplex (Face / Polygon): A bounded planar region enclosed by a contiguous loop of directed edges.

Euler's fundamental polyhedral formula establishes the invariant relationship among these primitives for any simply connected planar graph:

$V - E + F = 2$

Where $V$ represents the number of vertices (nodes), $E$ represents the number of edges (arcs), and $F$ represents the number of bounded faces (including the unbounded exterior face). In a bounded two-dimensional GIS plane where the infinite exterior face is excluded, the formula simplifies to:

$V - E + F = 1$

In GIS, topological data structures store these relationships explicitly in relational lookup tables rather than computing them on-the-fly from floating-point coordinate geometries. This architectural choice dramatically accelerates complex spatial operations such as contiguity searches, upstream/downstream network tracing, polygon overlay, and boundary verification.

---

2. The Three Core Topological Relationships

Every topological vector GIS architecture evaluates and manages three primary mathematical relationships among geometric primitives:

   [Adjacency / Contiguity]        [Connectivity]           [Containment / Area Definition]
   +-----------+-----------+       (N1)---[Arc 1]---(N2)            +------------------+
   |           |           |                         |              | Face A           |
   |  Face A   |  Face B   |                      [Arc 2]           |   +----------+   |
   |           |           |                         |              |   | Island B |   |
   +-----------+-----------+                        (N3)            |   +----------+   |
     Shared Boundary Arc                Arc-Node Linkage            +------------------+
     (Left Face / Right Face)          (Shared Junction Node)       Closed Edge Loops & Holes

1. Adjacency and Contiguity (Polygon-Arc Topology)

Adjacency defines which spatial regions share a common border. In a topological vector structure, an edge is not stored twice for two bordering parcels; instead, the edge is recorded once in an arc table alongside explicit topological pointers indicating its Left Face and Right Face based on the edge's digitization direction (from-node to to-node). Contiguity modeling enables instant identification of neighboring properties, demographic boundary analysis, and parcel fabric management without geometric buffering or intersection recalculation.

2. Connectivity (Arc-Node Topology)

Connectivity defines how linear paths join at common junction points to form continuous networks. Every edge maintains explicit attributes recording its From-Node (originating junction) and To-Node (terminating junction). In utility distribution systems (water, sewer, gas, electric) and transportation networks (streets, rail, flight paths), connectivity ensures that flow routing, shortest path algorithms (e.g., Dijkstra's algorithm), and network tracing execute through valid physical links rather than visually coincident but structurally disconnected lines.

3. Containment and Area Definition (Polygon-Edge Topology)

Containment defines which geographic entities reside within the interior of other features, as well as how closed linear boundaries enclose two-dimensional areas. Polygons are constructed not as standalone lists of coordinate loops, but as ordered sequences of directed edges that completely encapsulate a surface. Containment topology also formally resolves islands (inner rings or donut holes) and non-contiguous components (multipart polygons) by managing nested loop pointers and exterior/interior boundary hierarchies.

---

3. Vector Data Models: Spaghetti vs. Topological Models

The history of GIS data architecture is defined by the tension between lightweight graphical convenience and rigorous topological data integrity.

The Spaghetti Model

Introduced in early CAD and mapping packages, the spaghetti model treats geographic features as visual drafting entities. Each point, line, or polygon exists in absolute spatial isolation. If two adjacent cadastral parcels share a 500-meter boundary, that boundary is recorded independently in Parcel A's geometry and Parcel B's geometry. Inevitably, manual digitizing, coordinate precision truncation, or projection transformations cause the two lines to diverge slightly, yielding microscopic gaps (unclaimed spatial voids) or overlapping sliver polygons. The system has no innate knowledge that the two parcels touch.

The Topological Model

In a topological data structure, coordinates are tied directly to primitives. The boundary between Parcel A and Parcel B exists as Edge 101. The topology table explicitly declares: Edge 101: From Node 5, To Node 8, Left Face = Parcel A, Right Face = Parcel B. If a technician updates the path of Edge 101, both parcels update simultaneously and perfectly in unison, guaranteeing zero boundary divergence, zero slivers, and zero unassigned gaps.

---

4. Planar Enforcement and Geometric Exclusivity

Planar enforcement is the foundational spatial axiom governing two-dimensional topological GIS models. It enforces two strict geometric criteria:

1. Intersection Invariance: Every point of intersection between two or more 1-dimensional edges must occur at an explicit, shared 0-dimensional node. No two lines may cross without creating a junction node.
2. Exhaustive Planar Partition: In a polygon coverage or continuous parcel fabric, space must be completely and non-overlappingly partitioned. Every coordinate $(x, y)$ in the plane belongs to exactly one polygon face, and no two polygons in the same planar domain may overlap.

      VIOLATION OF PLANAR ENFORCEMENT                  PLANAR ENFORCEMENT APPLIED
   (Edges cross without junction node)             (Intersection resolved at Node N5)
   
        Line 1              Line 2                      Line 1              Line 2
          (N1)              (N3)                          (N1)              (N3)
            \                /                              \                /
             \              /                                \              /
              \   Cross    /                                  \   Node N5  /
               \          /                                    \    (o)   /
                \        /                                      \  /   \ /
                 \      /                                        \/     X
                  \    /                                         /\    / \
                   \  /                                         /  \  /   \
                    \/                                         /    (o)    \
                    /\                                        /   Node N5   \
                   /  \                                      /               \
                 (N4)  (N2)                                (N4)              (N2)

Exceptions to Planar Enforcement

While planar enforcement is essential for cadastral ownership, soil classification, electoral districts, and administrative zoning, real-world geography often contains non-planar entities:

• Transportation Overpasses and Underpasses: A highway bridge passing over a local roadway does not allow vehicles to turn directly onto the road below. Forcing a planar junction node would create an artificial traffic intersection. Non-planar network models handle this via elevation fields ($Z$-levels or $F_ZLEV / T_ZLEV$ attributes) indicating grade separation.
• Subsurface Utility Crossings: High-pressure water mains passing beneath telecommunication conduit banks without physical junction.
• Geological Overthrusts: Stratigraphic folding where ancient bedrock physically overlaps younger sediment formations.

---

5. Common Topological Errors and Cleaning Protocols

Digitizing analog paper maps, extracting features from raster imagery, or converting uncleaned CAD files into GIS vector formats inevitably introduces topological anomalies. Understanding these errors and their mathematical remediations is a prominent component of the GISP examination.

   COMMON TOPOLOGICAL DIGITIZING ERRORS
   
   1. Undershoot                  2. Overshoot                   3. Sliver & Gap
      Line A      Target Line        Line A      Target Line        +--------+  Gap  +--------+
      --------->   |                 --------+--> |                 | Parcel | (Void)| Parcel |
             (Gap) |                         | (Dangle)             |   A    |       |   B    |
                   |                         |                      +--------+       +--------+
                                                                        +----------+
                                                                        | Sliver   | (Double
                                                                        +----------+  Count)

Snap Tolerance and Cluster Tolerance Mechanics

To automate the detection and remediation of digitizing errors, GIS engines implement rigorous spatial tolerances:

• Snap Tolerance: The minimum distance threshold within which a moving cursor, vertex, or edge will automatically snap to an existing vertex, edge, or node during editing operations.
• Cluster Tolerance: The distance threshold within which all vertices and edges are considered coincident. During topological validation, vertices falling within the cluster tolerance are geometrically snapped together to a single mean location, and new nodes are inserted at all intersecting edges.

[!CAUTION] Exam Trap Alert: Excessive Cluster Tolerance Disasters. Setting a cluster tolerance too high collapses legitimate, closely spaced features. For example, if a parcel boundary has a narrow 0.5-meter setback and the analyst runs topology validation with a 1.0-meter cluster tolerance, the boundary vertices will snap together, permanently distorting property boundaries, shrinking acreage, and corrupting cadastral records. Always calibrate cluster tolerances well below the data's minimum mapping unit and positional uncertainty.

6. Topology Rules Across Feature Classes

Modern enterprise geodatabases enforce spatial business logic through configurable topology rules. Rather than relying on rigid, monolithic coverage architectures, modern systems allow administrators to establish relational constraints across one or multiple feature classes within a common feature dataset.

Standard Topology Rule Matrix

---

7. The Dimensionally Extended 9-Intersection Model (DE-9IM)

In spatial database theory and the Open Geospatial Consortium (OGC) Simple Features specification (ISO 19125-1), topological relations between any two arbitrary geometries ($A$ and $B$) are formally evaluated using the Dimensionally Extended 9-Intersection Model (DE-9IM), formulated by Christian Clementini and Eliseo Clementini.

Geometric Decomposition: Interior, Boundary, and Exterior

Every geometric entity (point, linestring, polygon) partitions 2D space into three pairwise disjoint sets of points:

1. Interior ($I$): All points belonging to the geometry excluding its boundary.
   • Point: The point itself.
   • Line: The points along the line between its endpoints (excluding the endpoints).
   • Polygon: The continuous area enclosed within the perimeter (excluding the perimeter itself).
2. Boundary ($B$ or $\partial$): The lower-dimensional set separating the interior from the exterior.
   • Point: The empty set ($\emptyset$, dimension $-1$).
   • Line: The two terminal endpoints (From-Node and To-Node; dimension $0$). If closed (ring), boundary is empty.
   • Polygon: The bounding linear rings enclosing the area and any interior holes (dimension $1$).
3. Exterior ($E$): All points in the coordinate space not belonging to the interior or boundary (dimension $2$).

The 3×3 Intersection Matrix

The DE-9IM compares each of the three components of geometry $A$ against the three components of geometry $B$, calculating the topological dimension of their pairwise intersections:

$\mathbf{DE\text{-}9IM}(A, B) = \begin{bmatrix} \dim(I(A) \cap I(B)) & \dim(I(A) \cap B(B)) & \dim(I(A) \cap E(B)) \\ \dim(B(A) \cap I(B)) & \dim(B(A) \cap B(B)) & \dim(B(A) \cap E(B)) \\ \dim(E(A) \cap I(B)) & \dim(E(A) \cap B(B)) & \dim(E(A) \cap E(B)) \end{bmatrix}$

Matrix Dimension Values and Wildcard Masks

The possible mathematical values inside each cell of the matrix are:

• $-1$ (or $F$): The intersection is empty ($\emptyset$). They do not intersect at all.
• $0$: The intersection contains at least one point, but no higher dimension.
• $1$: The intersection contains at least one 1-dimensional line segment.
• $2$: The intersection contains a 2-dimensional planar surface.
• $T$ (True): Any non-empty intersection, meaning dimension $\ge 0$ (i.e., ${0, 1, 2}$).
• $*$ (Wildcard): Any value is acceptable ("don't care").

                    DE-9IM 3x3 MATRIX EVALUATION SCHEMA
                    
                                Geometry B
                        Interior      Boundary      Exterior
                      +-------------+-------------+-------------+
           Interior   | I(A) ∩ I(B) | I(A) ∩ B(B) | I(A) ∩ E(B) |
                      +-------------+-------------+-------------+
Geometry A Boundary   | B(A) ∩ I(B) | B(A) ∩ B(B) | B(A) ∩ E(B) |
                      +-------------+-------------+-------------+
           Exterior   | E(A) ∩ I(B) | E(A) ∩ B(B) | E(A) ∩ E(B) |
                      +-------------+-------------+-------------+

Standard OGC Spatial Predicates Defined by DE-9IM

SQL spatial queries (such as PostGIS functions ST_Contains, ST_Within, ST_Touches) implement DE-9IM pattern matrices to evaluate geographic conditions:

Exam Trap Alert: The Touch Predicate. For two polygons to satisfy ST_Touches, their interiors must not intersect ($I(A) \cap I(B) = -1$). They may only touch along their boundaries. If two parcels overlap by even a microscopic fraction of a millimeter, their interiors intersect ($I(A) \cap I(B) = 2$), which instantly invalidates ST_Touches and converts the relationship into ST_Overlaps or ST_Intersects.