Depthmap Measures

Depthmap v3.03 Measures

AT March 2003

This is a list of the possible measures in Depthmap, with a short explanation of what they mean spatially.

They tend to be labelled using terms from mathematical graph theory, rather than space syntax, so it is

helpful to know a couple of graph theory terms before we start.

A graph comprises vertices (what we would commonly call nodes) and edges (which we would call

connections). In axial analysis, the vertices are the axial lines, and the edges are formed where they cross.

The vertices are coloured according to a scale, and the edges are not shown. The same is true for

Depthmap analysis. The vertices are points, typically of possible location, and the edges are the lines of

sight that connect them. As with axial analysis, the vertices are coloured, and the edges are not shown.

Edge

Vertices

Edge

Vertices

The Measures

The measures themselves are split into four groups: immediate (which are calculated when you make the

graph), local (which are calculated from vertices connected to one another), global (calculated over the

whole graph), and metric (calculated from physical distances, rather than visibility relationships).

Immediate measures

Neighbourhood size

(connectivity). Neighbourhood size is the number of vertices directly visible from a

point, and so approximates the visible isovist area. The set of vertices that are directly visible themselves

are called the

neighbourhood

.

Far distance

. The physical distance to the furthest visible vertex from a point.

Total distance

. The sum of the physical distances to all the visible vertices from a point (i.e., all the

vertices in the neighbourhood of the point)

Average distance

. The mean of the physical distances to all the visible vertices from a point (i.e., all the

vertices in the neighbourhood of the point).

Local measures

Clustering coefficient

. Has no analogue in space syntax. The clustering coefficient approximates the

convexity of the isovist from the current location. A convex isovist has a clustering coefficient of 1, while

a ?spikey? isovists tends to a clustering coefficient of 0. Technically, the clustering coefficient is the

proportion of connections that actually exist within the neighbourhood to the number of connections that

could possibly exist if all the vertices were intervisible.