Geography Reference
In-Depth Information
Additional metrics were examined using alternative software for terrain data
handling. For example, a
index (PRD) was calculated using
the Fragstats program, resulting in a metric for a landscape (or any categorised
polygonal dataset) which is higher the more individual patches of a class which
exist in the image (McGarigal and Cushman
2005
). The PRD metric presents the
number of distinct patches per 100 ha. Thus, a dissected landscape, classed for
example into 32 categories of height (i.e. a layer-tinted terrain model of a complex
area) will have smaller individual and more numerous adjacent hypsometric layer
tint zones, and a higher density of separate individual patches, compared to a
uniform sloped terrain which will have only as many patches as there are classes.
In this terrain, for example, Zone A has a PRD of 307.07, whilst Zone B, with its
more complex landscape has a PRD of 340.23.
Analysis of the terrain was also undertaken using the Landserf terrain data
handling software package. The fractal dimension of the surface in each zone was
calculated (Zone A, 2.12; Zone B, 2.20) confirming the higher disorder in Zone
B. Feature extraction and landscape feature detection is effectively undertaken in
Landserf, with elements such as pits, ridges, channels, passes, flat surfaces etc.
being identified and visualised. Graphical output from this routine indicates that
Zone A has a lower density of structural features (most of the detected ridges are, in
fact, walls and field boundaries rather than mining artefacts), whilst a greater
proportion of the pixels in Zone B can be categorised as forming channels and
ridges.
In Zone A, the Ridges and Channels form only 3.9 % of the DSM cells, whilst in
Zone B they constitute 16.6 %. The planar areas form 96 % of Zone A, but 81 % of
Zone B, which has many more peaks identifiable.
The work undertaken so far has demonstrated that terrain surfaces can be
captured effectively at appropriate scale and resolution for investigating their
structure. Disorder in the terrain can also be quantified, either absolutely (for
specific measures to be stored) or comparatively (to detect areas of relative disor-
der). Furthermore, background information about the nature of the terrain, its
formation and its modification, can be used to confirm the disorder inherent in
differing landscapes and land uses.
patch richness density
'
'
Representation
Once terrain disorder has been identified and quantified, the task of mapping it to
reflect the variability and complexity must be faced. It was suggested in the
Introduction that new methods of cartographic representation must be sought and
established for the most efficient mapping of disorder. Historical and contemporary
maps of the area studied in this paper are illustrated and considered here.
The early topographic mapping shown in Fig.
2a
, with data (including contour
values) collected solely by field observation, demonstrates the use of
two-dimensional mimetic symbols to represent breaks of slope, patches of spoil,
