Geography Reference
In-Depth Information
Data Processing
The DSM was examined using a range of software, in order to prepare and modify
the data, and analyse its properties. A primary parameter which can be obtained
from the surface model is a slope map, which can visually be used to detect zones of
differing levels of dissection. The slope map for the relevant zones is shown in
Fig.
1b
. As can be seen, the zone of sandstone (Zone A), which has been subject to
less mining activity and spoil heap creation, has fewer steep areas and a more
uniform surface. The mean slope in Zone A is 10.11
, whilst in Zone B it is 14.69
.
The comparative values of average slope may well be the simplest and most
suitable metric to establish variable disorder in this landscape.
However, further confirmation was sought of the distinction between the two
zones, by examining several more landscape indices. The Terrain Ruggedness
Index (TRI) was initially established for large-area landscape characterisation to
assist in wildlife management (Reilly et al.
1999
). It was calculated and mapped in
this exercise using the Raster Calculator within ArcGIS, the output demonstrating
the differences which result from measuring the height differences between adja-
cent pixels in the DSM. By extension, this is also a function of slope but quantifies,
more directly, dissection of the terrain surface and the degree of difference in height
of all eight neighbouring cells to the target pixel.
The mean value of the TRI for Zone A was measured at 19.31 whilst Zone B had
a higher mean TRI value of 22.55. There are, in fact, many different indices
available for characterising surface roughness (see, for example,
http://
gis4geomorphology.com/roughness-topographic-position/
). A further index
applied to this dataset is sourced from terrain analysis work presented by Hobson
in 1972, and coded as a Python script for incorporation into ArcGIS by Sappington
(
2008
). This index is more comprehensive than the TRI metric, in that it takes
account of aspect in addition to slope—clearly, consideration of variable orienta-
tion of equal slope values around a point, for example, should yield improved and
more faithful measures of dissection. The resultant index (called vector ruggedness
measure, VRM, by the script author) was assessed for Zones A and B: once again,
an overall mean figure for VRM shows variability, with Zone A calculated at
0.0013 and Zone B at 0.0044.
The measurements taken so far indicate that it is possible to develop realistic
measures of terrain variability from LiDAR-derived digital surface models, at
sufficiently large scale. The scale must be set to consider the impact, in this small
area, of relatively minor features—small spoil heaps, depressions indicating capped
pit-shafts, and surface features such as tracks and specially-dug drainage channels.
The figures show that a comparison can be made between nearby zones with
differing landscape use histories, and it may be possible to develop models of
landscape form and genesis which can be transferable across regional and national
landscape characterisation studies. In this case study, a distinction has been drawn,
using simple indices of disorder, between an area relatively untouched by human
activity, and one which has been comprehensively altered by anthropogenic mining
practices.
