You are here

Finding the best combinations of terrain attributes and GIS software for meaningful terrain analysis

TitleFinding the best combinations of terrain attributes and GIS software for meaningful terrain analysis
Publication TypeBook Chapter
Year of Publication2015
AuthorsLecours, V, Simms, A, Devillers, R, Edinger, E, Lucieer, V
EditorJ., J, Zb., Z, H., M, T., H
Book TitleGeomorphometry for Geosciences
PublisherInstitute of Geoecology and Geoinformation, International Society for Geomorphometry
CityAdam Mickiewicz University in Poznań
Keywordsbenthic habitat, BTM, Geomorphic seabed classes, Geomorphological mapping, geomorphometry, GIS and oceanography, seafloor geomorphology

Tools that derive terrain attributes from digital
elevation models are common in geospatial software. Their
accessibility permits applying geomorphometric techniques to a
wide range of applications. These tools however, can be
considered “black boxes” where the analysis and comparison of
the internal workings of the technique are vague and cannot be
assessed. Selecting the most effective set of tools for a given task
can thus be challenging. This work presents a method for
selecting an optimal set of terrain attributes that can help nonexpert
GIS users make the best use of geomorphometry. The
selection of terrain attributes aims to remove redundancy
between attributes and maximize the amount of information
given on a surface. We derived 230 terrain attributes from an
artificial surface using 11 software. This approach is twofold: a
pre-selection based on the ranking of attributes was first
established using stepwise multicollinearity measures, followed
by a final selection of attributes from a principal components
analysis (PCA). The results show that using 13 independent
terrain attributes can explain up to 83% of the variance for that
particular surface: the combination of common attributes that
are available in most GIS (i.e. aspect, basic curvatures, slope and
a measure of rugosity) can explain 67% of the surface variance.
The method proved efficient to reduce a high-dimensional list of
terrain attributes to identify combinations of 13 attributes or less
that can be used by non-expert GIS users.