28 Chapter 2
Calculation of surface characteristics such as slope is, however, more realistic with a
raster model than with a TIN model.
Attribute Data Models
In a vector model, each point, line or polygon is assigned the corresponding
attributes of geo-objects they represent. In a raster model, each pixel is assigned the
corresponding attributes of geo-objects it represents. A map layer in a vector or raster
model represents attribute data for a particular variable. For example, a map layer of
stream sediment sample points could represent labels of each point, whilst another map
layer of the same points could describe concentrations of a specific element at each
point. Parts of a study area without data for certain attribute are, in either vector or raster
model, null or undefined.
In a computer system, or more specifically in database terminology (see below),
attribute data represent observed and measured properties of geo-objects. Attribute data
have spatial, temporal or thematic characteristics. Spatial attributes pertain to properties
that vary in space and their variations can be characterised by location, topology and
geometry. Temporal attributes pertain either to age of geo-objects or to a period of data
acquisition. Thematic attributes are neither spatial nor temporal properties but pertain to
some forms of classifications to which geo-objects can be related, for example,
lithology, mineral deposit-type or mineral occurrence (i.e., presence or absence), faults
of certain orientations, etc. In many GIS studies related to mineral exploration, temporal
are thematic attributes are usually considered to be non-spatial.
Attribute data can be classified as either continuous or discrete variables. Continuous
variables take on any value (i.e., real values), whilst discrete variables take on only
certain values (i.e., real integers). Spatial attribute data are mostly continuous variables,
whilst non-spatial data are mostly discrete values. Element concentrations are examples
of a continuous variable, whereas stream order is a discrete variable. Because modeling
of geo-objects involves discretisation of continuous variables and quantisation (i.e.,
numerical representation for quantitative integration) of discrete variables, it is more
didactic to classify attribute data as quantitative or qualitative variables.
Attribute data representing numerical magnitude of geological, geochemical or
geophysical properties are quantitative variables. Data of quantitative variables are
usually measured on either ratio or interval scales and are mostly represented by
continuous values but can also take on discrete values. For example, element
concentrations and temperature are continuous variables measured on ratio scales and on
interval scales, respectively. In contrast, surface reflectance/absorption properties are
continuous variables measured on ratio scales but can be represented discretely as real
integers [0,255] in raster images. Quantitative variables are important forms of attribute
data because they can be manipulated by mathematical operations or transformations,
which are essential to spatial analysis.
Data of qualitative variables usually take on discrete values or labels according to
either ordinal or nominal scales of measurements. A percentile classification of element
concentrations is an example of ordinal measurement scale. Lithology is an example of