1 Spatial Modelling of the Terrestrial Environment - Georeferencial
1 Spatial Modelling of the Terrestrial Environment - Georeferencial
1 Spatial Modelling of the Terrestrial Environment - Georeferencial
Create successful ePaper yourself
Turn your PDF publications into a flip-book with our unique Google optimized e-Paper software.
116 <strong>Spatial</strong> <strong>Modelling</strong> <strong>of</strong> <strong>the</strong> <strong>Terrestrial</strong> <strong>Environment</strong><br />
assign surveyed elevations to <strong>the</strong> nearest DEM elevation, provided <strong>the</strong> point was within<br />
a given search radius. The larger <strong>the</strong> search radius used, <strong>the</strong> more check data points are<br />
assigned DEM elevations. However, this reduces confidence in <strong>the</strong> resulting height discrepancies,<br />
as <strong>the</strong> average lateral distance between DEM and survey point is increased. In<br />
this study, <strong>the</strong> search radius was set to 0.5 m (to give a search diameter <strong>of</strong> 1 m), <strong>the</strong>reby<br />
matching <strong>the</strong> DEM grid spacing.<br />
6.3 The Meaning <strong>of</strong> Error and <strong>the</strong> Treatment <strong>of</strong> Error in Digital<br />
Elevation Models<br />
6.3.1 The Definition and Quantification <strong>of</strong> Error<br />
Perhaps one <strong>of</strong> <strong>the</strong> most important aspects <strong>of</strong> error is a widespread variation in how it is<br />
defined and managed in <strong>the</strong> measurement <strong>of</strong> surface topography. By far <strong>the</strong> most common<br />
reference to surface quality is in terms <strong>of</strong> its accuracy (e.g. Davison, 1994; Neill, 1994;<br />
Gooch et al., 1999). However, in engineering surveying (e.g. Cooper, 1987; Cooper and<br />
Cross, 1988), it is very common to consider error more generically, in terms <strong>of</strong> <strong>the</strong> quality <strong>of</strong><br />
topographic data, with accuracy describing one subset <strong>of</strong> data quality. There are <strong>the</strong>n three<br />
types <strong>of</strong> error: systematic error, blunders and random errors (e.g. Cooper and Cross, 1988;<br />
Lane et al., 1994), and <strong>the</strong>se are thought to control data accuracy, reliability and precision,<br />
respectively. Systematic errors occur when a measurement is used with an incorrect functional<br />
model (Cooper and Cross, 1988). Cooper and Cross give <strong>the</strong> example <strong>of</strong> how <strong>the</strong> use<br />
<strong>of</strong> <strong>the</strong> basic collinearity equations derived from <strong>the</strong> special case <strong>of</strong> a perspective projection<br />
(e.g. Ghosh, 1988) will result in systematic error as <strong>the</strong> functional model (<strong>the</strong> collinearity<br />
equations) does not include those effects that cause deviation from <strong>the</strong> perspective projection<br />
(e.g. sensor distortions). The functional model provides an inaccurate description<br />
and leads to systematic errors. Cooper and Cross (1988) note that <strong>the</strong>re did not appear to<br />
be a widely accepted term to describe <strong>the</strong> quality <strong>of</strong> a dataset with respect to systematic<br />
errors and recommend <strong>the</strong> term accuracy. Blunders or gross errors arise from an incorrect<br />
measuring or recording procedure. Cooper and Cross noted that when measurements were<br />
made manually, it was relatively easy to identify and rectify gross errors through independent<br />
checks on measured data. However, with automation <strong>of</strong> <strong>the</strong> measurement process,<br />
this is more difficult, but still needs to be undertaken. The quality <strong>of</strong> a dataset in terms <strong>of</strong><br />
blunders is defined in terms <strong>of</strong> its reliability (Cooper and Cross, 1988). Third, random<br />
errors relate to inconsistencies that are inherent to <strong>the</strong> measurement process, and cannot be<br />
refined by ei<strong>the</strong>r development <strong>of</strong> <strong>the</strong> functional model or through <strong>the</strong> detection <strong>of</strong> blunders.<br />
The quality <strong>of</strong> a dataset in terms <strong>of</strong> random errors is defined as its precision.<br />
Whilst <strong>the</strong>se definitions largely relate to <strong>the</strong> engineering surveying approach to data<br />
quality, <strong>the</strong>re is some difference in terms <strong>of</strong> <strong>the</strong> definitions used in statistical analysis.<br />
Everitt (1998) defines: (1) accuracy as <strong>the</strong> degree <strong>of</strong> conformity to some recognized standard<br />
value; (2) precision as <strong>the</strong> likely spread <strong>of</strong> estimates <strong>of</strong> a parameter in a statistical model;<br />
and introduces (3), bias, as <strong>the</strong> deviation <strong>of</strong> results or inferences from <strong>the</strong> truth. It is clear<br />
from <strong>the</strong>se definitions that <strong>the</strong>re is some confusion over <strong>the</strong> meaning <strong>of</strong> accuracy and <strong>of</strong><br />
bias: <strong>the</strong>y appear to be <strong>the</strong> same thing according to Everitt (1998). This implies that <strong>the</strong><br />
distinction between accuracy and bias is a subtle one. Any observation may differ from its<br />
correct value, appearing to be inaccurate, but one observation does not allow us to decide