3.6.1Geographical information accuracy and uncertainty
The geographical information quality requirements for each entity are
described with positional accuracy, attribute information exactness,
completeness, and data origin. Geographical information quality elements
and accuracy assessments are discussed in the following paragraphs.
Lineage [16], or data history, relates to source data acquisition and
processing methods as well to conducted coordinate transformations. The
age of source material is important for all geographical information, but in
case of topographic information, the content does not change rapidly. How
old data can be used as source material depends on the particular case.
Source data should be as new as possible. In some cases, 5 years is
considered maximum. Old but otherwise accurate geographical information
can be very valuable if used in the map revision process. Of course,
topographic information is valuable even if older than 5 years, since it does
not change so much.
Positional accuracy is very important for all geographical information. It
defines spatial location exactness in all directions. Positional accuracy is
related to source data accuracy and map production methods [17]. In telecom
applications, the need for horizontal positional (i.e., planimetric) accuracy is
less than in conventional surveying applications [2]. Several steps in the map
production process include positional error sources. Positional location
errors in the final digital map product may cause major problems for
network quality [5]. Planimetric errors may cause inappropriate base station
locations, and errors in height information may cause overly pessimistic or
optimistic coverage predictions. Positional accuracy should be inspected
using known point locations or available field measurement data. Note that it
is often necessary to convert GPS measurements or other ground truth
information to the appropriate coordinate system before making an error
estimate, i.e., a mean square error calculation. If such field measurements are
available during the map production process, they may be used to correct
positional errors and thus improve data compatibility.
Attribute accuracy [18] describes the certainty of the geographical data
properties, such as the exactness of the land-use classes derived from optical
satellite images. An attribute check can be done by using an error matrix. In
an error matrix, accurate reference data is compared with classified digital
map information. This method describes how accurately image classification
is done in reference data areas, but not how good it is elsewhere in the scene.
Classification accuracy has a significant influence on macrocellular coverage
prediction. Classification errors may arise, for example, from using old
source material in areas where new residential areas have been built. This
has a direct effect on coverage prediction and the result is too optimistic if
changes are not taken into account.
Completeness of the data [19] delineates the perfection of the data
contents. This is an important factor when producing building and road
DIGITAL MAPS 113
vectors. Completeness is related to mapping resolution, i.e., the smallest
mapping unit. This means that objects smaller than a certain threshold value
defined in the project plan will not be assigned.
Logical consistency [20] compares the number of the correct attributes,
objects, and their relationships to the data specifications. Topological
consistency is important for the vector information. In building vectors, for
example, two single buildings can be close to each other but cannot overlap.
Road information should be a continuous road network without topological
errors, as illustrated in Figure 3-16. All road lines or edges should meet at
exactly the same coordinates in the intersections or nodes, and all road lines
should intersect only at node points [5].
Semantic accuracy [17] outlines how well the data describes the reality.
Semantic accuracy refers to the characteristics of the geographical objects.