system; have temporal changes occurred in spite 
of a correct classification during the field 
verification phase i.e. type of crop changed 
from maize to potatoes and finally have there 
been any errors in the field verification phase 
itself. 
- data completeness, though decisions taken on the 
extent of the data base content: this aspect of 
completeness, indicating the extent to which 
features on the ground are actually included in 
the data base, is related to the problem of 
generalisation in the traditional mapping 
environment. Here, in order to keep the maps 
readable, fewer and fewer features can be 
represented as the scales decrease and therefore 
they are simply not picked up. This type of 
  
reasoning does, however, not apply to data bases 
and so, in terms of assigning a quality label 
for this type of completeness, the content 
requirements of different data bases will have 
to be quantified in relation to the maximum 
completeness. E.g. assume that a 1:50,000 map 
series has been found to have an average 
generalisation percentage of 30%. If now these 
1:50,000 maps are digitised to create a data 
base, the quality indicator for the completeness 
of the digitised data based only on data base 
content will be 70%. 
Data with a security classification also fall 
into this category, but the effect on the 
completeness percentages will depend on whether: 
a) the security label prevent general access to 
these elements, whereby the data is de facto 
entirely removed from the data base. If this 
concerns 5% of the features in an area, the 
data completeness percentage become 95% due 
to security restrictions. 
b) the above solution has the disadvantage of 
yielding gaps when a graphical plot is 
produced. To overcome this, some agencies 
retain the feature (in simplified form in 
open areas) and give it a general 
classification such as government building or 
government area-restricted. With regard to 
the simplified form of the security object in 
open areas, this will generally only include 
the boundaries of the complex, plus the 
location of through roads, if the complex has 
more than one entrance. 
Besides eliminating gaps, this alternative 
also increases the data completeness 
percentage. 
Some authors advocate that data security be 
treated as a separate item in attribute 
accuracy standards, but the majority of W.G. 
members were in favour of retaining it as an 
integral element within data completeness, as 
affected by decisions on the data base 
content. 
- data completeness, expressing the degree to 
which features which should have been included 
in the data base, have actually been picked up. 
The data completeness depends on factors such as 
the quality of the field verification phase i.e. 
have all missing features been picked up; the 
temporal validity of the data i.e. the extent to 
which all changes have been recognised and new 
features have been picked up; the lack of 
598 
accessibility to the data due to security 
restrictions and finally, missing and thereby 
incomplete data due the fact that the input data 
has been generalised. This will occur, for 
example, when the individual buildings have been 
generalised by the built-up area symbol on a 
medium scale map and this map is subsequently 
digitised. 
In these guidelines, attribute accuracy will be 
expressed as a percentage of the correctness and 
completeness of the attribute, in relation to the 
true situation. There are three of these quality 
indicators for attribute accuracy: data 
classification, data completeness related to data 
base. content and data completeness related to 
errors in data collection. This implies that the 
final figure for attribute accuracy is the product 
of the three figures obtained for the indicators 
e.g. with figures of 85% for data classification, 
90% for completeness (content) and 80% 
for completeness (errors), the attribute accuracy 
figure will be 0.61 or 61%. 
Besides determining the attribute accuracy for 
each data item class, summaries will also be 
produced of groups of items falling within 
different accuracy classes, in order to provide 
useful indicators for map revision needs. 
Attribute accuracy is only readily verifiable in 
the field during the field completion or field 
verification phase. An exception to this is the 
figure for data completeness related to data base 
content, which must come from the organisation’s 
statistical records. 
3.4 Data history 
The data history describes the standards 
themselves and summarises the methods used in data 
collection and data quality verification. It is 
probably the most important element in the data 
quality standards, simply because it allows users 
accessing the data to correctly interpret the 
implications of the quality measures used. 
3.5 Logical consistency 
Logical consistency is a general term for fidelity 
in representing features in a data structure. It 
thus does not belong to the data collection 
parameters described in these standards and will 
not be treated further here. 
3.6 Data security 
Instead of taking this as a separate item, data 
security has, in these standards, been included as 
a component of the quality indicator data 
completeness, in relation to data base content, 
described in paragraph 3.2. 
3.7 Data decay 
The decay rate of data has been mentioned in 
paragraph 3.2 as being an important factor, 
affecting the quality of both data classification 
and data completeness. Nevertheless, decay rate is 
probably not important enough to be included as a 
separate item defining quality within the 
^ - ee