rst 
of 
ic 
ps 
nd 
n 
le 
ly 
he 
es 
le 
Or 
n 
geometry at the user's side, or a system based on unique 
identifiers, maintained at both the producer's and the 
user's side. 
Change-only information can be produced through 
graphical comparison of features or through a selection on 
database entry date. Research has been carried out on 
update deliveries of large scale databases (Gray, 1995) 
and some methods are already operational. The Dutch 
Cadastre has an automatic update delivery system, where 
users can choose their own update frequency. Change- 
only information is extracted from the master database 
through selection on database entry date (Lemmen, van 
Oosterom, 1995). 
4. TOP10VECTOR REVISION 
Nowadays, three revision cycli of respectively 4, 6 or 8 
years exist for 1:10,000 mapping, depending on the 
characteristics of the area. From 1997 onwards, a four 
year revision cycle will be carried out: 25% of The 
Netherlands will be revised completely every year. As a 
result, cyclic revision will take place, where all feature 
types will be updated. TDN does not have the intention 
to store historic data in the datasets. 
This year, TDN started to update the first eight 1:10,000 
map sheets on the basis of the existing TOP10vector- 
datasets. The revision process is subdivided in the 
following steps: 
1. data-collection 
2. processing 
3. checking 
4. delivery 
The process is visualized in figure 4. 
4.1. Data-collection 
Data collection for TOP10vector-revision does not differ 
from  data-collection for the initial TOP10vector- 
production. A positive film is produced from the existing 
dataset. Non-existing topography is marked on the film, 
new topography is annotated on the enlarged aerial 
photograph. This is done during the preparation stage in 
the office and during the reconnaissance survey in the 
field. 
4.2. Processing 
Digital orthophotos are produced from the enlarged 
annotated photographs. The method is already described 
in paragraph 2. A copy of the existing dataset is used for 
the update process. The marked features on the film are 
deleted from the dataset and new features from the 
orthophoto are added through heads-up digitizing. The 
date of deletion or addition of a feature is not captured. 
The year of aerial photography is stored for each feature, 
895 
but this attribute value is not used to produce change-only 
datasets. All features, unchanged and new, are checked 
on validity. Therefore, the whole dataset gets one 
reference date. 
4.3. Checking 
During this step, a check is carried out whether all non- 
existing features are deleted and new features added. 
Change-only datasets are produced for that purpose. The 
existing TOP10vector-dataset is automatically compared 
with its updated version. Change-only information is 
extracted in order to produce two types of check-plots. 
One type shows all deleted features, the other all new 
features. Checking is done by comparing the marked 
features on the film with the plot showing the deletions 
and the annotations on the photograph with the plot 
showing the additions. On the basis of these comparisons, 
corrections are made, after which TOP10vector-revision 
is finalized. 
4.4. Delivery 
The updated dataset can be delivered to users. It is also 
possible to compare the updated dataset with the old 
dataset in order to produce change-only datasets for the 
revision of TOPSOvector and for users. 
5. CHAN GE-ONLY DATASETS 
Change-only datasets are produced for check-plots, for 
delivery of updates to users and for updating 
TOP50vector. This paragraph describes the method of 
change-only extraction. 
The individual files (BAS, AREA, etc.) of the old and 
updated datasets are, one by one, automatically 
compared: unchanged features in terms of geometry and 
attribute values are deleted in both files, which results in 
files with deleted and added features. During the 
comparison, a tolerance value can be used in order to 
treat features with small geometrical changes as 
unchanged. Five different reasons can be distinguished 
why features appear in the change-only datasets: 
1. changes in data collection rules and content 
specifications 
2. changes in topography 
3. corrections/improvements in the existing dataset 
4. changes due to edge-matching of sheets in different 
years 
5. changes due to structural changes (splitting or merging 
of features). 
Changes in rules: It is clear that every change in the 
content specification will have its effect on the change- 
only datasets. Changes in specifications should be avoided 
as much as possible. However, we deal with a new 
International Archives of Photogrammetry and Remote Sensing. Vol. XXXI, Part B4. Vienna 1996