The International Archives of the Photogrammetry, Remote Sensing and Spatial Information Sciences. Vol. XXXVII. Part B6b. Beijing 2008 
218 
we must specify the baseline, here we use some certain old 
version map/database as baseline map. So change-only 
information is different according to different baseline map. In 
principle we can designate many baseline maps according to 
different applications to reduce data volume. We can use 
change-only information based on baseline map to reduce the 
volume of updating data. Change-only updating model is based 
on incremental information description according to baseline 
map and user data status, and the updating must be carried out 
to add change-only information on client data. In result the 
updating will result in modification of customer database to be 
consistent to current spatial data. 
Change-only updating becomes the trend of spatial data 
updating because of its model flexibility and guarantee of 
geographic data currency. Change-only updating of spatial 
database includes three components, that's detection and 
extraction of change-only information based on version 
databases, management and distribution of change-only 
information, updating of client database. Collection of change- 
only information is the foundation of change-only updating. 
According to the given national foundational information 
system, the paper finds the changes and change-only 
information through the overlay analysis and contrast of 
different snapshot with different version databases. We name 
the old and new version database respond to time stamp. 
The main framework of change-only information and updating 
can be delivered by figure 1. We first normalize the version 
databases to determine the candidate geographic datasets to 
match the corresponding objects. Then we analyze the 
relationships and identify the change taxonomy among objects 
between old version database and new version database through 
geometric, thematic and semantic analyses. With the definition 
and data structure of change-only information we then can 
detect and extract the changes of geographic data, and store and 
manage the change-only information. Last we can publish the 
change-only information as standard exchange file to update 
and integrate with client databases. 
There are many key techniques and difficulties in the change- 
only updating. First we should build rapid index and query 
mechanism to locate and match spatial objects efficiently in 
different version databases with the very large volume 
geographic data. Fortunately there are many spatial index and 
database techniques that we can utilize, such as R-tree, Oracle 
software. The second problem is to develop efficient method to 
extract change-only information. We use object match method 
to determine the difference between old objects and new objects 
through geometric, social and spatial properties. We consider 
that there are quantitative changes and qualitative changes with 
the changes of geographic features. It’s not enough to 
distinguish them only by the geometries of spatial objects. We 
should take the semantics and neighbourhood of spatial objects 
to the changes type. The representation and modelling of 
change-only information is also the key problem, which should 
resolve the representation difference of objects in old database 
and new database to support extraction and storage of change- 
only information. Importantly, we must insure that the client 
users can identify and parse the change-only information to 
integrate it with their own database. 
3. DESCRIPTION OF CHANGE TYPES 
Before the extraction of change-only information, the 
corresponding change relationships of realistic entities before 
and after change should be made clear. Spatial-temporal model 
based on change-only information focuses on the changes or 
events of spatial objects, and the corresponding relationships of 
spatial objects should be matched and analyzed. These 
corresponding relationships is analyzed based the representation 
of spatial objects based on different version. In the users’ view, 
spatial data updating is the conceptual description of changes 
with geographic entities. In other words, the users pay more 
attentions to changes of geographic semantics. Claramnut et al. 
(1996) gave three types changes: basic processes, 
transformation processes and movements that include eight 
detail types. This analysis is suitable for single spatial object, 
but it can not deal with the multi objects and groups in spatial- 
temporal change. For road element in the paper, we take 
Claramnut’s view as reference to analyze the spatiotemporal 
road features, and take multi-objects into consideration, analyze 
the change states of realistic entities, divide the changes into 
seven types: appearance, disappearance, splitting, connection, 
reshape, semantics change and modification, stability. Table 1 
illustrates the details changes of spatiotemporal road features 
with the descriptions and examples. 
Spatial 
database 
version! 
< 
Version match 
version2 
/ 
Object 
location 
U 
Version n 
Figure 1. Framework of change-only information and updating