International Archives of the Photogrammetry, Remote Sensing and Spatial Information Sciences, Vol XXXV, Part B2. Istanbul 2004
Internati
operation has to be chosen for each attribute. For example, if questions. However, more processes are required to answer a inne
two new road polylines were found to be matched to one road state-based query. be t
polyline in the old road map, the name of the old road could be
“assigned” to the two new road features, while the “length” In a typical vector GIS, there are some additional problems 3. The
value should be “departed” to the two new ones. with a change-based approach. Firstly, difficulties will occur in all t
calculating the differential value of non-spatial attributes, such netv
4.6 Modeling the changes as "name", "class" and "pavement". Secondly, although the
nr change representation proposed in [Mountrakis, et al, 2002] is AY Ald
Organize the newly-updated road database and historic simple and straight forward for a raster-based GIS or a polygon bast
database using a spatio-temporal model for road network. A theme, it is not sufficient for a polyline change representation. the
réasonable spatio-temporal model for road network is still Thirdly, to store the difference between two lines usually or
under development. requires the same storage as to store a copy of one of the the
original lines. sati:
5. ROAD CHANGE REPRESENTATION IN GIS We would argue that the choice of a state-based approach or a
change-based approach is really an application-dependent 4) The
It is still an open issue to model changes in GIS. Many problem. In the current prototype system, a state-based fox
problems should be addressed in a productive environment. approach is used to organize all the non-spatial attributes and a fous
Several considerations regarding the change representation of change-based method to model the spatial changes of the road
the road network are discussed in the following subsections network. 5) Bot
which are not necessarily complete. easi
5.4 Entity ID issues
5.1 Decomposition of non-spatial attributes
In a pragmatic system, road features will have a number of non- Ina CIS, gag h Scograp ; kc erty is assicned a un ique idenifiee
spatial attributes. Some of these attributes (e.g. length, width) (i to distinguish t frem ali other geographic entities Toc The «wi
will change as the road's geometric or spatial attributes change ol i of am ID hase eren effect on he Spatic-temperal S
while the others (e.g. name, pavement materials) will change database design and implementation. Unfortunately, there is no i i h
independently of the spatial attributes. Therefore, we suggest ünique Way to mas deciston on When te assem a REN ID Or isi
that we distinguish these two types of non-spatial attributes in when it fas to remain the Sam It Ian application oriented M e th
road spatio-temporal modeling to minimize the total storage issue. The hokey el asstumung ad ID in a esten information GR >
and improve the access efficiency. system is definitely different from the method in a road lows
management system system.
5.2 Choice of spatial data structure
If a spaghetti structure is used to store the spatial data of a road,
i.c. a series of coordinates of points is used as in ESRI Shape
Generally speaking, there are three kinds of spatial objects of
concern in a road management system. They are polylines,
nodes, and vertexes. A polyline is a representation of a real
world road. Each polyline will have an ID to relate uniquely Ed IRS Ot
file or Autodesk DXF file, many problems might occur in both : ; : ; Benton
e Y ss in , Hiany proni E 5 od with the real world road. Any information about the road will El InS Ot
spatial and temporal data processing. For example, difficulties be related to the same ID if all the changes are within the
to maintain topological consistency may occur. In addition, condition to maintain the same object ID according to the
storing a copy of the whole series of coordinates will be needed specific application. On the contrary, the vertexes have no extra |
%
even if only part of the polyline changes. So a topological
spatial data structure is a better suggestion to store spatial data
of the road networks.
5.3 State-based vs. change-based
In early approaches, representations of objects (states) were
stored at different time instances. In this case change
information besides the positional information and a new vertex
ID could be assigned whenever its position changes. However,
this is not true in a cadastral information system, in which every
point along the cadastral boundary (i.e. vertex) is important to
the management units. To assign a new road node ID or not
really depends on how many non-spatial attributes should be
attached to the node entities and whether these attributes
change along with the position change. For example, in a real
= Laver Informaiton
Uunert Cope
{RS Oftawas 010
; ; ; i Ready
information was handled indirectly, as it was not stored but world a road junction could be reconstructed and its position
could be calculated using the stored data. More recently may be changed. But some other attributes of the road junction Figur
change-oriented approaches have been introduced, focusing
mostly on qualitative attributes of geospatial entities [Peuquet
and Wentz, 1995] and variations of these attributes [Hornsby
and Egenhofer, 2000]. Along this line, Mountrakis, et al (2002)
also proposed a differential change model in which the change
information is treated as explicit information.
incorpor:
In fact, each option offers benefits and costs. A state-based Based on the above discussions, the main features of the
spatio-temporal database can answer a state-based query proposed spatio-temporal data model for the road network
efficiently, e.g., how did the road network look in 1998? could be summarized as follows:
However, it has difficulty in answering a change-based query, Road cl
e.g., how many roads changed their centerlines during 1990 to 1) While the change information of a road line or/and a road remotely
1992? In which sub-region did the road network change to the node is recorded in the model, the change information of a projects
rater ri Dp? (? he > T ach: ». haee < Pan 5 . E ^
greatest degree: On the other hand, a change-based approach vertex is disregarded. In other words, any change of an challeng,
may have fewer problems in answering these kinds of extractio
may still remain the same as before, such as the name or traffic
signs. So we would suggest that we keep the same ID for the
road node even if its position has been changed.
5.5 Summary of the spatio-temporal modeling for the road
network
The sys
processit
updating
used for
the histo
732
