ISPRS, Vol.34, Part 2W2, “Dynamic and Multi-Dimensional GIS”, Bangkok, May 23-25, 2001 
262 
PRIMARY SPATIAL CHANGES 
Hong SHU 1 , Christopher GOLD 2 and Jun CHEN 3 
1 National Lab for Information Engineering in Surveying, Mapping & Remote Sensing 
Wuhan Technical University of Surveying and Mapping 
129 Luoyu Road, Wuhan 430079 
China 
Tel: +86-27-87881292 
Fax: +86-27-87863229 
Shu_Hong@hotmail.com 
2 Department of Land Surveying and Geo-Informatics 
Hong Kong Polytechnic University 
Hung Horn, Kowloon 
Hong Kong 
Tel: + 852 - 27665975 
Fax: + 852 - 23302994 
E-mail: LsCMGold®polyu.edu.hk 
3 National Geomatics Center of China 
No. 1, Bei Shengcun, Zi Zhuyuan, Beijing 100044 
China 
Jchen@nsdi.gov.cn 
KEY WORDS: Spatio-temporal Database, Spatio-temporal Semantics, Spatial Changes. 
ABSTRACT: Spatio-temporal semantic modeling and spatio-temporal data storage/access are two fundamental aspects of spatio- 
temporal database management systems. In this paper, spatio-temporal semantic modeling is emphasized. Essentially, spatio-temporal 
semantics refer to two perspectives of spatio-temporal scale and spatial change over time, or say, absolute space-time and relative 
space-time perspectives. In the perspective of spatio-temporal scale, geographic objects are positioned with time and spatial 
coordinates, and some measurement functions are accompanied. In the perspective of spatial change, a sequence of changes present 
human an image schema of time. In human cognition, relative knowledge is first apprehended, then absolute knowledge is to refine 
relative knowledge. Likewise, human beings first perceive changes or events, then estimate absolute time, i.e., assign time or date 
numbers to a sequence of events. In spatio-temporal databases, spatio-temporal scale and spatial change are viewed as two kinds of 
equally important semantics. Spatio-temporal coordinates are easy for computer to calculate, but spatio-temporal changes are suited for 
human beings to develop spatio-temporal reasoning. To enrich spatial change semantics, we have completed three parts of work here. 
First, the existing taxonomies of spatial changes are reviewed. Second, three levels of spatial change, spatial property, individual and 
scene changes, are proposed. Primary changes of spatial property include location change, distance change, direction change, size 
change and shape change. Six kinds of primary individual change are functionally given in terms of mapping relationships between 
source objects and result objects. The primary scene change is spatial distribution type change, which can partly be described with 
different statistic quantities. Finally, we attempt to cognitively understand our spatial change categorization. At the same time, it is 
believed that complex changes are composed of primary changes. 
1 MOTIVATION 
In the early 1990s, Gail Langran, in her recognized book 'Time 
in Geographical Information Systems”, began initial studies of 
temporal GIS (TGIS) from concepts to implementations. In 1996, 
CHOROCHRONOS project started up the studies of spatio- 
temporal databases (STDB) in aspects of space-time concepts, 
data models, query languages, physical data structures, 
indexing methods and DBMS frameworks. In the fields of 
database, artificial intelligence and GIS, an increasing number of 
papers are devoted to TGIS or STDB. However, so far few 
results have truly been brought into practice (Timos Sellis, 
1999). This somewhat indicates that a gap exists between our 
wide theories and a large number of spatio-temporal phenomena 
modeling requirements. In our opinion, one efficient way to 
bridge this gap is to further research of spatio-temporal semantic 
modeling, because semantics root in our understanding of 
practical problems. Furthermore, only from the viewpoint of 
geometry, time has no difference with space, and it can be 
regarded as another dimension. A lot of past spatial or 
geometrical database techniques can be easily extended to 
spatio-temporal databases in the form. One of our research 
motivations is to shorten distance between our theoretical 
studies and practical requirements. 
On the other hand, among ongoing research, two topics are 
receiving more attentions recently. They are moving object data 
types (Sistla, A.P., O. Wolfson, et al., 1997; Erwig, M., R.H. 
Gueting, et al., 1997) and constraint-based spatio-temporal data 
models (Chomicki, J., P.Z. Revesz, 1997; Grumbach, S., P. 
Rigaus, L. Segoufin, 1998). Moving object types consist of 
moving points and moving regions, which have been embedded 
in Gueting’s geo-relational algebra. Mathematically, moving 
object types are defined as mappings of time to spatial extent. It 
is easy to see that moving object types directly aim at location 
change modeling. A typical example is moving vehicles, which 
change their locations over time continuously. In constraint- 
based spatio-temporal data models, space and time are 
modeled as two variables in a linear constraint equality or 
inequality, denoted by f(x,y,t)>=0. f(x,y,t)>=0 represents an 
object (identifier and thematic values) with space-time 
constraints. Alternatively, f(x,y,t)>=0 is transformed into h(x, 
y)>=g(t), where h(x,y) represents the spatial extent of an object 
and g(t) is a function of time. Usually, h(x,y) itself is treated as a 
constraint of coordinate variables x and y. h(x, y)>=g(t) means 
spatial extent change over time as well as thematic value 
change. Likewise, constraint-based spatio-temporal data models 
basically describe the location change of an object. The 
difference between moving object types and constraint-based 
spatio-temporal data models is that the former is at the high level