ISPRS, Vol.34, Part 2W2, “Dynamic and Multi-Dimensional GIS”, Bangkok, May 23-25, 2001 
159 
THE RESEARCH OF THE INFINITELY VARIABLE MAP SCALE IN GIS 
Yifan LI 1 Shaopeng SUN 2 
1 Nautical Science and Technology Institute 
Dalian Maritime University, Dalian, China, 116026 
Tel:86-411-4729297 
E-mail: yifanli@163.net 
2 Nautical Science and Technology Institute 
Dalian Maritime University, Dalian, China, 116026 
Tel:86-411-4729297 
E-mail: sun_sp@usa.net 
KEY WORDS: Variable Map Scale, Spatial Data Model, Spatial Data Warehouse, Web-GIS 
ABSTRACT 
Through the research of the infinitely variable map scale in GIS, this paper bring forwards a set of design philosophy of infinitely variable 
map scale in GIS from data modeling, data storage to data release. This design philosophy can provides powerful application support to 
research and application of Digital Earth. 
1. INTRODUCTION 
The accessing, management and application of data that 
obtained from survey earth, have become the key technology in 
the research of the digital earth. But surveying earth adopts 
multi-scale method, so the data was gathered and accessed 
repeatedly, making vast waste. With the application of Digital 
Earth, we need extract information from the largest scale spatial 
database, map with any scale and any theme, and achieve the 
query and analysis function in geography information system. 
This is infinitely variable map scale information generalization 
technology. 
The infinitely variable map scale technology includes map 
automatic generalization and information automatic 
generalization. The map automatic generalization has been 
researched for twenty years and has gotten achievements. The 
map automatic generalization is mapping according to scale and 
theme, and symbolizes the map. This technology focus its main 
purpose on the automatic generalization of map's graphics, but in 
GIS, its main purpose is not only limited in the map visualization, 
but also to achieve the query and analysis function of GIS, so 
only the automatic generalization cannot satisfy the infinitely 
variable map scale information generalization technology of 
Digital Earth, we need information automatic generalization. The 
information automatic generalization mainly accomplishes 
information filtration, semantic integrality checking, topology 
integrality, consistency checking, topology revert, attribute and 
weight information reconstruction when extracts and generates 
data from large scale spatial database to small scale spatial 
database. It is need to reconstruct data structure if there is 
difference in subject and purpose. This paper brings forward a 
new resolution in information automatic generalization, and 
achieves the infinitely variable map scale technology of GIS by 
combing map automatic generalization. 
2. KEY TECHNOLOGYS AND RESOLVE APPROACH 
The data management of infinitely variable map scale is the 
necessary condition of multi data infusion and emulate & virtual 
technology. But the related geographic data are multiple and 
complicated, the general rule of spatial data cannot be described 
and expressed exactly. Therefore, we need the support of 
advanced map generalization model and database model. The 
data management of infinitely variable map scale is an 
information processing technology to use a big scale (e.g. 
1:50000) spatial database as basic data source, auto add or 
reduce information content of spatial object in designated region 
with the change of scale, in order to make the self adapting of 
compression and recur of spatial geographic data and map scale. 
The distributed data storage of spatial data warehouse, analysis 
and extraction of spatial data provide the theoretical basic of data 
management of infinitely variable map scale. In addition, the key 
technologies of data management of infinitely variable map scale 
also should include: 
2.1 Infinitely Variable Map Scale Oriented Data Model 
There are two kinds of spatial data model in GIS: vector model 
and raster model. The basic unit of vector model is zero 
dimensional object (point). Points were integrated into high 
dimensional object. Vector model objects include point, line and 
region. Point and line are collection of one or many joined and 
sequent points. These objects are close to the boundary of region, 
and form a continuous region. Raster model is using serried 
square grid to discretize data of designated geographic region, 
and represents as pixel matrix. The position coordinates of object 
are implicated in row-column of pixel matrix, and the features of 
object are represented by gray scale of pixels. Point is expressed 
as a pixel; Line is expressed as a cluster of adjoint pixels; Region 
is expressed as pixels that fill in inside of the outline. With the 
development of GIS and Remote Sensing, researchers want find 
an integrated data model, and explore an integrated data 
structure. Hybrid model or Composite model is the third data 
model that combines vector and raster, and absorb their merits. 
Procession of infinitely variable map scale spatial data need fast 
compression and recur. Raster model can complete the 
compression and recur quickly and vector model can restore 
topological relationship, geometrical relationship and attribute 
relationship of GIS. Therefore, we should adopt hybrid model 
which combines vector and raster model as our infinitely variable 
map scale spatial data. In this hybrid model, we use vector model 
to deal with the geographic relationship of compressed infinitely 
variable map scale information, and use raster model to achieve 
the visualization of geographic data. 
2.2 Spatial Data Warehouse 
The functions supported by spatial data model and spatial 
database determine the realization of functions of GIS. Thus, in 
order to realize infinitely variable map scale function, we need 
spatial database that has map generalization and information 
generalization ability. 
Based on definite theme, spatial data warehouse integrates data 
from different databases, so data is integrity in structure. It can 
extract data from different time measure, such as instant, period 
of time, all time. Thus, we can deal with multi-application in 
different fields. Spatial data warehouse integrate temporal 
attributes and spatial attribute tightly. Through building analysis