LARGE SPATIAL OBJECT HANDLING IN GEOGRAPHIC INFORMATION SYSTEMS 
Wenjin Zhou 
UNISYS SYSTEM 9 GIS 
61 Middlefield Road 
Scarborough, Ontario, Canada M1S 5A9 
ABSTRACT 
We have seen the increasing availability of large spatial objects in Geographic Information 
Systems(GISs) in the past two decades, mainly due to the interest of multi-media information systems. 
These large objects may come from remotely sensed imagery, scanned maps, digital terrain models, and 
simulation model output. Due to the large amount of data and rich semantics involved in the large objects, 
traditional methods used to handle text and graphics data will not work well for large spatial objects. 
Therefore how to effectively and efficiently store, manage and process large spatial object presents a 
special challenge to the current generation of GISs. 
This paper presents techniques that can be used to handle large spatial objects. After analyzing the 
nature of large spatial objects, we decompose them into three components: row data, Direct Related 
Attribute Data and Derived Attribute Data in order to make them more manageable in GISs. We then 
investigate GIS query patterns on large spatial objects and present an object oriented data model that uses 
objects, methods, rules, and class browser to handle large spatial objects in the context of GISs. We also 
present five abstraction techniques that can be implemented to effectively and efficiently support large 
objects in GISs. Finally we use several practical examples to demonstrate the effectiveness of the 
suggested methods. We conclude the paper with a short discussion of system performance and further 
research issues. 
KEY WORDS: Large spatial object, Object oriented model, Data base, Query Patterns, Abstraction, GIS 
  
  
  
  
  
  
  
  
  
  
  
  
  
  
  
  
  
  
  
  
  
  
  
  
  
Landsat Image Class Digi Imagery. 
image_name: string a" Ÿ 
resolution: array[double] : 
origin: point Satellite Air Photo Video 
size: array[integer] Y N 
plane_no: integer 
time: date Single Band Multi Bands 
quality: integer 
footprint: polygon Sa 
icon: pixelmap Color Multi-Spectrum 
  
  
  
  
  
  
  
  
Figure 1 (a): Class Browser Figure 1 (b): Class Hierarchy Browser 
212 
In the pe 
has been 
spatial c 
terrain m 
interest 
informat 
informat 
roughly 
category 
two Or 1 
processir 
Base Ma 
image ok 
can be | 
Wegener 
category 
systems | 
oriented 
systems 
Deux [19 
However 
in GISs : 
systems : 
image 
indepenc 
the DBM 
limits t 
managen 
unnecess 
because 
method, 
data str 
abstracti 
inefficier 
data set 1 
together 
interacti 
In this p: 
highligh: 
object ori 
digital 
investig: 
Several r 
amount 
efficiency 
Support 
practical 
perform: 
techniqu 
discussic 
research