ISPRS, Vol.34, Part 2W2, “Dynamic and Multi-Dimensional GIS”, Bangkok, May 23-25, 2001 
317 
Fig 2 Distributed GIS CORBA ORB Server Group Figure 
3.1 Projection transform ORB 
According the concept of spatial reference system of Geodata 
Model, attributes are tied with features by coordinate. The spatial 
reference system is function which associates locations in space 
to geometries of coordinate tuple in mathematical space, usually 
a real valued coordinate vector space. 
The Conversion ORB’s Interface is described in the file: 
ProjectionTransform.idl 
interface Projection 
{ 
attribute string ProjectionName; 
attribute long ProjectionNumber; 
àttribute double originateX; 
attribute double originateY; 
void LLtoProjection(in double latitude,in double lantitude, out 
x,out y); 
void ProjectiontoLL (out double latitude,out double lantitude, in 
double x,in double y); 
} 
Fig 3 Projection Transform Model 
Projection transform ORB supplies operation for coordinate 
transformation, and allows points to be transformed between any 
ellipsoid coordinate system and projection coordinate systems 
such as Gauss and UTM. It provides for a "domain of validity" for 
each coordinate system transformation, as most transformations 
are valid only over a portion of the Earth. Exceptions are raised 
when a transformation is invoked outside its domain of validity. 
3.2 Format Transform ORB 
Geographic data has been collected in digital form for more than 
30 years. The overall rate of collection increases rapidly with 
advances in technologies such as high resolution satellite-borne 
imaging systems and global positioning systems. Geodata 
formats tend to be complex, more complex than other kinds of 
digital data formats, because of the range of information they 
must be able to represent. Usually, the complexity that begins 
with the underlying digital format imposed by a particular 
software application or acquisition method is incremented by the 
complexity of higher level descriptions, conventions, and rules 
imposed by the individuals, organizations, and disciplines using 
the software. Because spatial data exist in a wide range of 
incompatible and often vendor-proprietary forms, and geographic 
information systems (GIS) usually exist in organizations as 
isolated collections of data, software, and user expertise. So I 
developed the format transform ORB including interface for 
exchanging diverse geodata then applications can access 
remote spatial databases and spatial processing resources in 
real-time. 
3.3 Spatial Database Storage ORB and Query ORB 
Geospatial data are structured based on the Open Geodata 
Model of the OGIS[4], Spatial entities are represented as 
features. Features are the primary units for access, management, 
manipulation and interchange. A feature is specified by 
geometric and attribute properties. Based on the Open Geodata 
Model, three basic spatial entity types are used to specify zero, 
one, two and three dimension map features. But spatial data in 
files are difficult for query. For quick query they are indexed in 
spatial databases. Our spatial data structure is based on spatial 
occupancy. R-tree is designed to organize a collection of 
arbitrary spatial objects. Each node in the tree corresponds to 
the smallest d-dimensional rectangle enclosing its son nodes. 
We developed the main spatial query methods including buffer, 
overlay and so on. They are listed in Table 1. 
Table. 1 Spatial query method for distributed GIS 
Function 
Output 
Overiay(Ri,R 2 ) 
Overlay two maps: Ri,R 2 
Buffering^,) 
Buffer of line, polygon and point 
Distance(Ai,A 2 ) 
Distance of point Ai and point A 2 
Within(Ai,A 2 ) 
If a line,point or polygon A, within A 2 Q 
then return true 
lntersect(Ai,A 2 ) 
If line A, and Line A 2 intersect n then 
return TRUE 
Contain(Ai,A 2 ) 
If polygon A^ontain polygon,line or 
point A 2 Dreturn TRUE; else FALSE 
Adjacent(Ai,A 2 ) 
If polygon A,and Polygon A 2 adjacentD 
then TRUE; else FALSE 
Because query in a distributed GIS involves a large number of 
spatial operations, it is more complicated than query in a normal 
distributed relational database. When query for remote data, 
data distribution and transformation costs must be considered. 
3.4 Application ORBS 
Application ORBS is the kernel of distributed GIS. Main 
geoprocessing is finished by application ORBS. They can be 
divided as two groups. One is the general application group