mode 
it and 
lel, to 
abase. 
using 
chnics 
tional 
(f the 
se the 
aster 
1. (2) 
e. (3) 
d the 
chnics 
tabase 
abase, 
more 
CHS 
stems. 
ology 
tabase 
n and 
deling 
te and 
ion of 
> and 
It use 
ıl data 
MIX, 
tween 
> also 
n this 
le. 
)E the 
nodels 
iscrete 
ral or 
ling a 
International Archives of the Photogrammetry, Remote Sensing and Spatial Information Sciences, Vol XXXV, Part B4. Istanbul 2004 
  
; 1 
surface. n 
2.1 ArcSDE Data Dictionary"! 
The ArcSDE data dictionary includes tables that maintain infor- 
mation about the feature classes and feature datasets. 
22 Raster Dataset Management Approach d 
ArcSDE tiles the pixels of the raster into blocks according to a 
user-defined dimension to make it easy to research on spatial a- 
nalysis and application either in a special area or in a whole arca. 
With a certain of compression method provided by ArcSDE, we 
can save disk space and decrease the I/O of network. The user- 
defined raster block and compression method determine the 
need of the disk space. At the same time, ArcSDE populates the 
raster blocks table according to a declining resolution pyramid. 
Spending more disk space it improves the display speed. 
ArcSDE store the raster as a series of tables in a database. They 
are RASTER_COLUMNS Table, Business Table, Raster Table, 
Raster Band Table, Raster Band Auxiliary Table and Raster 
Block Table. 
2.2.1 RASTER COLUMNS: RASTER_COLUMNS table is 
an ArcSDE system table. It records the raster information about 
the database name in which the raster store, the business table 
name the raster belong to, The date the raster column was added 
to the business table etc. 
2.2.2 Business Table: Business table is closed to the real world. 
It regards the raster as one object, and describes the attribute of 
this object. Business table contains the raster column. It is a 
foreign key reference to the raster table. 
2.2.3 Raster Table: Raster table has a description and a bitmap 
set according to the characteristics of the raster. And raster_id is 
the primary key of the raster table and unique sequential 
identifier of each image stored in the raster table. 
2.2.4 Raster Band Table: Each image referenced in a raster 
may be subdivided into one or more raster bands. The raster 
band table stores the raster bands information, such as the tile 
size, the original coordinate of the band, etc. Raster_id column 
of the raster band table is a foreign key reference to the raster 
table's raster_id primary key. 
2.2.5 Raster Band Auxiliary Table: Raster band auxiliary table 
stores optional raster metadata such as the image color map, 
image statistics, and bitmasks used for image overlay and 
mosaicking. The rasterband id column is a foreign key 
reference to the primary key of the raster band table. 
2.2.6 Raster Block Table: Raster blocks table stores the actual 
pixel data of the raster images. The raster is tiled into a lot of 
blocks of pixels. One tile corresponds to one record of raster 
block table. The rasterband_ id column of the raster block table 
is a foreign key reference to the raster band table's primary key. 
A composite unique key is formed by combining the 
rasterband. id, rrd. factor, row. nbr and col. nbr columns. 
ArcSDE realizes the Raster database Management. With the re- 
lation of the tables the description of the raster has its own in- 
tegrity. At the same time, ArcSDE ensures the safety saving of 
raster dataset by an appropriate database backup and recovery 
procedure. 
371 
2. 3 Feature Dataset Management Approach"! 
ArcSDE uses a compressed binary format to store geometry in 
either an Oracle binary LONG RAW or BLOB data type. Arc- 
SDE stores compressed binary spatial data in the POINTS colu- 
mns of the Feature Table. Compressing the geometry offers effi- 
cient storage and retrieval of spatial data by reducing the size of 
the geometry. ArcSDE verifies the geometry, compresses it and 
sends it to the Oracle instance, where it is inserted into a feature 
table in compressed binary format. Compressing the geometry 
on the client offloads the task from the ArcSDE server and red- 
uces the transmission time to send the geometry to the ArcSDE 
server. Storing compressed geometry data reduces the space 
required to store data by as much as 40 percent. 
A feature dataset is a collection of feature classes that share a 
common coordinate system. The simple feature class can be 
inside or outside of feature dataset. In order to describe the re- 
lationship of feature dataset and feature class ArcSDE uses a 
series of ArcSDE system table. The feature class is stored in a 
database as several tables too. They are GEOMETRY_ COLU- 
MNS Table, Business Table, Feature Table and Spatial Table. 
2.3.1 GEOMETRY COLUMNS: GEOMETRY.. COLUMNS 
table is an ArcSDE system table. It records the feature class 
information about the database name in which the feature class 
store, the business table name the feature class belong to, The 
date the raster column was added to the business table, 
Geometry type of the feature class etc. 
2.3.2 Business Table: The business table contains attributes and 
a spatial column. The spatial column is a key to the feature and 
spatial index tables. The relationship between the business table 
and the feature table is managed through the spatial column and 
the FID column. 
2.3.3 Feature Table: The feature table contains the geometry 
attributes information of the feature entity, such as the type of 
geometric feature stored in the spatial column, the number of 
points defining the geometry, the envelope of the geometry etc. 
The feature table stores the geometry, annotation, and CAD in 
the POINTS column. The POINTS column may be defined as 
either LONG RAW or BLOB. 
2.3.4 Spatial Table: The spatial index table defines the grid 
range and extent of all geometry in an ArcSDE feature class. It 
is an assistant table for Building a spatial grid index. The spatial 
index table contains two indexes. One index is on the SP. FID 
column, which contains the feature ID. The other is a composite 
index that includes all of the columns of the spatial index table. 
Since all of the columns of the spatial index table are indexed, 
the values of the table are read from the leaf blocks of the index 
and not the table data blocks. Therefore, when considering how 
to position the tables and indexes to reduce disk I/O contention, 
you should be concerned about the positioning of the indexes of 
the spatial index table but not the table itself. 
3. SPATIAL DATABASE DESING 
3.1 Spatial Data Preparing 
The data of Digital Line Graphic (DLG), Digital Orthophoto 
Map (DOM), Digital Elevation Modal (DEM) and Digital 
Raster Graphic (DRG) are needed during our spatial database 
design.