ArcUser Online
 

July - September 2002
Search ArcUser:
 
ArcUser Main Current Issue Previous Issues Subscribe Advertise Submit An Article
 

Raster Data Management with ArcSDE
Continued...

Raster Catalogs

A raster catalog displays many individual rasters as a seamless layer without having to mosaic them. Raster catalogs are useful if the data in the catalog will be updated frequently or if there is a requirement for maintaining data in areas where images overlap. Definition queries can be used to select the specific images that will be returned to the user. The fields required for queries should be added to the raster business table so they can be queried by the client application. Rasters to be stored in a raster catalog must have the same spatial reference and should have the same spectral properties (i.e., colormap).

ArcSDE supports two types of raster catalogs-referenced and embedded. Referenced raster catalogs reference separately stored raster layers. Embedded raster catalogs store individual rasters in one business table. The best type of raster catalog to use will depend on how the raster data will be used.

Mosaics

Raster mosaics are images that have been created by the merging of two or more individual images. Mosaics are useful if the underlying data is not changing or any overlap of the images is inconsequential. One other benefit is the speed with which they can be displayed. Mosaics are the fastest raster data type to display. The creation of large mosaics can be time-consuming due to the processing that is required. Preprocessing steps that may be required depend on the loading tools used, but may include pixel alignment, color balancing, and projection. Processing undertaken when loading includes the creation of pyramids and calculation of statistics, typically completed after all the individual images have been loaded.

Neither sderaster or the ArcGIS tools will allow the mosaicking of rasters that have colormaps as they may differ; however, the sderaster command will allow the loading of rasters with colormaps by ignoring the existence of a colormap. Colormaps can then be later added to the mosaic using sderaster.

Referenced Raster Catalogs

Column NameData Type
IMAGEString
XMINFloat
YMINFloat
XMAXFloat
YMAXFloat
RASTER_IDInteger
Figure 3: Fields required for a referenced raster catalog

This type of raster catalog is a stand-alone database business table that references rasters stored elsewhere on disk or in the database. This table must be created manually in the DBMS being used. The six required fields are listed in Figure 3.

Running the sample SQL code in Figure 4 in the SQL execution environment for a DBMS creates a raster catalog named "mycatalog." The value for the IMAGE field is the full path and file name of the file on disk or the fully qualified database name (i.e., []..) for the raster in the ArcSDE database, as it appears in ArcCatalog's treeview. The RASTER_ID field is used for individual images stored in an embedded raster catalog. The value is the rastercolumn_id of the raster.

CREATE TABLE mycatalog (IMAGE varchar(50) NOT NULL, XMIN float(53) NOT NULL, YMIN float(53) NOT NULL, XMAX float(53) NOT NULL, YMAX float(53) NOT NULL, RASTER_ID integer)

Figure 4: Sample code for creating a raster catalog

Embedded Raster Catalogs

This type of raster catalog is a business table with a raster column that has multiple rows corresponding to individual rasters. An embedded raster catalog is created by loading multiple rasters into the same business table using the Raster to Geodatabase tool in ArcGIS (available from ArcCatalog and ArcToolbox), or the sderaster command in ArcSDE. An embedded raster catalog has an advantage over a referenced raster catalog in that it requires fewer unique layers that must be stored and accessed by clients.

Tools for Converting Raster Data

Both a wizard and a tool, available from the ArcCatalog and ArcToolbox interfaces, supply raster to geodatabase conversion functionality. A command line based tool, sderaster, is also available. Figure 5 provides a comparison of the characteristics of two interface-based methods versus the command line method. To learn more about the tools read the articles "Loading ArcSDE Rasters Using ArcGIS 8.2" [PDF-275 KB] and "Load Rasters in ArcSDE Using sderaster" [PDF-335 KB] in this section.

Platform Support

Available for all the platforms ArcSDE is supported on (Compaq Tru64, HP HP�UX, IBM AIX, Redhat Linux, SGI IRIX, Sun Solaris, Windows NT, and Windows 2000).

ArcGIS tools are supported on Windows NT, Windows 2000, and Windows XP.

User Interface

Command line tool with no graphical user interface (GUI)

The ArcGIS tools are Windows applications that have a GUI.

Input Formats

Supports importing TIFF or BSQ formats

Can load a wide range of formats.

Automation

Can be automated using operating system shell scripts

Can be customized using ArcObjects, sample scripts from the ArcObjects Developer Help; Offers a batch mode that can convert multiple rasters at one time.

Data Loading Options

Allows user to specify custom Pyramid Levels, Remove Colormaps on loading, Specify No Data values and whether or not to build pyramids or statistics at load time

Can specify whether or not to build pyramids or statistics at load time

Data Exporting Options

Allows the clipping and extraction of a raster layer

Can clip and extract portions of a layer with the use of ArcObjects

Mosaicking Requirements

Requires rasters to be spatially referenced, have the same bit depth and same number of bands, exact pixel registration, and same cell size.

Specifies raster must be spatially-referenced, have no color map, and have the same bit depth

Figure 5: Comparison of sderaster command and ArcGIS tools

Conclusion

With ArcSDE, large raster data warehouses can be developed using industry standard DBMS technology. This allows many concurrent users to quickly access raster data for query and analysis as well as allowing for user-defined data extraction parameters when needed. For more information on raster management in ArcSDE, visit the Support Center for white papers, a discussion forum, and other resources.

Contact Us | Privacy | Legal | Site Map