Raster datasets, particularly those for satellite imagery, are generally much larger than vector datasets and often require many terabytes of storage. In addition to large source datasets, the multiple versions created during traditional processing steps are saved and exacerbate storage issues. This aspect alone has made storing, managing, and distributing imagery challenging when using traditional methods.
In most cases, image data is most valuable when it is new. It rapidly depreciates. Depending on the application, its value can decline substantially over a matter of hours or days. However, for change detection and other applications, older data becomes useful for comparisons with current data. A related issue is the need to make imagery available to many users rapidly.
Imagery represents the surface of the earth at a specific moment and, in that sense, it is fixed. Unlike vector data, it is not edited although it may be processed to create different products that will be used for visualization or analysis.
The imagery used in GIS can be mosaicked tiles; rectified scenes; or raw, unrectified data. Mosaicked tiles are an arrangement of images that have been processed and cut into tiles. Digital orthophoto quarter-quadrangle (DOQQ) and controlled image base (CIB) are familiar examples of imagery supplied in mosaicked tiles. Imagery in this format provides a backdrop for vector data. Rectified scenes are acquired by satellites at a specific time for a specific sensor and are rectified to the ground using a spatial reference system. Nonrectified scenes, from satellite sensors or aerial photography, contain the most information content. However, this data requires geometric and radiometric enhancement before being used in a GIS.
Not Just More but Better Imagery
Not only has the amount of imagery data available increased in recent years owing to the launch of new satellites, such as DigitalGlobe's QuickBird and GeoEye's IKONOS, newer satellites supply better-quality imagery. These satellites carry sensors capable of greater resolution. Bit depth or spectral resolution has also increased from 8 bits to 12 bits per channel. Some satellites have hyperspectral sensors that capture hundreds of spectral bands.
An abundance of imagery has made it less expensive to purchase. Lower cost and greater availability of higher-quality imagery data has spurred the development of new GIS functionality. Now imagery is commonly used for analysis and data verification as well as a backdrop for other data layers.
Supporting the Imagery Demand Cycle
Because imagery represents the conditions on a specific date and is not edited, there is a constant requirement to update imagery. However, greatly increased collection capabilities aren't the entire solution.
This reality has been recognized by the U.S. intelligence community. Constantly updating geospatial imagery requires addressing the entire tasking, processing, exploitation, and dissemination (TPED) cycle that produces geospatial imagery information about natural and man-made objects. TPED is an end-to-end cycle that encompasses recognizing needs and planning and directing activities to meet those needs; converting raw image data into one or more usable formats; extracting information and fusing it with other types of geospatial information; and, finally, transferring and storing image data and derived information. This comprehensive approach is driven by the government's need to assess military and national security threats in a constantly changing environment.
ArcGIS 9.3 on the server, desktop, and mobile platforms and in online services supports TPED for processing, exploitation, and dissemination of imagery. An accompanying article in this section, "Visualizing an Enterprise Approach to ImageryManaging, analyzing, and serving imagery in ArcGIS 9.3," describes in greater detail improvements in working with imagery data across platforms to enhance geospatial knowledge and allow this knowledge to inform action.