Imagery and GIS. Kass Green
Russia, China, India, and Israel all have constellations of satellites that are fully funded by their government agencies but whose imagery use is strictly restricted to security agencies.
The passage of the 1992 Land Remote Sensing Act made it possible for US commercial companies to build, launch, and operate satellite sensors able to collect high-resolution imagery globally. Although fully commercial, the first companies to launch high-resolution systems received large contracts from the National Geospatial Agency of the Department of Defense for imagery. As a result, the funding for the imagery is part government and part commercial. The commercial companies distribute the imagery through licensing agreements that restrict either the amount of time the imagery is available for use or the sharing of the imagery with other organizations. This quasi-public/private funding model for high-resolution satellite imagery with licensing restrictions has since been replicated by several companies (e.g., DigitalGlobe, Airbus, Planet, and DMC constellations).
Access
Organizations make imagery available in a variety of ways. It can be delivered on a hard drive, downloaded from the web, or served as image services. Because imagery files are very large, access can be problematic and can affect the cost of working with imagery. Free imagery with no license restrictions can still be difficult to use if its access is cumbersome.
Before digital sensors, imagery was accessed as hard copy negatives and photographs. Reproduction of the negatives and photographs was very expensive and, as a result, access to them was limited. With the adoption of digital sensors, digital imagery was initially accessed from tape, and then from hard drives and CDs, and processed first on mainframe computers and then on desktop computers.
Until recently, the most efficient way to deliver and gain access to high-spatial-resolution imagery for analysis was still by shipping hard drives and then using on desktop machines or serving the imagery locally. With increases in Internet bandwidth, imagery is increasingly accessible by FTP download or direct access from cloud storage. In this way, imagery can be downloaded to desktop machines or directly used in the cloud infrastructure.
Over the last five years, several imagery providers and software companies have begun to host imagery in the cloud and offer direct visualization, analysis, and processing of the imagery. Most notable is Esri’s Landsat services, which obtain Landsat imagery hosted on Amazon Web Services and provide access and on-the-fly processing of large collections of multitemporal multispectral Landsat imagery that is updated daily as imagery is acquired by the USGS. Google also hosts archives of Landsat imagery and provides processing to educational and research organizations.
Case Study — the Effects of Price and Licensing on the Use of Landsat Imagery
The history of Landsat imagery is a good example of how organizational characteristics affect imagery use. Landsat satellite imagery is moderate resolution, multispectral, and funded by US taxpayers. NASA launched the first Landsat satellite in 1972. The spatial resolution was coarse (80 meters) and included only four bands (green, red, and two infrared bands). Technological barriers slowed the use of the imagery because the knowledge base was small, little image processing software existed, and the files were huge for that time, requiring mainframe computers. Most users were NASA or academic scientists and government agencies. Landsats 2 and 3 were similar to Landsat 1.
In 1979, the Landsat program was moved from NASA to NOAA. In 1982, Landsat 4 was launched and included a 30-meter resolution instrument that collected seven bands of imagery, adding two middle-infrared and one thermal band. A similar system, Landsat 5, soon followed in 1984. However, Congress passed the Land Remote Sensing Commercialization Act of 1984, which directed NOAA to migrate Landsat imagery distribution from the federal government to the private sector with the hope that revenue from imagery sales would support the continuation of the Landsat program. As a result, the cost of Landsat imagery increased from $2,800 per scene from NOAA to $6,000 per scene from the commercial company EOSAT, and use of the imagery was license restricted. The demand for imagery sharply declined, as did Landsat research and innovation (Draeger et al., 1997).
In 1992, Congress passed the Land Remote Sensing Policy Act (Public Law 102-555), which ended Landsat commercialization by designating the USGS to take over distribution of Landsat 7 imagery when it was launched (Landsat 6 failed to reach orbit). The act required that imagery be priced at the cost of fulfilling user requests and have no licensing restrictions. Landsat 7 was successfully launched in April 1999, and the USGS initially set the price of a scene at $600. The lower price of Landsat 7 imagery forced the company distributing Landsat 4 and 5 data to match the price of Landsat 7 imagery. Unable to run Landsats 4 and 5 profitably, the company returned its rights to distribute Landsat 4 and 5 imagery to the federal government in 2002. The lower price and unrestricted licensing for all Landsat imagery resulted in a dramatic increase in the operational use of Landsat imagery, with government revenue from image sales growing from $4 million in 1999 to $11 million in 2002. However, access to the imagery was still cumbersome and slow, requiring the manual ordering and writing of CDs.
With improvements in the web and automation of the USGS distribution processes, the agency made Landsat imagery free and downloadable from the web in 2009. As a result, the use of Landsat imagery skyrocketed from 20,000 scenes to 2,000,000 scenes a year, and commercial companies such as Esri are hosting Landsat imagery and processing services, which further increases global access to the imagery.
Summary — Practical Considerations
In this chapter, we have learned how imagery is differentiated by a combination of technical and organizational characteristics. An image’s sensor and platform determine its technical characteristics—its spectral, radiometric, spatial, and temporal resolutions, as well as its viewing angle, and extent. In summary:
Spectral resolution—Terrestrial, airborne, and satellite platforms can and do carry all types of sensors. Currently, panchromatic, multispectral, and hyperspectral sensors can be found on terrestrial, airborne, and satellite platforms, as are active and passive sensors.
Radiometric resolution—Older sensors will often have lower radiometric resolution than newer sensors because newer sensors can take advantage of continual improvements in digital arrays, memory, and storage.
Spatial resolution—Airborne systems are more commonly used to collect high-spatial-resolution imagery than spaceborne systems if the infrastructure to support aircraft is available and if the aircraft have access to airspace. If access to the air is limited, satellite systems or drones can be used to collect high-resolution imagery. Moderate-and low-spatial-resolution imagery is best captured from satellites.
Temporal resolution—Geostationary systems offer the highest temporal resolution, but at either a lower spatial resolution (e.g., weather satellites) or a smaller extent (e.g., video cameras at ATM machines) than airborne or satellite systems. Airborne systems are more flexible than satellite systems and are limited only by aircraft access and fuel capacity. Additionally, cloud interference can be avoided by positioning airborne systems below the cloud ceiling or by timing flights to avoid cloud cover (e.g., flying after fog has burned off in a coastal area). However, the marginal cost of mobilization for each image is higher for airborne systems than for satellite systems.
Extent—Depending on the resolving power of the sensor, high-altitude platforms will generally result in greater area imaged per exposure (i.e., larger extent), but at coarser spatial resolution than platforms operating at low altitudes. Airborne systems are usually more effective than satellite systems in collecting long and sinewy project areas.
Technical characteristics are not the only factors differentiating imagery types from one another. Often more important are the organizational characteristics, which will determine an image’s price, licensing, and accessibility. Choosing what imagery to use in a project requires making trade-offs between technical and organizational characteristics. In the next chapter, we will learn how to match imagery characteristics with user requirements to decide what type of