CIVIL LAND OBSERVATION SATELLITES

Introduction

The first, and still the most dramatic, example of seeing where we live from the vantage point of space was the image of a pale blue and white globe hanging all alone in an infinite expanse of darkness taken through the window of Apollo 8 on its way to the Moon in December 1968. Three and a half years later on 23 July 1972, NASA launched ERTS-1, later renamed Landsat-1. This was the first civil imaging satellite that had enough resolution, 80 meters, to image human-scale activities; that is everything bigger than a football field. Its more capable successors now provide anyone, anywhere, the ability to see images of motorcycle size objects any place on the globe at almost any time. As of November 2002, there are 23 satellites in orbit whose resolutions range from 30 meters down to 0.6 meter. They are being operated by the United States, France, India, Korea, Canada, China/Brazil, the European Space Agency, and three private corporations, two U.S. and one Israeli. The number of satellites and the number of their national and private sponsors show that civil land observation satellites have arrived at the point where they are now another permanent payoff of the space age. It will take much longer than the Weather, Communication and Global Positioning System (GPS) satellites for their full economic and social effects to be felt, but they are already forcing national and international discussions on the effects on nations, corporations, and individuals of the worldwide transparency that these satellites will inevitably bring (1).


The Big Picture

Figure 1 provides the orbital history of every satellite whose sensors can provide Earth images, both optical and radar, and whose resolutions are equal to and better than Landsat 1′s 80 meters. It shows quite clearly that there have been three distinct periods in the history of civil land imaging satellites. For 13 years,the United States’ five optical and one radar satellites were the only such systems in orbit that provided civil data. The second period started with the launches of Russia’s Resurs-01-1 in late 1985 and of France’s SPOT 1 early in 1986 and extends up to the first launch attempts by American commercial systems in 1997. During this period, there were 20 foreign launches and one failed U.S. attempt. The third period begins in 1997 when high-resolution commercial systems entered, includes NASA’s return to Landsat and research launches, and continues to the present day. The following discussion will provide some of the details of these periods to illuminate why they occurred as they did and will provide some sense of where the development and use of civil land imaging satellites might take us in the future. First, however, there are a few facts about satellite and sensor technology that will be helpful in understanding their development.
Orbital history of all optical and radar land imaging satellites that have Landsat or better resolution. This figure is available in full color at http://www.mrw. interscience.wiley.com/esst.
Figure 1. Orbital history of all optical and radar land imaging satellites that have Landsat or better resolution.

The Technology

This article discusses the special set of Earth orbiting satellites designed to acquire detailed images of the global land surface. It is the high-resolution subset of the larger family of Earth sensing satellites that image the land and oceans on a kilometer scale and measure the characteristics of the atmosphere to record the weather and to explore the complex interrelationships among the atmosphere, the oceans, and the land surface that cause the weather and our climate. The following describes the satellite, sensor, and data technologies involved. Satellites. Essentially all current and planned imaging satellites operate in sun-synchronous polar orbits and cover Earth about 14 times a day, as illustrated in Fig. 2. Sun-synchronous means that the satellite crosses the equator at the same time on every orbit. This characteristic means that all images have the same Sun angle and, therefore, shadow characteristics thus eliminating one interpretation variable. (The Sun angle changes as the seasons change.) The crossing times vary between 9 and 10:30 a.m. The current satellite orbits range between 460 and 904 kilometers; the lower range is chosen by balancing the need for a high enough orbit to ensure a long satellite lifetime and a lower orbit for higher resolution. The higher limit is a balance between the desire for the broad coverage area provided by high orbits and the higher resolution provided by lowering the orbit.
Sensors. Except for one Russian intelligence film based satellite (Spin-2) used for a commercial mission, all commercial optical sensors have been and are electronic imagers; they focus the light from the ground onto the telescope’s focal plane, either by sweeping it across a small line of sensors in a system called a whisk-broom scanner or by using the satellite’s movement along its orbit to sweep the image across a much larger set of sensors. This latter design, called a push-broom scanner, is the dominant sensor used today with one prominent exception: Landsat 7 has retained the whisk-broom design first flown on Landsat 1.
However, from the viewpoint of the uses to which the data are to be put, the choice of scanner type is not as critical as the four basic imaging characteristics of the total satellite system, that is, spatial resolution (how small an object can be perceived), scene size (how large an area is captured in each image), temporal resolution (how quickly a scene can be imaged again), and spectral resolution (the number and location of the spectrum bands measured). Each will be discussed in turn.
Typical Landsat daily ground trace.
Figure 2. Typical Landsat daily ground trace.
Spatial Resolution. The size of the imaged feature on the ground that the satellite can resolve is largely a function of the sensor’s telescope design and the satellite’s orbital altitude. Resolution is the most important of the three system characteristics for most users because it defines the scale and, therefore, the type of applications the system can serve. Scale is particularly important in making maps. Figure 3 provides a comparison of images of the U.S. Capitol taken at the resolutions of current satellites.
Scene Size. The width on the ground imaged by the sensor/satellite system is called its swath and serves as a surrogate for scene size since the along track image can be taken continuously and scenes are created for data management reasons by arbitrarily cutting the strips into lengths about equal to their swath width. Large scene sizes are essential for the repeat imaging of the regional and global areas that is required to study the changes in our land cover and land uses. In general, the higher the resolution, the smaller the swath. Temporal Resolution. For fixed nadir-pointing satellites like Landsat that can only image the land beneath their orbit, the time to return to image a given area is almost entirely a function of its swath width. The TIROS weather satellite and its 1000-meter resolution AVHRR imager has a ground swath of 2500 kilometers and images the entire globe daily. Landsat has a 185-kilometer swath and repeats its global cycle every 16 days.
For satellites that can point their sensor to either side of the ground track, the return visit time is a function of both the basic orbit repeat time and the range of the off angle pointing. SPOT uses a mirror to point the sensor and can provide a variety of scene location patterns (see Figure 4) as well as increase its 22 day nadir image repeat time to 3 or 4 days. The smaller higher resolution satellites point their sensor by tilting the spacecraft to achieve 2 to 3 day repeat cycles. However, because these systems have very small swath widths, between 8 and 16 kilometers, they would take 6 months or more to image the globe. Please keep in mind that the world’s average 50% cloud cover means that in practice getting repeat cloud-free images could take from 2 to 3 times longer than the days quoted above.
Comparison of image quality as a function of the resolutions of current satellites. Reproduced by the kind permission of Northrop Grumman Information Technology Inc. This figure is available in full color at http://www.mrw.interscience.wiley.com/esst.
Figure 3. Comparison of image quality as a function of the resolutions of current satellites. Reproduced by the kind permission of Northrop Grumman Information Technology Inc.
Example of acquisition optimization over France showing the impact of cloud cover forecasts and the satellite's mirror-pointing capabilities (each square is 60 km on the side). Reproduced by the kind permission of SPOT image. This figure is available in full color at http://www.mrw.interscience.wiley.com/esst.
Figure 4. Example of acquisition optimization over France showing the impact of cloud cover forecasts and the satellite’s mirror-pointing capabilities (each square is 60 km on the side). Reproduced by the kind permission of SPOT image.
 Imaging modes available from Radarsat.
Figure 5. Imaging modes available from Radarsat.
The location, pattern, and size of radar images are controlled by electronically programming the radar’s antennae and can easily be varied in a number of ways. This flexibility is illustrated in Fig. 5.
Spectral Measurement. The number and the placement of the spectral bands that are sensed are critical to the ability to classify surface features automatically, be they vegetable, mineral, or man-made. This feature of multispectral sensors has extended the use of images beyond their original mapping and object identification functions. The repeated classification of the continuing changes in forests, rangeland, and farmland at the state, country, and global scales, so important to understanding our environment, could not be accomplished by the eyes of photo interpreters alone. Figure 6 presents the number and location of the spectral bands for the major satellite classes. Figure 7 provides examples of the way Landsat band data can be combined in three band combinations to bring out various features in the scene.

The Road to Landsat

As noted before, the practical beginning of civil land imaging satellites was the 1972 launch of Landsat 1. But the process that preceded and finally resulted in that launch was long, complex, and acrimonious. It is important because its history illustrates many issues that are still not fully resolved to this day.
Humankind has probably recognized the military advantage of occupying the high ground since people first descended from the trees. As early as 1783, it was obvious to Benjamin Franklin, when, as Ambassador to France, he observed the first Montgolfier balloons in flight over Paris and predicted their use in ”conveying Intelligence” about ”an Enemy’s Army” (2). After the Wright brothers demonstrated flight, the incredibly rapid development of the airplane in World War 1 was due almost entirely to the importance of its unique reconnaissance capability. (The development of fighter aircraft, dogfights, and aces resulted when each side tried to deny the enemy the use of air reconnaissance.) Bombers dominated aircraft use in the Second World War, but reconnaissance was still a major air function, and infrared imaging was developed out of the need to identify camouflage. The cold war caused the United States to develop the fastest and highest flying aircraft in the world for the sole purpose of photoreconnaissance.
Number and location of the spectral bands for several satellites. This figure is available in full color at http://www.mrw.interscience.wiley.com/esst.

Color images generated by assigning different TM bands to the blue, green and red guns of the color monitor. Provided by Stacy Bunin and Mitretek Systems. This figure is available in full color at http://www.mrw.interscience.wiley.com/esst.

Figure 6. Number and location of the spectral bands for several satellites.

Figure 7. Color images generated by assigning different TM bands to the blue, green and red guns of the color monitor. Provided by Stacy Bunin and Mitretek Systems. This figure is available in full color at http://www.mrw.interscience.wiley.com/esst.
Although the Soviet launch of Sputnik in October 1957 shocked the American public, the only surprise to the science and military communities was that the Soviets did it first. In 1955, both the Soviets and the United States had announced plans to orbit an Earth satellite to support the 1957/58 International Geophysical Year (IGY). The American announcement mentioned the beginning of the ill-fated Vanguard Program that President Eisenhower, in his press release congratulating the Soviets scientists on Sputnik I, pointedly noted was separate from our military launch vehicle programs. President Eisenhower’s private response to Sputnik was to increase the funding for an Air Force program first organized in 1953 that, after 12 failed attempts, resulted in the successful launch and recovery of the film carrying capsule of Discoverer XIV on 30 August 1959. The images were not as clear as the U-2 aircraft photos but covered Soviet areas never seen by spy planes. The purpose of the program was kept top secret until 1995 when President Clinton declassified it under its proper name of Corona. (See the article on the Military Use of Space in the topic for more details.)
The Russians soon flew their own spy satellites, and in a kind of bilateral ”don’t ask, don’t tell” spirit, the Russians finally, but tacitly, accepted Eisenhower’s ”Open Skies” policy. This policy actually followed the aircraft reconnaissance focused Open Skies policy and is more accurately called the ”satellite data nondiscriminatory dissemination policy.” It supports the freedom of all satellites to image the surface of any and all countries by assuring access on an equitable basis by the imaged country to all images of its territory. (See discussion in the ”Getting Data to Users” section.)
There were, of course, prophets. In 1946, well before Sputnik, the United States Army Air Corps requested that RAND Corporation consider how objects might be inserted into orbit (3). The study resulted in a report, Preliminary Design of an Experimental World-Circling Spaceship (4). The proposed midget moon, or ”satellite,” would provide an observation aircraft [sic] which cannot be brought down by an enemy who has not mastered similar techniques.” The public sector had its own prophet. In 1951, 6 years before Sputnik 1, Arthur Clarke, a science fiction writer, proposed that a satellite could be inserted into orbit over the North and South Poles while Earth revolved beneath it, and that this satellite would permit humans to view the planet in its entirety (5). (Clarke is also credited with being the first to propose using satellites for communications.)
The civil satellite potential was not neglected. In April 1960, the National Aeronautics and Space Administration (NASA) and the Department of Defense (DOD) launched the Television and Infrared Observational Satellite (TIROS-I) into a polar orbit, inaugurating the first experimental weather satellite. This system generated the first television-like pictures of the entire globe in a systematic and repetitive manner. This ongoing series of TIROS satellites became operational in 1966 as the TIROS Operational Satellites (TOS) and in 1970, the National Oceanic and Atmospheric Administration (NOAA) renamed them the Polar Orbiting Environmental Satellites (POES). (For a full account, see the Weather Satellites in this topic.)
In spite of the deep secrecy of the Discoverer Program, the military early recognized the need to involve the scientific community in developing techniques for using satellite data. In February 1962, at the Navy-sponsored First Symposium on Remote Sensing of the Environment, Dana Parker, one of the organizers, focused his inaugural address on the fundamentals of the electromagnetic spectrum, a subject that suggested that multispectral data could provide information that went beyond the ”eyeball” analysis of photographic data on which all reconnaissance analysis was based. His talk heralded the multispectral interpretive approaches responsible for the selection of the Multi Spectral Scanner (MSS) flown on Landsat 1 and the current capability to analyze continental and global scale Earth cover changes (5).
The person who was probably the most influential in realizing and promoting the value of multispectral sensing was Robert N. Colwell, a professor of forestry at the University of California at Berkley. In the 1950s, Colwell and others showed that agricultural crops, trees, and even different soils possess a telltale electromagnetic signature that could be measured from any height and used to identify the object. He and his American colleagues were not alone in such thoughts. A Soviet scientist named E. L. Krinov had measured the spectral signatures of some 370 natural and man-made objects in the 1940s, and he, too, had aerial mapping in mind. In a prescient article in American Scientist in 1961, Colwell wrote, ”Just as our musical appreciation is increased greatly when more than 1 or 2 octaves are exploited, so also is our appreciation of the physical universe through multiband spectral reconnaissance, which already can exploit more than forty ‘octaves’.” He and his colleagues at Michigan and Purdue were instrumental in the selection of the MSS launched on Landsat 1 (2). Colwell’s students were pivotal in making the rapid development of multispectral sensing possible.

Getting Landsat Started

The concept of a dedicated, unmanned land-observing satellite emerged in the mid-1960s. It was stimulated by the early interest of the ONR, and aided by NASA’s studies and aircraft sensor experiments starting in 1963 in its Office of Manned Space Flight’s Earth-Orbital Apollo Extension System under P. C. Bad-gley. These efforts bore direct fruit in Earth-sensing experiments on Skylab in the mid-1970s, but were more important to the Landsat decision because Bad-gley also funded the U.S. Geological Survey (USGS) and the U.S. Department of Agriculture (USDA) to define practical uses for space-derived images (6). By 1996, Badgley, then working in NASA’s Office of Space Science and Applications, and Leonard Jaffe, in charge of its Application programs, were well aware of the desire of the USGS and USDA for the speedy development of an Earth resources satellite, but NASA upper management showed little inclination to address the desires of the potential user agencies. What attention they were giving to Earth application needs was centered on planning to use the Apollo spacecraft as a manned Earth observation system. (This should not be surprising. At this same time, the U.S. Air Force, in spite of the already very active unmanned observation program, was in the first stages of the Manned Orbiting Laboratory (MOL) program which would use the Gemini spacecraft to support a space station filled with both optical and radar sensors. MacNamara canceled it probably because of its rising costs, but also because there was no clear argument that the presence of a crew in orbit would add value to the reconnaissance data. He was also aware that the loss of a manned spy satellite due to enemy action would require a much more dangerous response on our part than the elimination of a simple satellite.)
While NASA’s manned program developed and tested some of the relevant technology and the startling photographs from Gemini and later Apollo gave scientists insight into how valuable satellite imagery could be (7), the real credit for the Landsat satellite must go to the USGS and in particular to its then director, Dr. William T. Pecora. Disturbed by the continuing lack of progress by NASA, Pecora set up a press conference on 21 September 1966 at which his boss, Interior Secretary Stewart Udall, announced the Interior’s plans for an Earth Resources Observation Satellite program (EROS) to be run in cooperation with NASA. James Webb, the NASA Administrator, made aware of the press conference on the previous day, arranged an immediate meeting with President Johnson, who reaffirmed that the development of space technology was NASA’s responsibility. Secretary Udall was informed by NASA on the day after the press release that NASA had been assigned ”lead agency” responsibility for civil experimental space applications. The USGS announcement had stressed its development of an operational system, but NASA managed to maintain that any satellite system, even if it carried sensors considered ready for operational use, was experimental (6). This has been governmental and NASA policy to this day, though it has several times required some imaginative interpretation of “experimental.”
Although unsuccessful with its own “operational” satellite system, the Department of the Interior (DOI) continued to press NASA for a satellite. The overall goal of the proposed EROS program was to acquire remotely sensed data from satellites in the simplest possible way, deliver these data to the user in an uncomplicated form, and ensure their easy use (8). NASA did not believe that the problem was technically that simple and was especially negative about the schedule that the USGS was proposing. DOI seemed to have lost the battle, but the result was that NASA accelerated its planning for a small unmanned satellite.
However, it was not to be that easy. NASA was initially frustrated by the Bureau of the Budget (BOB) that rejected NASA’s FY 1968 budget. In the fall of 1967, NASA and the DOI budget request for FY 1969 funding for the project was again turned down. It took a direct appeal by NASA Administrator James Webb to President Johnson to have the request restored and a rescheduling of the elements in that budget by Congress to allow NASA to start on the initial contractor studies for the satellite. DOI’s budget request was still refused, which caused a delay in its plans to initiate a data processing and distribution facility. This delay would cause problems in getting Landsat 1 data to user communities for several years after the launch.
Though NASA finally was given its FY1970 budget request as well, the BOB was still fighting hard to stop the program; one internal BOB memo written in June 1968, while the FY 1970 request was being developed, suggested that the Landsat satellite proposal should be eliminated and replaced with an Earth resources aircraft program (6). Because the EROS satellite was being developed to prove that satellites images could be beneficial to a wide range of governmental and public activities, the BOB had required NASA to make a cost-benefit analysis of its potential value. The argument that aircraft could do the same jobs better and at less cost was an issue that plagued Landsat then and later as it presented its plans for the follow-on Landsats. Amron H. Katz, in an article in Astronautics and Aeronautics, made this argument most forcibly in June 1969, and he was still arguing the superiority of aircraft through 1976, long after the launch of Landsat 1 (6). At this time, Katz’s argument is in the process of being tested in the marketplace by the recently launched commercial satellites whose 0.6 to 1-meter resolutions can compete directly with many (but not all) of the imaging tasks performed by aircraft-based sensors.
Interior changed the “S” in its Earth Resources Observation Satellite (EROS) program from Satellite to System and placed it under the direction of the USGS. The EROS mission was to archive and distribute remotely-sensed data and to support remote sensing research and applications development within the DOI. To carry out the EROS responsibilities, the USGS built the EROS Data
Center in Sioux Falls, South Dakota, in 1972. This location was chosen after a competition with a location in Mississippi. Both were being considered because their central locations would enable a receiving antenna to record satellite passes over both the East and West Coasts. The receiving antenna was finally installed at the EROS Data Center (EDC) in 1998, the delay being due to the commercialization of the program since the contractor chose Norman Oklahoma as the location for its receiving station.

The Multispectral Scanner Years, 1972 to 1984

Table 1 provides the complete history of the Landsat system, its sensor complements, and operational periods.
Landsat 1 carried two sensors, the return beam vidicon (RBV) and the MSS. The RBV was a television camera designed for cartographic applications and was the sensor of choice of the USGS and originally NASA. The MSS was designed to identify natural features using spectral analysis and came to NASA in an unsolicited bid from a group at Hughes Aircraft headed by Virginia Norwood. The Hughes scanner triggered quite a debate. ”Mapmakers like myself were very suspicious of the multispectral scanner, which we could not believe would have geometric integrity,” admitted the USGS’s Alden Calvocoresses. ”We were wrong on that one.” ”People were so emotionally opposed to a mechanical device,” Norwood recalls. The argument which went on for more than a year, was resolved by flying both. However, the six-band sensor originally proposed by Hughes had to be reduced to a four-band system because of weight restrictions. In retrospect, the decision was to prove a crucial one. Soon after the launch, the TV cameras became afflicted with an unexplained electrical problem and had to be shut down within the month. The scanner, designed to work for 1 year, was still functioning, moving mirror and all, when the satellite was turned off 6 years later. Figure 8′s schematic of the MSS shows how the mirror scans across the track while the satellite motion provides the other dimension. Figure 9 shows the whole satellite with the locations of the MSS and RBV noted.
The launch of Landsat 1 in the summer of 1972 settled the issue. The quality of the images, including their geometric fidelity provided by the multi-spectral scanner (MSS), was a surprise to many in the community and was quickly followed by the recognition that its data scale, resolution, and four color bands provided new, unique, and very useful ways to see and understand our geography.

Table 1. History of the Landsat Satellites

Satellite Launched Status as of July 2002 Sensors
Landsat 1 7/23/72 Ended ops 1/06/78 MSS & RBV
Landsat 2 1/22/75 Ended ops 2/25/82 MSS & RBV
Landsat 3 3/05/78 Ended ops 3/31/83 MSS & RBVa
Landsat 4 7/16/82 Standby MSS & TM
Landsat 5 3/01/84 Operational MSS & TM
Landsat 6 10/05/93 Launch failure ETM
Landsat 7 5/15/99 Operational ETM +

Scanning geometry of the multispectral scanner
Figure 8. Scanning geometry of the multispectral scanner
The satellite data were received on Earth in digital form. However, nearly all of the initial MSS products were provided as color and black and white prints and negatives because the user community had little experience with or training in analyzing data in its digital form. In addition, their computer facilities were only marginally up to the job of processing the amount of data involved, and the interpretive algorithms were still in an early developmental stage. There was a cost difference that certainly had some effect in the later years; film costs varied between 8 and 15% of the cost of digital tapes. Film dominated the product sales in dollars until 1984, but even then 35,000 film scenes were delivered compared to 5,000 digital scenes. Digital sales increased rapidly after that until, in 1993, EOSAT eliminated its photographic product line because it was not economically viable (9). Finally, 21 years after Landsat 1, the digital applications made possible by the MSS, had worked their way into the marketplace. Today, all of the 24,000 annual sales are in the digital format.
The quality of the photographic images was more than sufficient to inspire a rush to explore their uses for many applications. For the first time, mapmakers had images that covered 100 miles on a side with near orthopho-tographic geometry. (The edge of the scene was imaged at only 7^° from the vertical. Figure 10 shows just how impressive this capability is. It is one Landsat scene covering 185 km x 170 km, virtually 100 miles on the side, showing the Chesapeake and surrounding land, including Baltimore and Washington, D.C. Aircraft montages for that same area would involve hundreds of individual images, each of which would have had to be corrected to match their 30 or 40° edge angles with their neighbors.) Geologists were able to recognize innumerable new fault lines revealed by the consistency of the lighting and shadow angle across the whole scene. There was even a minor coffee-table topic rush to present the public with pictures of both familiar and far away places made strange and beautiful by the many colors generated by manipulating the four color bands.
Platform configuration, Landsats 1, 2, 3.
Figure 9. Platform configuration, Landsats 1, 2, 3.
The first sizable use by industry was by the major mineral and oil exploration companies. They were targeted early in the program by JSC’s Earth Resources program office and, after some initial skepticism, formed Landsat analysis groups that used Landsat images to update maps and to identify surface features that warranted further exploration. It was equally important for them to be able to obtain images anyplace on the globe without their rivals knowing about it. (Early in the program, EDC established the policy of keeping the data buyers’ identities private.) The industry put together an advocacy group, the GEOSAT Committee, that served as its representative in the many Landsat debates that were to be a feature of the Landsat program over the years. Robert Porter, who resigned his position as Director of the NASA Earth Resources Program in 1970 to found his own company, Earth Satellite Corporation, to apply Landsat analysis to commercial needs, estimated that at least $1 billion worth of oil had been found using Landsat data (6).
100 mile square Landsat image of Chesapeake Bay and environs. This figure is available in full color at http://www.mrw.interscience.wiley.com/esst.
Figure 10. 100 mile square Landsat image of Chesapeake Bay and environs.
The boost this application gave to those in NASA who defended the Landsat program would be short-lived; the exploration companies generally needed only one look. Landsat covers the whole globe 24 times a year, every year, and its ”killer application” must be one that needs that kind of repetitive coverage. The management of things that grow, trees, grass, and food, seemed to fit that bill and were, of course, the intended target of the multispectral analysis for which the MSS was designed.

Transferring Landsat Technology

NASA’s Office of Space Applications (OSA) was split from the Office of Space Science and Applications (OSSA) in 1971 and retained the Landsat program that had been transferred to OSSA from the Office of Manned Space Flight in 1966. OSA also developed the first communication satellite, the first weather sensors and satellites, both polar and geosynchronous and the first and still the only U.S. civil radar satellite, Seasat (See Radar section). However, it recognized from the start that its function went beyond the development of satellites and included the development of the required ground systems and data analysis technologies that would enable users to apply the satellite data to their problems. OSA’s success in these basically technical problems can be rated excellent for the satellites and sensors and after a very slow start, good for the ground systems and analysis algorithms. But it soon was forced into the social and political engineering role of technology transfer. OSA’s results in this function are still being debated. The best study of this activity and its results for the period up to 1990 is Pamela Mack’s ‘Viewing the Earth: The Social Construction of the Landsat Satellite System’ (6), which has been the source of many of the facts in this article. As will be noted later in this article, the social and political issues concerning Landsat and its successors became even more complex after 1990.
NASA’s initial approach to the technology transfer issue was to fund studies in universities and user agencies. The university studies were most often concerned with developing and testing analytical techniques to exploit the manipulation of the signals from the four bands to identify crops and to measure their yield and health. The user agencies were concerned about defining operational and land management functions that could be done better, and more economically using Landsat images, or new but needed applications that could not be done at all without Landsat. NASA requested proposals even before Landsat 1′s first launch. Expecting 40 or so principal investigators, it was inundated by more than 500 proposals. Wanting to encourage the broadest use possible, it set up peer review panels in several disciplines (chiefly geology, hydrology, and geography). The panels chose more than 300 investigators (6).
The large number of investigators made NASA carefully consider what the data distribution policy should be. Because there was the possibility of using an image to find value on someone else’s property, NASA decided to release all data to the public immediately to avoid charges of unfair discrimination. Data distribution policy is a crucial issue still facing the civil program, involving as it does the complex relationship between the commercial interests that require exclusive distribution to earn their profits and the government’s desire to provide data to scientists and government agencies at prices commensurate with public good” support.
The results of these and following studies were presented in a series of symposia hosted by GSFC starting in March 1973. The presenters were asked to report on ”user identified significant results” in agriculture/forestry, environment, geology, land use/land cover, and water. Each of the Proceedings approached 2000 pages. The studies produced valuable scientific results from the time of the first launch, but they also revealed that some of the capabilities promised by the magic of multispectral analysis fell a bit short when faced by the heterogeneity of the real world and the radiometric and resolution limitations of the MSS. Though many crops could be identified, the accuracy of the identification varied widely by crop type, growth status, atmospheric conditions, and field size. There was much to learn, but the reports demonstrated the value of multispectral analysis for a wide variety of surface analyses. These analyses were also able to identify the limitations on analysis imposed by the limited resolution and number of spectral bands of the MSS, thus giving support to the need to develop a better scanner. The development of the new scanner, the Thematic Mapper, is described below.

The Large Area Crop Identification Experiment (LACIE)

The NASA technology transfer effort did not rely only on the numerous individual studies that NASA continued to fund throughout this first period. In the 1970 and 1971 growing seasons, the Johnson Space Center (JSC), the USDA, and Purdue University used aircraft equipped with a variety of sensors, including a multispectral scanner that simulated the MSS, to fly over 210 test areas in seven Corn Belt states. The program detected the corn blight early enough to check it by pesticides and, as a result, led to the operational use of low flying aircraft to solve the farmers’ problem (6). It did not provide Landsat with a direct application, but it did provide valuable experience in dealing with the problems of crop identification and health detection in the real world, which were important in developing NASA’s Large Area Crop Inventory Experiment.
In July 1973, Congress asked NASA why it had failed to do more to predict the failed Russian wheat crop of 1972. Congress was concerned because American farmers lost potential profit by being unaware of the condition of the Russian crop. NASA saw an opportunity to reply with a plan when Robert MacDonald, who had led the corn blight team at Purdue and was then the Landsat Chief Scientist at JSC, suggested the large-scale agricultural experiment that later became LACIE. LACIE was certainly large scale; it monitored the entire Russian wheat crop and the U.S. Great Plains for three years, involved a contractor force of more than 200 Lockheed analysts, and cost between $10 and $15 million a year. In 1997, its third year, it met its goal of 90% accuracy, successfully predicting the poor Russian wheat harvests in that year. But as a technology transfer mechanism, it failed. USDA, in response to the direction of the OMB (see below) formed the ArgriStar program in close cooperation with NASA, but chose to focus it on improving and testing the techniques of crop identification rather than continuing the country production estimates of the LACIE program. Over the many years since, it has developed new and different uses of Landsat data and has been the largest civil agency Landsat data purchaser.

The End of NASA Application Activities

LACIE’s last year was 1997, as it was for most of OSA’s Landsat technology transfer programs. The Office of Management and Budget (OMB, the new name for the BOB) informed NASA that it would no longer support NASA activities that benefited other agencies on the theory that if they were really beneficial to those agencies’ functions, the agencies should budget for them. NASA was being told to return to its experimental and scientific mission. This was part of the commercialization of the Landsat drive discussed in the next section. By the end of 1978, OSA had virtually ceased all application efforts, replaced its application managers with science managers, and embarked on studies to define the scientific requirements for Earth observation satellites. This effort resulted in the creation of the Office of Earth Sciences and the development of the Earth Observation Systems (EOS) program designed specifically to support the interagency Global Change program. Landsat was eventually to be considered part of this program, but it received little management attention, which resulted in little science or application research funding during Landsat’s commercial phase. NASA did, however, continue its development of the Thematic Mapper throughout this period, as befitted its technology development role (11).

The Thematic Mapper

As noted before, during the 1972 to 1978 period, NASA developed a new and improved multispectral sensor, named the Thematic Mapper (the TM) to emphasize that it was designed to measure and map the basic features or themes of the earth’s surface automatically, that is, forests, plains, rivers, lakes, cities, farms, etc. Originally the sensor designers at GSFC reacted to the success of the MSS by proposing two alternate improvements, a 10-meter high-resolution pointable imager (HRPI) and a 30-meter resolution TM. HRPI was to be a small sensor using a new technology, multilinear array sensors, the scanner technology adopted by the French for their SPOT system in 1978 and for virtually all (except Landsats 6 and 7) land imaging sensors since then. HRPI could be programmed from the ground to point to targets along or to the sides of the flight path. This pointability provided 2 to 3 day repeat looks at a given target. It also made it possible to generate stereo images. However, high resolution came at the price of a small field of view, 10 kilometers. The decision between the two was a fairly easy one for NASA because the scientists liked the large area coverage of the MSS and DOD made it plain that 10 meters exceeded their threshold for civil system resolution. It is interesting to speculate on the path remote sensing would have taken if HRPI had been chosen. HRPI had the pointability and higher resolution that have been the capabilities that finally emboldened the private sector to fund commercial systems in the mid-1990s.
The TM was selected because its 185-kilometer wide, 16-day global repeat coverage replicated that of the MSS, a very important requirement of the science community and user agencies. The user agencies and the scientific community were called upon to determine the exact spectral bands for the TM. The consultation process culminated in a 3-day conference at Purdue under the direction of Dr. David Landgrebe, the director of their remote sensing group. Three teams, each composed of sensor designers, data analysis experts, and data users, debated the merits of all possible band combinations. There was surprising consensus among the three teams, and the selection of six bands was essentially unanimous (12). Much later in the sensor development process, the geological community led by Alex Goetz, feeling they had been slighted in the band selection process because of NASA’s focus on agricultural users, requested that a band he had shown was sensitive to minerals, be added. Fortunately, the addition was found technically feasible, and a seventh band was added. The increased capabilities of the TM came with significant weight and size increases over the MSS. The TM weighed 258 kg versus the MSS’ 64 kg and it measured 1.1 x 0.7 x 2.5 meters compared to the MSS’ 0.5 x 0.6 x 1.3 meters. Figure 11 presents a picture and cutaway of the TM.
The user community also was able to convince the OMB that any satellite carrying an experimental sensor should also carry the MSS to allow testing the new sensor relative to the established and understood characteristics of the MSS. While this was a scientifically reasonable position, the user agencies were also worried that they would have budget problems upgrading their computer systems to meet the greatly increased data volume generated by the TM. Thus, Landsats 4 and 5 came into being. Figure 12 presents a drawing of Landsats 4 and 5.
Landsat-7 ETM + (note dimensions are 1.8 x 7 x 2 meters). This figure is available in full color at http://www.mrw.interscience.wiley.com/esst.
Figure 11. Landsat-7 ETM + (note dimensions are 1.8 x 7 x 2 meters).
Landsat 4 and 5 observatory configuration.
Figure 12. Landsat 4 and 5 observatory configuration.

Commercialization 1984-1992

When the TM on Landsat 4 was launched in 1982, NASA could declare Landsat a technical success. Landsat was also an international success as the look-alike programs of France, India, Russia, and Japan were soon to demonstrate. However, as is the fabled fate of prophets in their own land, NASA failed to convince any other agency or any commercial entity to develop its own system. The user agencies, which were vociferous in telling NASA what they needed to make the program better and, above all, their need for it to be operational, were very cautious in their testimony to Congress on the value of Landsat, lest they be asked to use part of their budget to build the next system.
The basic issues had been there from the start: should there be a civil operational Landsat and which agency should fund and manage it? The large benefit-cost ratios derived from the BOB-requested economic cost studies and used by NASA to justify the continuation of the program, inadvertently added a third issue: should the system be commercialized? In June 1978, the Office of Science and Technology Policy published the results of a study of policy options for civilian remote-sensing satellites. The study concluded that the President should make a commitment to an operational Earth resources satellite system and suggested that NASA have responsibility for that system (6). User agencies and the states disagreed that NASA be the manager. In November 1979, President Carter issued Presidential Directive 54, which gave NOAA temporary responsibility for operational Earth satellites and called for the National Oceanic and Atmospheric Administration (NOAA) to develop a plan for turning over Earth resources satellites to private industry. The decision to have an operational remote sensing system and to have it managed by NOAA along with its management of weather satellites was roundly applauded. However, the suggestion that it should later be turned over to private industry sparked a strong controversy that has continued to this day. A 1980 report by a task force of state representatives explained that ”the establishment and operation of a land remote sensing satellite system is viewed as a public service in the same context as census, cartographic, geological and meteorological data which are supplied by the Federal government” (Recommendations of the National Governor’s Association April 30th 1980). As required by the directive, NASA turned over the operations of Landsat 4 in 1982 after assuring that it was functioning as designed.
Carter’s steps toward gradual privatization were quickly trumped by Reagan whose administration moved rapidly to transfer the Landsat program to the private sector. Reagan attempted to include the weather satellites in this process but strong objections from Congress and a scandal concerning questionable lobbying and contact between COMSAT (which lobbied strongly because the privatization mechanism was based on its successful commercialization of communication satellites) and the Department of Commerce resulted in PL 98-52 which prohibited the transfer of weather satellites to the private sector. Reagan signed the final law, Public Law 98-365, the Land Remote-Sensing Commercialization Act of 1984.
In January 1984, NOAA solicited bids to operate the existing Landsats, retrieve and develop a market for their images and, aided by government subsidies, to build and operate future systems. Both of the two final bidders, EOSAT a partnership between Hughes, the MSS contractor, and RCA the satellite contractor, and Kodak, required government subsidies of about $500 million spread out over a 6 to 10-year period. This was a reasonable amount based on several NOAA-funded studies that agreed it would take at least 10 years before the sales of Landsat images could support the development of a satellite. OMB did not want to supply federal funds. Reagan’s chief of staff split the difference, and the contract was limited to $250 million of government funding. Kodak withdrew, and EOSAT was selected. However, the cost cap kept contract negotiations going until both parties agreed to a final amount of $350 million, and the contract was signed in September 1985.
With that signing, the Landsat community breathed a sigh of relief because it seemed that Landsat was finally on a course toward the long desired goal of operational status. This was short-lived, however, as the impact of the EOSAT 200% price increases and their restrictive licensing requirements became felt. This was especially critical for the academic community and the state users who were used to buying the data at almost the cost of reproduction. The result was that during the commercial period, studies to develop and test new uses of multispectral analysis using Landsat scenes almost came to halt. Most of the major advances in spectral analysis of satellite imagery in this period were made using the free 1-kilometer resolution data from the advanced very high resolution radiometer (AVHRR) on the NOAA weather satellites.
The Reagan administration continued its campaign to reduce spending on Landsat even after the contract was signed; this made it virtually impossible for EOSAT to continue its marketing efforts and to plan for the follow-on satellite. In 1986, the French government launched SPOT1, designed specifically as a commercial data gathering system. EOSAT was faced with selling 30-meter data against SPOT’s 10-meter panchromatic and 20-meter color data. In 1991, Land-sat’s sales were $32 million and SPOT sales topped $40 million. Note that neither number approached a level that would support a fully commercially financed system.
The first positive boost to the program occurred early in 1989 after NOAA directed EOSAT to turn off Landsats 4 and 5 due to lack of program funds. The newly elected Bush Administration was deluged with hundreds of letters from academic researchers, foreign governments, private companies, and concerned congresspersons demanding the necessary funding. Vice President Quayle, head of the newly formed National Space Council (NRC), canvassed the major user agencies and put together an ad hoc funding plan that kept the satellites in orbit. Unfortunately, there was still division in the administration, and the ad hoc process had to be invoked for the next 2 years. Landsat did start to gain backing in the government due to three factors: Bush’s Global Environmental Change initiative, the utility of Landsat and SPOT data during the Gulf War; and the technological impact of computer capabilities, especially Geographic Information Systems (GIS) and the satellite-based Global Positioning System (GPS). The first had impact because Landsat was the only consistent set of global data extending back to 1972, and the second because it caused the powerful military and intelligence communities to state openly the importance of Landsat for their missions. (By late 1980, the DOD was the single largest user of Landsat data in the world.) The third factor finally allowed science and application users to take full advantage of the digital data on affordable computers. The forces for change were set in motion.

Landsat Returns to the Government in 1992

Landsat commercialization was a failure. That is not just an opinion; Congress so declared it in the findings of the Land Remote Sensing Policy Act of 1992. Passed in October of that year, it was the result of a congressional process that was started by Representative George Brown and Senator Larry Pressler. Brown, Chair of the House Science and Technology Committee, was a dedicated environmentalist, and from that viewpoint, he declared that Landsat should be returned to the government to be treated as the weather satellites, as a public good (13). He introduced a bill in the house for the purposes of “Amending the Land Remote Sensing Commercialization Act of 1984…” and Pressler offered a similar bill in the Senate.
The Land Remote Sensing Policy Act of 1992 was a farsighted and fruitful act. It created the commercial licensing process that culminated in the current leadership of the United State in the high-resolution commercial market. It recognized the importance to this country’s scientific, resource management, security and economic sectors of continuing the then 20-year recording of the global land surface by the midresolution Landsat series and established “continuity” as the criterion for Landsat follow-on systems. That somewhat loosely defined criterion resulted in Landsat 7 and is the basis of the current plans for its successor, the Landsat Data Continuity Mission (LDCM). The law also mandated that the data be provided to everyone for the ”cost of fulfilling user requests,” responding to Chairman Brown’s desire to make Landsat data a public good. It gave the DOI the responsibility for creating and maintaining an archive of land image data, a responsibility that the DOI assigned to the USGS. Subsequently, the USGS entered into an interagency agreement with NASA to operate Landsat 7 and its data collection, archiving, and distribution functions. The law recognized those desiring commercialization by giving the DOC, appropriately consulting with the DOD, CIA, and the NRC, the task of licensing private companies to fly Earth-sensing satellites.
The administration was not idle while Congress debated. In March 1992, NASA and DOD released the management plan that Bush had requested. The plan called for the development of Landsat 7 to replace the EOSAT-built Landsat 6 at the end of its expected life and to continue the ”continuity” of the Landsat data series. The plan assigned the responsibility of the space segment for Landsat 7 to the DOD and the ground segment to NASA and provided a budget of $470 million for DOD and $410 million for NASA. After so many years of cross-purposes, it seemed that the Bush administration and Congress agreed at last on the role of civil remote sensing satellites.
The real breakthrough for the commercial satellite companies occurred as the result of a joint meeting of the House committees on Science and Intelligence in 1994. One company, Earth Watch (now Digital Globe), quickly obtained a license for a 3-meter resolution satellite. (Unfortunately, it failed soon after launch in 1997.) However, at the time of hearing, Lockheed had been waiting 6 months for an answer to its request for a 1-meter system. Congress wanted to know why. The testimony of all of the agencies was that the availability of 1-meter images to the world might pose a serious threat to our security and required more study. Congress suggested that they reevaluate their concerns in the light of the 2-meter satellite scene of the Mall that had been purchased on the commercial market from Russia. Two weeks later, President Clinton issued Presidential Decision Directive (PDD) 23 that authorized commercial satellite high-resolution imagery and set specific regulatory guidelines. It left the definition of high resolution to the decision of the agencies involved, but its intent was clear to all, and Lockheed had its 1-meter license. At the present time, two companies have licenses for systems of 0.6-meter resolution and one is already operational.
Though the DOD overplayed its hand a bit, it did have legitimate concerns about the availability of such data in time of war. The license contains a ”shutter control” clause for just that situation. However, shutter control’s legality is being questioned on the basis that it is prior constraint and not allowed by the Constitution. During the Afghanistan engagement, the National Imaging and Mapping Agency (NIMA) finessed the prior constraint issue by contracting with Spacelmage for exclusive use of all the data taken over Afghanistan during the critical months of the operation. Most of the data were made available to all 2 months later.

Landsat Returns to NASA

Hardly a year passed before the Landsat program was evaluated for a third time, principally because NASA felt that its budget would not cover building the extra capabilities in the ground data system that would be required by DOD’s addition of a high-resolution (5 meters) multispectral stereo imager (HRMSI). The National Science and Technology Council (NSTC) meeting in response to the launch failure of Landsat 6 and aware of DOD’s reluctance to continue a program without a HRMSI sensor, recommended developing Landsat 7 with only an improved TM instrument and establishing a new management structure, so that DOD could withdraw from the program. This resulted in Presidential Decision Directive/NSTC-3, dated 5 May 1994, reconfirming the Administration’s support for the program but giving NASA, NOAA, and the USGS joint management responsibility (White House, 1994). The Landsat Program Team (the LPM) was created by the three agencies. It proceeded to negotiate with EOSAT for new Landsat 4 and 5 product prices for the U.S. government and its affiliated users and began the process of defining and developing Landsat 7.
A project office was established at GSFC. It defined a new sensor, the ETM + (extended TM +), that replicated the TM bands but also included several improvements made possible by advances in technology since the earlier Land-sats. The improvements included adding a 15-meter panchromatic channel, sharpening the thermal band to 60 meters, and increasing sensor sensitivity and calibration accuracy. It also defined a data acquisition goal. The system was to be able to acquire the entire landmass of the globe on the average of four times a year. Thus, for the first time in Landsat history, the satellite was to be programmed to do what its supporters had long envisioned: acquire a scientifically useful annual archive of the seasonal and annual changes that take place on all of Earth’s land surface.
Landsat 7 was launched on 15 April 1999. Its data have met all expectations. Landsat 7 is the first and still the only midresolution system programmed to acquire and archive all of Earth’s land surfaces once per season. The data acquisitions are being collected under the guidance of a program developed by Dr. Sam Goward’s team at the University of Maryland. The program maximizes the probability of observing critical seasonal changes and reduces the number of cloud-filled scenes acquired from the system’s capability of 250 scenes per day. This is not a trivial task because clouds cover 50% of the globe, on average, and some of the most biologically active areas are covered 80% or more of the time. (For the past 2 years, scientists have been able to gather data at twice the frequency of the 16-day Landsat repeat orbit because Landsat 5 is still functional in its eighteenth year of operations.)
Equally important, especially to the global environmental change science community, as required by the 1992 law, all of the data are archived at EDC and are sold to everyone at the cost of fulfilling user requests, currently about $600 per scene for the standard product. A survey of EDC image sales for the first 10 months of Landsat 7 operations showed that sales averaged 2400 a month. An analysis of the 1000 separate purchasers revealed that when account is taken of the many commercial and academic buyers who purchase data for use on government programs, the government is the ultimate purchaser of more than 80% of the data sold. This supports the observation in the 1992 Law that Landsat data are effectively a public good.

The Landsat Data Continuity Mission (LDCM)

Goddard created a Landsat Data Continuity Mission (LDCM) team with the USGS in 2000 (NOAA had left the Landsat Program Team earlier because it did not get its Landsat budget) to formulate plans for a follow-on mission to Landsat 7. In response to Congressional mandates that NASA not develop any satellite until it could demonstrate that it could not buy similar data commercially, GSFC first developed a rigorous set of data specifications that would meet the requirement for data continuity: 30-meter multispectral resolution, 15-meter panchromatic, 185-kilometer swath, data quality equal to or better than Landsat 7, and delivery of an average of 250 scenes a day for 5 years. After several iterations with the science and application community, the resulting data specifications were made the basis of an RFP to industry for a 6-month study to provide several trade-off studies to NASA, including the cost of keeping the thermal data requirement. The contractors were also requested to provide the design of a total satellite, sensor, and ground system to the level of detail required for a traditional Program Design Review. Two contractors were chosen. Both bids were for systems designed to supply a commercial market that required much higher capability than NASA was requesting, but whose data could be degraded to meet the government’s requirement exactly. It is just possible that a formula has been found that will meet the requirements of both sides of the commercialization-public good debate. NASA will get a product that meets its scientists’ goals without having to remain in the satellite operations business, Landsat type data will be provided to all at the cost of reproduction, and industry will get the help it needs to develop new products that will broaden and add to its nascent high-resolution business.

Getting Data to Users

The Landsat history described so far left out a major element in the Landsat program; the ground system hardware and software that received, processed, stored, and distributed the data to users. The history of the development and operation of the ground-based systems is as full of trials, tribulations, decisions, and revisions as there are in the space segment’s story. The facts in the following summary have been taken (sometimes literally) from an excellent article by W. C. Drager et al. (9).
To Pecora (8), the success of Landsat depended on making the data available to users in a timely and efficient manner. NASA recognized that it was its responsibility to develop the hardware and software to process the data from its new instruments. The technical problems turned out to be more difficult than expected, and data processing systems were slow in reaching the ability to handle the amount of data that was being downloaded from the spacecraft and ordered by users. Also, NASA did not think that its job was to be operationally responsible for the processing, storage, and distribution of the data products. Thus prior to the launch of Landsat 1, NASA signed an agreement with the USGS to process, archive, and disseminate Landsat data at the USGS EROS Data Center (EDC). That agreement, together with similar agreements with NOAA during its Landsat tenure, was codified by The Land Policy Act of 1992 that directed DOI/USGS to maintain a national archive of land remote sensing data and, as noted before , to make it available to all users at the cost of fulfilling user requests. That remains the situation currently.
As the United States established the Landsat program, it made extraordinary efforts in the United Nations and elsewhere to gain worldwide acceptance of nondiscriminatory dissemination of remotely-sensed data from space. The United Nations adapted this position in 1986 as ”The Principles Relating to Remote Sensing of the Earth from Space” (10). This concept (often called the ”open skies” policy, although that name is more correctly applied to an Eisenhower initiative to exchange the rights to do aircraft remote sensing with Russia and other nations) was responsible for many early activities aimed at getting Landsat data into the hands of data-poor third-world countries. Soon after the launch of Landsat 1, U.S. development assistant agencies, including USAID and the International Development Bank (IADB), began offering sizable Landsat data grants and funding for personnel to third-world countries. In 1974, for example, USAID awarded $260,000 worth of Landsat data and training grants to 10 developing countries in South America and Africa. By 1981, that program had grown to approximately $40 million and included projects in 35 countries that ranged from forest surveys in Costa Rica to geological mapping in Monaco (11).
Landsat revenue history.
Figure 13. Landsat revenue history.
NASA and the USGS initiated another major “open skies” initiative, the establishment of Landsat data receiving stations in foreign countries. NASA agreed to downlink Landsat data directly to such stations every time the satellite passed within its range. In addition to disseminating the data to the widest possible user audience, the establishment of foreign ground stations had two other very practical reasons: they served as a very valuable backup for the failure of the onboard tape recorders that were the weakest link in the data system, and they provided a significant source of revenue to NOAA that offset its operational expenses. The early annual fee of $250,000 rose to $600,000 by 1998 (11). Figure 13 presents Landsat’s revenue history that shows the large part that station fees played. Each station has the right to sell all of the data it receives and the responsibility to archive that data for possible use by scientists; this responsibility has not been followed to the extent that the Landsat scientists would like. However, the foreign stations have been very active in distributing Landsat data. From 1979 through 1995, more than $234 million worth of data has been distributed worldwide; at least $103 million of that was through foreign ground stations (9).
Starting with Canada in 1972, 17 countries have built Landsat receiving stations over the years. The names and locations of 16 stations are presented in Fig. 14. (Argentina, Spain & Thaiwan created stations after the period shown on the chart.) In the midlatitudes, each station covers an area the size of the United States, and at the higher latitudes, a great deal more. The result is that the ground stations cover virtually all of the global land cover except a large portion of northern Asia, principally Russia and China, and central Africa. These stations, all of which are government run or supported, have become their country’s focal point for remote sensing. In recent years, many have established agreements with SPOT, Radarsat, and commercial high-resolution satellite operators.
An example of Landsat receiving stations and areas of coverage.
Figure 14. An example of Landsat receiving stations and areas of coverage.
There is no question that the Landsat program’s early and continued aggressiveness in promoting and supporting foreign ground stations has been the major element in developing the worldwide application of satellite data and of creating the many national infrastructures that are critical in exploiting the satellites’ unique global capabilities for both global science and the creation of a global market for commercial systems. This activity also became important to the Landsat program itself; the support of the international ground stations aided in the many debates on whether Landsat should be continued.

The International Period

The ”United States only” period ended rather abruptly in late 1985 as the foreign activities generated by the early Landsats culminated in the launch of multispectral satellites by Russia, France, Japan, and India, all within the following 2 years. Though it took a little longer, Seasat’s radar images also impacted the international activities starting with the launch of Russia’s COSMOS-1870 in 1987, followed a year later by the Japanese and ESA launches.
This ”international” period of very aggressive foreign development and launch of civil land-imaging satellites extended through 1996. During this second period, France, India, Japan, and Russia launched 14 optical systems. The United States’ only launch, Landsat 6, did not make orbit. It was only the incredible extension of the orbital operations of Landsats 4 and 5 beyond their 5-year design lives that kept the U.S. land imaging program alive. The radar situation was even more biased as the foreign countries aggressively pursued radar imaging and launched seven radar satellites in this period. Satellite Types. Before discussing the individual programs, it will be useful to note the data of Table 2. This table divides the current land imaging satellites into four classes; Landsat-like, frequent global coverage, high-resolution, small area coverage, multi & hyperspectral experimental, and radar. The columns provide the ground resolutions of the panchromatic and multispectral bands, the location of the color bands, and the image swath. The similar values of these parameters in each class illustrate the practical results of the trade-offs that must be made among the three principal observation variables, spatial resolution, spectral resolution and band coverage, and temporal resolution or swath width when designing a satellite/sensor system. The table also lists the wide field of view sensors carried by the Indian, French, China/Brazil, and U.S. satellites that provide daily or near daily coverage everywhere at low resolutions, usually 1 kilometer or a bit less. These sensors are similar to the imaging sensors carried on weather and ocean color satellites that are not covered in this article. (See Weather Satellites by Singer and Rao in this topic.)
Table 2. Resolution, Bands and Swath Width of Some Current Satellites

Satellite Orbit Sensor Resolution in meters SW
types -1-
Thematic mapper bai
Pan VNIR SWIR
1 2 3 4 5 7
nds MWIR TIR 6 Km
Landsat like, frequent global coverage
tmp4E-24 tmp4E-25 tmp4E-26
High resolution, small area coverage
tmp4E-27 tmp4E-28
Multi & hyperspectral experimental
tmp4E-29 tmp4E-30 tmp4E-31
tmp4E-32 tmp4E-33
tmp4E-34 tmp4E-35 tmp4E-36 tmp4E-37
tmp4E-38 tmp4E-39
Radar C band
tmp4E-40 50 100 100
Companion low res wide area sensors
Sat Inst. # Res Swath bds meters Km
IRS-1C,D WIFS 3 188 810 IRS-P6 AWIFS 3 70 720 SPOT 4,5 Veg. 4 1000 2200 CBERS WFI 2 260 900 Terra MODIS 2 250 2330 5 500 2330 29 1000 2330

The broad swaths of the Landsat class satellites enable them to cover the entire globe many times each year: Landsat and CBERS 23 times; SPOT, when using both sensors, 15; and IRS 18 times a year. The number of times that optical systems can be expected to obtain cloud-free images is one-half to one-third of those figures. The public sector is responsible for mapping and managing large (state to country to global) land areas, so it should not be surprising that all of the satellites in this category are government funded. The high-resolution systems have about one-tenth the swath of the broad area systems, and as a result, they cover the total land cover about twice per year. However, they can point off-track and thus can return to any given site within 1 to 3 days.
The majority of the satellite and sensor data in following sections were taken from Ref. 14. It is not much of an exaggeration to say that this chapter could not have been written were it not for the extraordinary work of H. J. Kramer. Mr. Kramer’s 1500 pages recording the engineering and operational characteristics of every sensor and satellite launched into Earth orbit from Sputnik to 2000 will be considered the essential reference volume of Earth and atmospheric observation programs for years to come.
France. The SPOT (Satellite Pour l’Observation de la Terre) series was created by the government’s Centre National D’Etudes Spatiales (CNES) and is operated by the SPOT Image Corporation. SPOT Image sells the SPOT images and data products worldwide through a commercial network of distributors and 23 ground receiving stations. The corporation’s main shareholders are the government’s Centre National D’Etudes Spatiales (CNES—38.5%) and ASTRIUM, the prime contractor for the SPOT satellites (35.6%) (15).
SPOT was designed for the commercial application market from the beginning. Its 10-meter Pan band was pointed to user mapping needs, and its ability to look from side to side made it possible to see any particular site seven times during its 26-day global coverage cycle (compared to Landsat’s once in 16 days, see Figure 4). This capability was developed to meet a commonly expressed user requirement to see a site often and at specific times. Four satellites are currently in operation that give customers the opportunity to see their sites every day. SPOT 1, 2, and 3 had only three color bands. This limited SPOT’s value for some types of multispectral analysis of special use in crop monitoring. However, the addition of a fourth band on SPOT 4 and 5 has somewhat mitigated Landsat’s advantage in such multispectral applications.
Spot 1 was launched in 1986, 2 years after Landsat was commercialized; the two commercial entities competed fiercely. In 1991, Landsat’s sales were $32 M and SPOT sales were more than $40 M. Neither figure is close to making the program self-sustaining. Since 1999, SPOT sales have faced serious competition from Landsat 7 because the 1992 law requires selling Landsat data at the ”Cost of Filling User Requests” (COFUR) which is presently $600 a scene, compared to SPOT’s roughly $2000 for a much smaller image. Landsat 7′s 15-meter Pan band also cut away much of SPOT’s 10-meter advantage. Based on the launch of SPOT 5 (4 May 2002), Spot Image is hoping that its 60-kilometer swath, 2.5 and 5.0-meter resolution images will fill a large mapping niche that neither the new 1-meter, 10-kilometer swath commercial satellites or Landsat 7 can fill. Perhaps even more salable is its currently unique ability to produce continuous stereo strips for the production of high accuracy three-dimensional maps. Spots 4 and 5 also carry the vegetation sensor that the European Union and several of its member countries developed to meet their agriculture monitoring requirements. This sensor provides a daily 1-kilometer global coverage similar to, but with much better spectral data than, the advanced very high resolution radiometer (AVHRR) on POES used by NOAA to produce daily and 10-day vegetation maps of the globe. However, the AVHRR data are provided at the cost of reproduction, as are the data from NASA’s moderate-resolution imaging spectroradiometer (MODIS) sensor that has even better spectral data than the vegetation sensor. Thus, the U.S. policy is again in the position of negatively affecting France’s commercialization objective.
SPOT plays an important role in the European Community (EC). The Commission of the European Communities is using SPOT to set up an environmental database covering the entire EC. SPOT data are now an integral part of the EC’s operational agricultural statistics system used to manage its common agricultural policies.
India. The Indian Space Research Organization (ISRO) manages the Indian Remote Sensing Satellite (IRS) system. The intent of the program is to support India’s National Natural Resources Management System (NNRMS). NNRMS supports the national economy in the areas of agriculture, water resources, forestry, and ecology, and the availability of the new high-resolution satellites makes studies of urban sprawl, infrastructure planning, and large-scale mapping possible. The basic use of the IRS data is to support such national infrastructure efforts, but ANTRIX Corp. Ltd., ISRO’s marketing arm, sells the data commercially and has franchised sales of its data through the Space Imaging Corp. in the United States (1).
As can be noted in Table 2, IRS-1C and D have four color bands, 23-meter resolution, and a 142-kilometer swath that make it comparable with Landsat in many ways. However, IRS-1C’s 5-meter panchromatic capability was the best resolution commercially available on a regular basis until the launch of SPOT 5. (Note that the Russians sold some of their archived intelligence satellite data scanned to 2 meters resolution in this period). Having launched its Technology Experiment Satellite (TES) in October 2001, India is now the only government orbiting a nonmilitary, 1-meter class mission. Its current plans include launching one more. Its high-resolution systems do not carry a multispectral sensor apparently because color is not required for mapping and tracking civil and military activity/construction. India also plans to match SPOT 5′s 2.5-meter relatively broad swath capability with its P-5 satellite.
Japan. Japan’s early interest in keeping up with the United States and other space-faring nations began with launcher development and communications satellites. As noted in Ref. 1, ”The success of Landsat led to Japanese interest in remote sensing… (and their)… remote sensing strategy therefore followed the so-called ”Landsat modeP—developing remote sensing satellites for scientific and research purposes.” From 1987 to 1996, Japan launched four satellites that had land imaging sensors of better than 50 meters resolution.
The first two, Marine Observation Satellites 1 and 2, launched on 19 Feb. 1987 and 7 Feb. 1990, respectively, focused on ocean and atmospheric data but also carried two land imaging sensors that had 50-meter resolution and 100-kilometer swaths. When combined with a 15-kilometer overlap, their total swath matched Landsat’s 185 kilometers exactly. The sensor’s 15-kilometer overlap, also provided stereo images. The next satellite, Japan Earth Resources Satellite (JERS-1), launched on 11 Feb. 1992, was devoted totally to Earth resources. It carried both optical and radar sensors; each had 18-meter resolution and a 75-kilometer swath. Its successor, the Advanced Earth Observing Satellite (AD-EOS), launched on 17 Aug. 1996, was like the MOS satellites, a combination land, ocean, and atmospheric measurement system. The land sensor, the Advanced Visible and Near-Infrared Radiometer (AVNIR), added a Pan band that had 8-meter resolution and the additional capability of tilting from side to side for faster site-return opportunities. The Japanese have consistently opted for large satellites that have many sensors (Table 2 lists only the land imaging sensors; each had several others). That is often the preferred scientific approach because it provides the opportunity for comparing multiple observations of the same site. The unfortunate early demise ofADEOS has caused a large gap in the Japanese satellite program. They have, however, ambitious plans to continue in the land imaging field and are currently building their Advanced Land Observing Satellite (ALOS) for launch in 2004. ALOS will carry both optical and radar sensors. Its two optical sensors contain advanced features that are required to meet the mission’s goals of mapping at 1:25,000 scale without the use of ground control points and with 3.5-meter vertical accuracy for contour plotting. The panchromatic remote-sensing instrument for stereo mapping (PRISM) provides continuous three image stereo (fore, nadir, and aft) and has 2.5-meter resolution and a swath of 35 kilometers. The nadir view has a swath of 70 kilometers. The advanced visible and near-infrared radiometer-2 (AVNIR-2) has four spectral bands at 10-meter resolution and a 70-kilometer swath and has the ability to swing off-track for quick return viewing. Its other major goal of 24-hour response for covering disasters requires both pointing and an all-weather capability of the radar system. It will be an impressive test of the value of a multisensor suite on a single satellite.
Japan has also supplied a major new land imaging sensor, the Advanced Spaceborne Thermal Emission and Reflection Radiometer (ASTER) that is currently flying on NASA’s Earth Observation Science (EOS) Terra Satellite. See discussion later.
Russia. Russia was the first to use CCD push-broom technology in space for the 18 June 1980 launch of the experimental Meteor-Piroda-5 satellite carrying the MSUE-E (multispectral scanning unit electronic) that had three spectral bands, a spectral resolution of 28 meters, and a swath of 28 kilometers. This successful test was followed up by incorporating several versions of this sensor into a series of land remote sensing satellites similar in function to Landsat, called the Resurs-O1 series. Operated by SRC Planeta, a large conglomerate of scientific and industrial centers, the four satellites provided moderate (45-meter) resolution multispectral images. Resurs -01 #1 was launched on 3 Sept. 1999, #2 on 20 May 1988, #3 on 4 Sept. 1994, and #4 on 7 July 1998. None is currently operational (14). Unlike France, India, and the United States, Russia did not commercialize these data.
However, Russia was the first to offer high-resolution imagery on the commercial market. The Soviet trade association Soyuzkarta offered 5-meter imagery in 1987, and in 1992, two Russian firms began to sell selected images whose resolutions were as low as 2 meters (16). As noted previously, the 2-meter commercial sales were important in pushing the Clinton administration into a commercial licensing policy that has granted licenses for 1- and 0.6-meter commercial resolution systems. The Russian data were digitized from film data taken by their military satellites. The Soviet and the Russian agencies concluded many sales distribution agreements with U.S. and European market organizations (17), but none has been commercially successful. The most aggressive of these programs was SPIN-2 (for Space Information, 2 meter resolution). In July of 1995 Sovinformsputnik signed a contract with a group of American companies, Aerial Images, Central Trading Systems and Lambda Tech International, for a bulk supply of 2-meter imagery of the southeastern part of the U.S. The contract envisioned American customers paying for the launch and operation of a Kometa film carrying satellite. The first attempt on 5/14/96 was lost due to failure of the Soyuz-U rocket. The second launch and flight from 2/17/98 to 5/2/98 was a success. However the commercial sales, in spite of a companion agreement with Microsoft for the development of a system for selling imagery on the Internet, failed to materialize, putting an end to future launches. However, data from a wide range of sensors, including two synthetic aperture radars, are still being offered (17) and are potentially useful to those seeking to chart land surface changes during the last 20 years.
In addition to making images from its military satellites available, Russia has many plans for future satellites (17), but all are apparently on hold at present due to the lack of available resources. However, it can be expected that their considerable experience in optical and radar technology and their ongoing launch and spacecraft activities will someday bring them back into the commercial imaging market.
China. China’s first venture into land satellite sensing was carried out in cooperation with Brazil. The China/Brazil (CBERS) satellite carries two multi-spectral sensors, the HRCC (high-resolution CCD camera) and the IRMSS (infrared multispectral scanner). The HRCC is a push-broom sensor that has five 20-meter resolution bands and the ability to image off-axis to provide repeat viewing in 3 days. The IRMSS is a whisk-broom scanner like the TM and has the same bands: 76-meter resolution in the reflective bands and 156-meter resolution for the thermal. Both sensors have a 120-kilometer swath. The IRMSS also is like the TM and the ETM in that it has an onboard calibration system that includes both an internal calibrator and a solar calibrator that provide 3% band to band and channel to channel calibration precision and absolute calibration accuracy of less than 10%; these figures compare favorably with the TM. The satellite also carries a wide field imager that has 250-meter resolution and an 885-kilometer swath. The two countries have apparently made no attempt to make the data available in the marketplace. The first CBERS was launched on 10/14/99 and the second is scheduled for 8/10/03. The two countries have also announced plans for CBERS 4 and 5 that will add a 5 meter panchromatic sensor to the multispectral capability of CBERS 1 and 2. China has also launched two satellites of its own, Ziyuan ZY-2A and 2B launched on 9/1/00 and 10/27/02 respectively. Both have panchromatic sensors, the first reportedly of 9 meters resolution and the second 3 meters.
Korea. Korea launched its first satellite on 10 Aug. 1992 as part of its program to gain space technology capability and has supported an increasingly aggressive program since then. KITSAT-3 (Korean Institute of Technology Satellite 3) launched on 26 May 1999 was Korea’s first medium-resolution land imaging satellite. Its Multispectral Earth Imaging System (MEIS) sensor has three 15-meter resolution bands and a 50-kilometer swath. The KITSAT series was followed by the KOMPSAT (Korea Multi-Purpose Satellite) series. KOMP-SAT-1, launched on 20 Dec. 1999, carries an electro-optical camera (EOC) that has a 6.6-meter Pan band and a 17-kilometer swath. It is designed to obtain cartographic maps of Korea at 1:25,000 scale. It also carries an ocean color sensor. KOMPSAT 2 is being developed for launch in 2004 to provide surveillance of large-scale disasters. The plan is for it to carry a 1-meter Pan and a 4-meter, four-band multispectral sensor (14).

The Current Period

The third period, 1997 to the present, is dominated by the entrance of commercial high-resolution systems, but includes the return of U.S. activity in the form of an R&D program and the launch of the long-awaited Landsat 7. In addition, as discussed before, Korea, China, and Brazil, the latter two in a cooperative program, joined the civil land imaging satellite community in this period.

Commercial Satellites

As can be seen in Table 2, high-resolution systems image objects that range in size from 1.8 to 0.6 meters (6 to 2 feet). These resolutions are about the same as the original classified systems of the early 1960s. The private sector dominates in this capability range. Three of the four high-resolution mission satellites launched successfully to date have been commercial systems.
U.S. commercial systems are the direct result of the 1992 Land Remote Sensing Policy Act and the subsequent 1994 Clinton administration policy that led to licensing commercial resolutions as high as 0.6 meters. The commercial sector, after ineffectually pursuing a market in Landsat and SPOT images is hoping that satellite imagery of 1 meter or less resolution can be profitably marketed at prices that will capture some of the very large aircraft remote sensing market and create entirely new markets based on the satellite’s ability to gather scenes anywhere, anytime, for anyone, clouds willing. The commercial companies are also well aware that their 0.6 to 1.0-meter resolutions are sharp enough to provide very useful information to the military and are pursuing both national and international defense markets.
These high resolutions have been achieved by restricting the width of their images to between eight and 16 kilometers and for those with color bands, limiting the spectral coverage to the VNIR bands. The narrow swath does restrict their utility for the broad area monitoring tasks of the Landsat type systems, but the satellites’ ability to point off-track provides the capability to image any specific target at 1 to 2-day intervals. This agile pointing enables the acquisition of stereo images and thus the production of three-dimensional images.
Three American companies, Space Imaging (SI), Digital Globe (DG) (formerly Earth Watch), and OrbView and one Israeli company, ImageSat International (ISI), are currently offering high-resolution images commercially. They have had their share of launch failures. Prior to the successful launches, Digital Globe lost its first two satellites at or shortly after launch; OrbView, SI, and ISI, their first. Launch insurance has enabled them to survive to date. All recognize that their business cannot survive with single satellites, and the three U.S. companies are each planning follow-up satellites. ISI has published a very aggressive plan for six more launches by 2006 of a new and improved system that essentially copies Digital Globe’s half-meter panchromatic and four-band multispectral capabilities. The timing of all of these future systems is somewhat uncertain because it depends on each company’s ability to obtain the necessary funding from their slowly growing sales or from the currently troubled financial marketplace.
The technical success of the commercial systems has proven that the private sector can meet the technological challenges in providing sophisticated data products. As each company struggles to turn its technical success into a financial success, it is faced with a variety of challenges. The very technologies that have made spacecraft imaged data possible, digital sensors and GPS (the satellite Global Positioning System), have also increased the capabilities oftheir aircraft-based competition. The satellite’s ability to get images of any and all countries is their prime advantage over their aircraft-based competitors. That feature is of interest to many multinational companies and to some nongovernmental organizations (NGOs). It is of great interest to governments for military and political reasons. The U.S. companies are pursuing agreements with foreign countries, but they face the perception that the ”shutter control” clause in their government license would subject the availability of data to the whims of U.S. foreign policy. Several foreign governments have already announced plans for their own high-resolution satellites (see The Future section). In the United States, the administration is well aware of military utility of high-resolution data and has instructed NIMA (the National Image and Mapping Agency) to use commercial systems to expand its capabilities and to reduce the requirements on its classified systems.
There may be another market niche. As noted previously, the Congressionally mandated commercial data buy for the Landsat Data Continuity Mission has prompted two companies, Digital Globe and Resource 21, to compete to provide NASA with Landsat data as a by-product of their own projected satellite data service, weekly crop health reports to farmers. These reports will enable farmers to take early remedial actions on the less productive areas, the so-called ”precision farming” approach made possible in part by another satellite service, GPS. These systems will have the coverage and spectral capabilities of Landsat but at higher resolutions, 7.5 or 10 meters and more frequent coverage, 4 or 8 days.
At this time, it must be concluded that the commercial land imaging satellite programs are still in their start-up stage and that their future is still uncertain. The interested reader is advised to seek out their web pages to keep up with their progress.

Multi- and Hyperspectral Tests

These systems, 3 funded by the U.S. government and one by ESA, are space tests of advanced sensors. As shown in Fig. 6, the sensors include more and narrower bands to provide increased sensitivity for analysis. Aircraft tests have demonstrated that measurements of the radiance of essentially all of the spectrum using the so-called hyperspectral sensor provides increased ability to identify a wide range of geologic material and crops as well as to discern crop health and potential yield. Two of the new sensors have added increased thermal coverage that has also been demonstrated to improve crop classification and health status.
ASTER (Advanced Spaceborne Thermal Emission and Reflection Radiometer), a Japanese sensor flown on NASA’s Terra mission, has twice the number of Landsat’s 7′s spectral bands. It is actually a set of three telescopes flown together. The VNIR system includes a backward pointing telescope that is combined with its nadir telescope to provide 15-meter resolution, continuous stereo data. The six shortwave IR (SWIR) bands and the five thermal IR (TIR) bands, all at 15-meters resolution, are opening new analytical capabilities.
The MTI (Multispectral Thermal Imager) was developed by the DOE’s Office of Nuclear Non-Proliferation to test the ability of multiband thermal data to measure the heat output of nuclear plant-cooling and thereby its power level. In addition to its four standard VIS 5-meter bands, it includes a unique series of 20-meter bands, three NIR, three SWIR, two MWIR, and three TIR. The DOE 3-year program includes imaging industrial, government, and natural sites. Unfortunately, its data have not yet received the attention of the broader research community they deserve.
EO-1: This NASA spacecraft is designed to test advanced technologies of potential use in the next Earth resources satellite. It carries two 30-meter resolution sensors: the Advanced Land Imager (ALI) with six VNIR bands, three SWIR bands, and one 10-meter resolution pan band; and a hyperspectral imager, Hyperion, which has 220 bands covering the VNIR and SWIR range. It has been placed in orbit near Landsat 7 to obtain comparative data sets. An extensive scientific and application analysis program is in process as this is written. It has proven so useful that EO-1 operations are being continued beyond their planned duration by selling data to several government agencies. The ALI push-broom approach resulted in a sensor that is one-quarter the weight and one-seventh the volume of the ETM + but provides three more bands and a 10-meter pan band (18).
PROBA: PROBA is a technology demonstration satellite designed to demonstrate many advanced satellite technologies in a manner similar to EO-1. It carries CHRIS, an experimental hyperspectral imager that can be programmed to produce either 19 VNIR bands with 18 meter resolution or 63 bands with 36 meter resolution. It also carries an 8 meter resolution panchromatic sensor.

Radar

Radar images through clouds and at night; this can be very important in those parts of the world like the tropics and Europe that are cloud-covered a high percentage of the time. Their sensitivity to a target’s size, shape, and material characteristics and to its moisture content provides an entirely different information set than an optical system’s measurement of reflected and emitted radiation. Radar sensors come in a variety of types based on their signal frequency and signal and reception polarization. To date, all satellite radar systems have been single frequency, which, like panchromatic optical data, are very valuable for mapping type functions but are not very good at classification functions.
Radar has not yet been used in civil Earth observation satellites to the extent that the optical sensors have. As the discussion on future systems below notes, this may change in the next decade. Figure 1 shows the orbit history of the civil radar satellites to date. Seasat was a multisensor satellite developed by NASA in cooperation with the Navy. Its SAR sensor is regarded as the first imaging SAR system used in Earth orbit (14). Launched 27 June 1978, it failed abruptly 106 days after. In spite of its short life, it was a great success. The quality of its data was responsible for ESA’s 1981 decision to favor development of a radar system. ESA was stimulated by radar’s ability to image through Europe’s frequent cloud cover (19). As can be seen in the figure, ESA has maintained the continuity of its radar program to the present. In the United States, radar data were apparently judged so potentially valuable for military use that civil radar satellites were discouraged, and there have been no U.S. civil radar satellites to date. The current Defense Department position limits civil systems to five-meter resolution, a restriction that caused Canada to cancel their contract with an American firm and turn to Italy to build its planned 3-meter system.
At present, the uses of single-frequency systems are many and impressive. A short list would include oil spill detection, ice movement monitoring, mapping (one of Canada’s prime applications), monitoring forests and agricultural land in the tropics, geological mapping aided by its oblique viewing angle and vegetation penetration capabilities, and disaster evaluation, especially cloud-covered floods. Using data from two overflights, radar satellites can provide interferometric maps of surface elevation changes of the order of a few centimeters.
Current and planned radar satellites come in two styles. ERS-2, Enviasat, and ALOS-1 carry many other observation sensors along with their SAR. Canada’s current and planned Radarsats are dedicated to their SAR mission.
A discussion of radar satellites would not be complete without mentioning the NASA-sponsored Shuttle-based radar sensor program. Although restricted from creating free-flying radar satellites, NASA’s Jet Propulsion Laboratory (JPL) was able to advance the technology in a series of five Shuttle missions carrying radar sensors. SIR A, flown in 1981, and SIR B, flown in 1984, carried L-band sensors. The third and fourth Shuttle missions, SIR C/X-SAR, flown twice in 1994 in cooperation with Germany and Italy, consisted of separate but connected L, C, and X-band SARs. The long and complex analysis program of that data has shown that multifrequency radar analysis can serve many of the applications that multispectral analysis has made possible for optical systems, which, in effect, could lead radar into closer competition with the now dominant optical world. The fifth Shuttle mission produced what may be space-based radar’s most useful accomplishment to date. The 10-day Shuttle Radar Topographic Mission (SRTM) was sponsored by NASA and NIMA and launched on 11 February 2000. It carried two SIR-C/X SARs separated by a 200-ft boom attached to the Shuttle’s bay. The data from the resulting continuous stereo view produced a topographic map of the land surface on which 95% of Earth’s population lives and which is 80% of the total global land surface. Surface elevation of all ground points 30 meters apart will be provided to scientific investigators, and elevations 90 meters apart will be open to the public.

The Future

Predicting satellite launch dates, even by those who are paying for them, is a risky business. Since 1995, the author has tracked the progress of more than 30 land imaging satellites. The delays between the first announced launch and the actual launch date averaged a bit more than 2 years, and nine of the 30 had an average of a 4-year delay. Keep this history in mind while reading the following discussion.
Even without firm knowledge of the status of many of the satellites being proposed, one thing is certain, that civil land imaging satellites will be spread over the globe. This is dramatically illustrated by Fig. 15. As its caption aptly states, ”Global competition is heating up.” There are 16 foreign countries that have plans for land imaging satellites. The 1-meter and better satellites, currently mostly the province of the U.S. commercial sector, will be seeing a significant part of this global competition in the form of government systems in France, Italy, Korea, and India, as well as the multisatellite plans of Israel-based ISI.
The most mature of these plans have been noted in the previous sections, but as this is being written, both governments and private corporations are discussing/announcing ever more ambitious plans. Rapid/Eye of Germany continues to promote its plans for a commercial four-satellite, 6-meter system to provide a crop monitoring service similar to that being proposed by the LDCM contractors. The French and Italians are discussing the development of a five-satellite, 1-meter satellite system to be launched by 2005/2006 made up of the French Pleiades two-satellite optical system and the Italian Cosmos-Skymed three-satellite radar program (14). The German government announced TerraSarX and TerraSarL, a 1-meter radar satellite program planned for launch in 2006. These somewhat competing programs are the subjects of much debate in the European Union planning process. However, should most of these plans carry through to launch, there could be 19 1-meter or better systems in orbit by the end of 2006, including three to five radar systems.
Global competition is heating up. Reproduced by the kind permission of John Baker and the Rand Corporation.
Figure 15. Global competition is heating up. Reproduced by the kind permission of John Baker and the Rand Corporation.
A new approach to combining daily coverage and medium resolution land observation capability by creating a constellation of small low cost microsatellites (as much as 1/5th the cost of current commercial systems) is being developed by the British company, Surrey Satellite Technology Ltd. (SSTL) (20). An early example is already in orbit. UoSat-12, developed and funded by SSTL, carried payloads from Singapore and ESA into a 64° inclination orbit on a Russian launcher on 12 May 1999. Its payload included one 10-meter resolution panchromatic sensor that had a 10-kilometer swath, one 975-meter resolution sensor that had a 1000-kilometer swath, and two 32.5-meter resolution multispectral sensors that had 33-kilometer swaths. SSTL is currently building a five-satellite Disaster Monitoring Constellation (DMC) that will be capable of providing 32-meter resolution, three-band color images of any location once per day (14). The first mission sponsored by Algeria was launched from Russia on November 28, 2002, and four more are scheduled in 2003 (21). The other sponsors are Nigeria, UK, Thailand, and Turkey. China and Vietnam are planning similar systems for launch in 2004 that carry a 4 meter panchromatic sensor. It is clear from the number of nations involved, that remote sensing of one’s land from space brings out a strong desire to have one’s own satellite for reasons that appear to be a mixture of security, practical applications, and simple national pride.

Summary

Of all efforts to date to use our ability to operate satellites in space for the betterment of life on Earth, the U.S. Landsat program certainly ranks among the most successful. It invented and implemented a whole new technology, a truly new and powerful way of looking at Earth through the analysis of its reflective and emitted radiation, a method that for the first time has made it possible and practical to monitor the entire surface of Earth on a regular basis at resolutions capable of observing and assessing human-scale activities. At the same time, the Landsat program and other government programs actively spread the data and the technologies involved throughout the whole world, with specific emphasis on third-world countries. The technical achievements are significant, but it may be that this worldwide sharing will be its enduring monument.
All of this was accomplished despite the difficulties the program encountered from before its birth to the present, difficulties related to national security issues, agency roles, delays in data delivery, funding uncertainties, and a failed attempt to commercialize the federal program. Its 30 years of synoptic global data remain the only systematic collection of consistent global land images. The current U.S. policies of having Landsat image the entire global land surface four times a year and of providing data to all for the cost of reproduction has yet to be emulated by any of the foreign satellite systems.
The rest of the world started their civil land imaging programs 15 years later, and they have launched more than twice the number of missions. France and India equaled or bettered the launch pace of the early Landsat years, and both countries have strong plans to continue and improve their systems. Japan and Russia started just as vigorously, but halted after their late 1990 launches. Japan has seemingly firm plans to restart its program in 2004, and Russia’s future plans remain uncertain. Although foreign optical satellite programs have been very impressive and in some ways more innovative than the U.S. program, their image acquisition programs have focused on getting data for sale in the marketplace rather than on creating a synoptic archive of global images for scientific use.
The radar history is entirely different. The U.S. short-lived Seasat radar mission provided sufficient data for Russia, Japan, ESA, and Canada to create radar satellite programs that, Russia excepted, have been maintained to the present. To date, all of the civil land imaging radar satellites have been foreign, and current planning statements indicate that this situation will continue into the foreseeable future. This is due to a combination of reasons, including Europe’s persistent cloud cover, Canada’s strong interest in mapping the changing ice fields of their North, and the reluctance of the U.S. defense establishment to allow civil radar systems.
Until the late 90′s all of the systems were funded by their respective governments. As a result of the U.S. 1992 Land Remote Sensing Policy Act that permitted the development and operation of civil satellites with resolutions in the meter or less range, the pull of the marketplace entered the land imaging satellite game. This was a significant break with the tacit international policy of keeping high-resolution ”spy quality” data as the exclusive province of governments. This opening of high-resolution data to the civil sector was helped by Russia’s commercial sale of 2-meter data, but was vigorously protested by France as well as by U.S. defense agencies. Congress’ motivation was based on the belief that meter-level systems could compete with aircraft data for that multibillion dollar market and that sharing the cost of such satellites with a commercial market would dramatically reduce the government’s data costs. At this time, it still is uncertain if purely commercial systems will find a sufficient market share to ensure their future.
In any case, the secrecy genie is out of the bottle. The availability of meter-scale data in the open marketplace has opened the door to data uses far beyond the scientific and large-scale land management functions of the Landsat resolution data. The door has been opened to global information transparency that can change the relationships between nation and nation and between citizens and their governments (1). The political, scientific, and commercial currents during the next 25 years of Earth-observing systems will be no easier to chart than were the first, but the systems they spawn will certainly advance the understanding and informed use of the planet’s resources and aid in establishing the trust among nations, that will make global environmental management possible.

The Internet

The above suggested reading will be useful. However, the field is a rapidly evolving one and the reader is well advised to make use of the internet. The NASA and USGS home sites have directions to sites specifically devoted to Landsat and other satellite imagery. All of the commercial companies have their own sites, easily located by any search engine. The same is true of all of the countries involved in satellite sensing. Finally, news about the status of any particular satellite can often be found by typing its name into Google.

Next post:

Previous post: