Just as the Arecibo S-band and the Goldstone X-band upgrades had propelled radar astronomy into new directions, in 1986 a second upgrade planned for the Arecibo radar and the restoration and upgrading of the Goldstone radar stimulated new shifts in the planetary radar paradigm. Instruments and hardware continued to drive the field. Fresh techniques, either developed by radar astronomers or borrowed from other fields, namely ionospheric and radio astronomy research, allowed radar astronomers to solve new problems on the terrestrial planets. Also, the bizarre radar signatures of the icy Galilean satellites appeared once again, though closer to home on the terrestrial planets, and suggested new problems to solve.
Planetary radar astronomy survived at JPL in a tenuous state as a testbed for DSN technology and as a mission-oriented activity. That state depended largely on support from specific upper-management individuals, Eb Rechtin and Walt Victor. After Rechtin left JPL and Victor transferred out of the Deep Space Network to the JPL Office of Planning and Review in December 1978, radar astronomy became vulnerable to extinction. The DSN Advisory Group, headed by Rechtin, had judged that radar astronomy was no longer the testbed of DSN technology.
The Goldstone radar was in desperate need of repair, and the old equipment had become very hard to maintain; very few people knew how to work with it. By 1980, much of the equipment was old and not functioning properly. During experiments, for instance, entire runs of data would be flawed or lost as a result of computer malfunctions. "You simply had to bite the bullet and rebuild the whole damned thing, particularly the data acquisition systems," Ray Jurgens explained.1 However, nobody wanted to pay the cost of the needed repairs and upgrades.
Reviving the Goldstone radar so that planetary radar astronomy could once again prosper at JPL required that the activity have a new rationale. The initial arguments for funding needed new equipment focused on the value of radar to NASA flight missions and to planetary geology. In 1979, E. Myles Standish, Jr., who was in charge of the JPL planetary ephemeris program, wrote a memo to Richard R. Green, who had recently been promoted from the radar group to Advanced Systems, to explain that if no radar experiments were conducted, then the accuracy of the ephemerides for the terrestrial planets would suffer, and JPL would not be able to meet its ephemeris commitments to either the Galileo or Magellan missions.2
Obtaining mission approval had been a requisite for acquiring antenna time for radar experiments. As George Downs explained, "Dick Goldstein always wanted to  connect us with a project. I believe he felt that if we tried to get constituency from geologists alone, we wouldn't make it. Well, he was right."3 In 1979, George Downs asked USGS geologist Henry J. Moore to write a letter in support of the Goldstone radar; Moore wrote to Arden L. Albee, Caltech professor of geology and the new JPL Chief Scientist.
Albee was sympathetic and met with members of the JPL radar group, Ray Jurgens, George Downs, Stan Butman, and Rick Green, on 24 January 1980. As a result of the meeting, Albee wrote to the NASA Office of Space Science recommending a line item for radar astronomy in the fiscal 1982 budget. Thomas Mutch, associate administrator, NASA Office of Space Science, replied that he could not raise the annual allocation; the funding level would have to remain level.4
Some support for resurrection of the Goldstone radar could be counted on coming from the Planetary Radar Working Group, which consisted largely of geologists in the USGS and academia plus smaller numbers of individuals representing SAR remote sensing and NASA Headquarters, as well as radar astronomers Pettengill, Campbell, Goldstein, and Len Tyler. The Planetary Radar Working Group met in conjunction with the AAS Division for Planetary Sciences and the Lunar and Planetary Science Conference and discussed priorities in radar astronomy at Goldstone and Arecibo. Of course, the fate of the VOIR mission, not the Goldstone radar, was the focal point of discussions.5
With support from the Planetary Radar Working Group, Ray Jurgens and George Downs wrote a proposal requesting about $1.8 million to purchase a VAX computer to reduce radar data. They submitted it to the NASA planetary geology office because they thought planetary geologists would be the prime users of the data. In retrospect, George Downs judged that reviewers saw the proposal as a threat to their own funding and did not give it good reviews, while those who saw the project's usefulness gave it good reviews.
Although Jurgens and Downs did not get the amount requested, the NASA Office of Space Science and Implementation did grant them enough to buy a new VAX-700 and about $150,000 a year to analyze radar data. Radar astronomy also achieved a modest level of recognition in 1982. The original 1971 NASA Management Instruction governing ground radio science, now considered obsolete, replaced the term "radio science" with "Radio and Radar Astronomy." Such was the state of radar astronomy when Downs left in 1982. 6
In 1983, radar astronomy acquired a new advocate, Nicholas A. Renzetti. Originally a DSN manager responsible for the interface between the DSN and its flight customers, starting in 1975 with the Viking and Voyager launches, Renzetti gave less attention to flight projects and more attention to applications of radio technology to non-flight projects, such as geodynamics, the Search for Extraterrestrial Intelligence, radio astronomy, and starting in 1983, radar astronomy. Renzetti took on the task of convincing the Office of Space Science and other NASA Headquarters departments that it was in NASA's interest to support the Goldstone radar as a scientific instrument.7
The new rationale for funding the Goldstone radar, as defined by Renzetti, would be its use as a scientific instrument. In his campaign to garner support for the Goldstone radar, Renzetti was assisted by Steve Ostro, who took a position at JPL in late 1984, after leaving Cornell. They negotiated a new task in December 1987, in which the Goldstone radar would be treated as if it were a facility, not as a science task, with an annual budget  of about $200,000 for hardware improvements. The NASA task underwrote the interface between the DSN and the radar astronomers. The objective of the new task, called the Goldstone Solar System Radar (GSSR), was to support planning, experiment design, and coordination of data acquisition and engineering activities for all Goldstone planetary radar astronomy. As Steve Ostro explained, "This has been the financial backbone for the Goldstone radar, and it is separate from the DSN."8
At the same time, Renzetti created a part-time position, the Friend of the Radar. The holder of that position was to carry out a number of duties, including NASA flight project science and liaisons with Arecibo Observatory, but most importantly interfacing with the scientific community. Tommy Thompson performed those duties until he became Magellan Science Manager in 1988, when Martin A. Slade replaced him. Slade had been a graduate student of Irwin Shapiro at MIT and had had some exposure to radar astronomy during summer jobs at Haystack. His main previous research interests, however, lay elsewhere.9
The creation of the GSSR task and the Friend of the Radar were only first steps in addressing the core issue of funding the Goldstone radar on the basis of its use as a scientific instrument. Renzetti took tentative, unsuccessful steps to open up the Goldstone radar to outside researchers in order to operate it as a national research facility. He approached Von Eshleman and two others from outside JPL to propose radar experiments at Goldstone. Renzetti also proposed to Tor Hagfors, NAIC director, that a single peer review panel assess radar experiment proposals for both the GSSR (as the Goldstone Mars Station or DSS-14 now came to be called) and Arecibo. Moreover, hoping to acquire a facility budget for GSSR on a level with that of Arecibo, Renzetti proposed to Hagfors that Arecibo and GSSR present a common front to NASA, rather than appear as competing facilities.10
But it did not make sense to pursue the common budget, Renzetti reasoned, as long as the GSSR was not a national facility. The annual amount requested from NASA to make the GSSR a "first-class scientific instrument," $500,000, was not well received at NASA Headquarters. In comparison, the NASA budget for the Arecibo radar was only $362,000 in 1986.11 Nonetheless, Renzetti, who felt there was a built-in bias in favor of Arecibo at high-level NASA meetings, submitted a formal proposal to make the GSSR a national facility, but it never got off the ground.12
A chief critic of the proposal to turn the Goldstone radar into a national research center was Dewey Muhleman of Caltech. He called parts of the proposal "ludicrous" and declared that it would do "nothing for Science, the Nation, NASA nor, in the long run, JPL." Moreover, he pointed out, the heavy scheduling of the antenna for spacecraft work militated against the plan. "I strongly favor," he wrote, "the idea of getting Radar Astronomy at JPL out of the closet of component development and into the light of pure science."13
Gradually, that was starting to take place. During a JPL administrative reorganization in the fall of 1987, the Office of Space Science and Instruments (OSSI) was created with Charles Elachi as its head. Elachi was a seasoned radar engineer with decades of SAR experience. After he obtained a modest level of funding, $150,000, from NASA Headquarters, Elachi named Steve Ostro manager of Planetary Radar Science and authorized him to allocate the funding.14
 "At that point," Ostro explained, "I had a little bit of authority. I had the program office backing me. I acted as somewhat of a filter on proposals and papers, when I could, and I acted as the voice of science for radar." Ostro agreed with Muhleman's perception that JPL placed too much emphasis on hardware and not enough on doing science. The science community in general, he pointed out, viewed the GSSR as state-of-the-art electronics, but saw Arecibo as producing state-of-the-art planetary radar data. The objective, Ostro declared in 1988, "is, a year from now, to have a sparkling list of GSSR radar articles that have appeared in high-quality journals."15 Despite such sterling intentions on the part of Ostro and Renzetti, keeping JPL, DSN, and NASA management aware of the Goldstone radar's scientific achievements and potential has been a Sisyphean task. In contrast, the value of radar astronomy was established from the outset at the Arecibo Observatory.
Here was an important difference between the two facilities that had a profound impact on the development of radar astronomy at each site. Even more important, however, was the fact that Arecibo had acknowledged and formalized the existence of radar astronomy from the start; whereas JPL purposely had denied radar astronomy any formal existence. The difference has had long-term implications that has favored radar astronomy science at Arecibo, while holding it back at JPL.
New hardware and fresh leadership enabled radar astronomers to make new discoveries about Mars, Mercury, and the asteroids with the Goldstone radar. The major hardware upgrade did not arise from a concerted campaign on the part of Renzetti and Ostro to improve the state of radar astronomy at JPL, but rather, in a fashion typical of the history of planetary radar astronomy, came from outside radar astronomy, namely, the Voyager mission to the outer planets.
The Voyager upgrade of the main GSSR antenna, known within the Deep Space Network as DSS-14, involved enlarging the dish diameter from 64 (210 ft) to 70 meters (230 ft), increasing the surface accuracy, and improving the receiving system. These measures increased the sensitivity of the DSS-14 significantly. Tracking and acquiring data from the Voyager spacecraft, as they encountered Uranus and Neptune, stretched the capacity of the Deep Space Network. During the Neptune encounter, the Voyager X-band radio signal would be less than one-tenth as strong as during the Jupiter encounter in 1979 and less than one-half as strong as during the Uranus encounter in 1986.
A study to enlarge all the DSN 64-meter antennas to 70 meters already had been undertaken as early as 1973 in preparation for Voyager when it was still called Mariner Jupiter/Saturn. After completion of design work in 1984, the upgrade of the DSS-14 began in October 1987 and concluded in May 1988.16 When Steve Ostro arrived at JPL in 1984, the DSS-14 lacked the threshold of sensitivity to do meaningful asteroid research. Upon completion of the initial upgrade phase, however, Ostro made his first successful asteroid observations with the DSS-14 in May 1986, when he detected echoes from 1986 JK, an asteroid only just then discovered by Eugene and Carolyn Shoemaker.
The Voyager upgrade had a profound impact on the practice of radar astronomy at JPL; it provided the GSSR the sensitivity needed to carry out research on a whole new set of targets (and to begin solving new sets of problems). Not only did the GSSR gain the ability to undertake significant asteroid research, but when linked to the Very Large Array in New Mexico, as we shall see later, it became a new radar research tool.
Despite these major upgrades, the GSSR had serious problems as a scientific instrument. The site lacked dormitory and cooking facilities for visiting or even JPL scientists, and the drive to Barstow 50 miles away on winding roads after a night of observations was dangerous. These deficiencies and dangers persist today. Furthermore, the radar itself was far from user-friendly. "It was just impossible to work," Ostro explained. "For example, the  VAX that is used for data acquisition at Goldstone is not good for radar astronomy for various technical reasons. It has been improved a lot since the mid-1980s, but even now it is difficult, for example, to stamp your data with a high-precision UTC [from the French for Coordinated Universal Time] time tag. For this kind of work, the first thing you need on your data is a UTC time tag."17
The Arecibo Observatory stood in sharp contrast to the Goldstone radar. It had proper quarters for visiting scientists, and the radar was far more user-friendly. Moreover, in 1986, the Arecibo Observatory proposed a major upgrade of the radar that would benefit both ionospheric and radio astronomy research and planetary radar astronomy. The Arecibo upgrade stirred Renzetti to seek funding for a Goldstone radar upgrade equal to the cost of the one megawatt transmitter proposed for Arecibo.
Renzetti lobbied the DSN and NASA hierarchy for funding for a Goldstone one megawatt transmitter, which JPL engineers initially estimated would cost $12 million. A good argument for DSN use of the radar would not fly; the rationale had to be its use for scientific research, Renzetti realized. The radar upgrade, to be completed in fiscal year 1993 and costing $10 million over two years, appeared in the DSN budget for fiscal 1989. JPL viewed the price tag as "pared to the bone." In the end, Congress approved the expenditure not as a specific radar upgrade but as an ambiguous improvement of the DSN. This ambiguity freed DSN management to use the radar transmitter money to purchase low-noise supercooled masers to improve antenna sensitivity for Galileo's encounter with Io.18
It was not clear, moreover, that radar science at JPL needed the one-megawatt transmitter. The estimated cost of the transmitter now stood at $16 million. Steve Ostro believed that if the cost were reduced below $8 million, the improved science capacity would justify the expense. The high cost reflected JPL administrative and DSN operational support requirements that added several million dollars to the cost.19
Ostro favored upgrading the GSSR antenna's subreflector to improve its ability to make asteroid observations. The long time needed to rotate the subreflector, which was never designed to act as a transmit/receive switch, compromised short round-trip-time asteroid observations. "The most powerful scientific rationale for the longterm support of GSSR," Ostro argued, was work on near-Earth asteroids. The estimated price tag for the transmit/receive upgrade, which involved turning the transmit-only horn into a horn capable of switching quickly back and forth between transmit and receive, was $485,000.20
Instead of going directly through the DSN hierarchy for a radar upgrade, Renzetti changed his strategy. The one-megawatt transmitter and two other radar improvements (construction of a transmit/receive horn and modernization of the data acquisition system) were submitted to a panel of outside scientists for review. Gordon Pettengill chaired the Goldstone Planetary Radar Science Review Committee, as the panel was called. It included planetary astronomers and geologists, as well as Don Campbell and Tor Hagfors from the Arecibo Observatory.21
 Although invited to join the Committee, Muhleman declined. He "took the attitude, well, this is one more panel, it can't be that important. How about if I don't come? Let me know how it comes out. That was a terrible mistake. It really was....The JPL viewpoint was not represented."22 More importantly for Muhleman, his viewpoint was not represented, and he paid the price. The Committee met on 8 August 1991 and presented its conclusions later that month. The Committee applauded "the efforts currently underway by JPL management to broaden the usage of the Goldstone facilities (including observations jointly with the VLA) to include members of the larger North American and global planetary communities."
Of the three improvements, the committee gave the highest priority to the single-horn, fast-transmit/receive-switchover system. That improvement would serve asteroid work only. "At a lower, but still high, priority," the committee endorsed the modernization of the data acquisition system and recommended that the output protocols and formats of the new system be coordinated with those of the Arecibo planetary radar. Each of these two improvements had a modest cost of about $500,000 spread over one to two years.
The one-megawatt transmitter, the committee judged, "seems less attractive as an upgrading option than the first two presented." The cost was too high for the amount of sensitivity gained. The value of the transmitter upgrade, the committee decided, lay in observing Titan, "but we do not find the scientific argument compelling for what appears to be a fairly narrowly focused study of a single object. We note also that the improved transmitter is unlikely to be available in time to provide data that materially assist in the design of the Cassini Mission."23
Titan, however, was of the highest research interest to Dewey Muhleman. "In my absence," he complained, "this panel frankly wrote a silly report. It just really made me sick to read it. It said that the only advantage of going to a megawatt on the Goldstone antenna was to be able to do Titan better with the VLA. Nothing else was really important. That is ridiculous. For everything we do, our integration time would be cut down by a factor of four by doubling our power to a megawatt. We would be able to do much more on each one of these objects and quite frankly continue to rival Arecibo after the upgrade."24
The struggle at JPL to gain recognition for the GSSR as a scientific instrument stood in stark contrast to the effort to upgrade the Arecibo telescope. Both NASA and the NSF already recognized Arecibo as a national research center, and the rationale for any upgrade would be on the basis of scientific need. Furthermore, the Arecibo upgrade stood to benefit all research at the facility, radio and radar astronomy and ionospheric research, not just planetary radar astronomy. Other factors eased the process of garnering support for the Arecibo upgrade, including the method of funding what was, in relative terms, a low-cost project.
The Arecibo upgrade was a package of five interrelated improvements: 1) installation of a ground screen to virtually eliminate noise from the surrounding earth; 2) adjustment of the reflector surface to enhance antenna gain; 3) correction of the pointing system; 4) replacement of the accumulation of radio astronomy line feeds with a single reflector feed possessing large bandwidth, low loss, high gain, and continuous frequency coverage from 300 MHz (1 meter) to 8 GHz (3.75 cm); and 5) doubling the S-band transmitter power to one megawatt. The total effect of these changes was to  increase radar sensitivity by a factor of 10 to 50 (about 20 times on average), to double its range or to detect objects 10 times smaller than previously possible.25
The principal objective of the upgrade, however, was to solve a problem that had plagued the telescope since its creation--the problem of spherical aberration. Unlike parabolic dishes, the Arecibo spherical antenna did not focus waves in a single point. The antenna feed system designed by the Air Force did not work efficiently, and though later feeds improved the telescope's performance, they did not perform up to the level of a Gregorian reflector, the solution recognized as early as the 1960s. Named for the astronomer John Gregory, a Gregorian reflector is concave and placed above the prime focus of a telescope. In a Cassegrain system, the type used, for example, on the Goldstone DSS-14, the reflector is convex mounted below the prime focus.26
Designing the Gregorian optics was a daunting task. A Cornell graduate student had considered the use of Gregorian optics, an option also studied by the AFCRL's Antenna Laboratory.27 Frank Drake, director of the NAIC from 1971 to 1981, nurtured the Gregorian reflector idea and attempted unsuccessfully to gain financial support to gather together the necessary antenna expertise to submit a formal proposal to the NSF.28
Design of the Gregorian reflector did not begin until 1984, after Tor Hagfors became director of the NAIC in late 1982. After serving earlier as director of the Arecibo Observatory following the departure of Gordon Pettengill, Hagfors spent a number of years in Scandinavia building the EISCAT facility, before returning to Cornell to head the NAIC.29
EISCAT (European Incoherent Scatter Association) is a European consortium headquartered at Kiruna, Sweden. Inaugurated by the King of Sweden in August 1981, the EISCAT facility is a high-power radar installed at sites in Norway and Finland for the study of the Earth's ionosphere, upper atmosphere, and magnetosphere at high latitudes. Germany, France, and the United Kingdom bore the greatest share of its construction costs (25 percent each), while Sweden (10 percent), Norway (10 percent), and Finland (5 percent) contributed the rest.30
Under the direction of Tor Hagfors, the NAIC initiated systematic studies of several major antenna upgrading projects in 1984. As part of the upgrading project, the NAIC concluded consulting agreements with a number of antenna experts. Among them were Alan Love, who had designed the telescope's first circular feed, and Sebastian von Hoerner. Morton S. Roberts, director of the National Radio Astronomy Observatory (NRAO), and a member of the Arecibo Advisory Board, suggested that the NAIC hire as a consultant von Hoerner, a well known antenna expert working for the NRAO. The project appealed to von Hoerner's imagination, and he set to work designing the Gregorian optics and laying out the initial description of the shape and size of the reflector. He also realized the need for a tertiary reflector.31
In addition, Hagfors brought in Per-Simon Kildal, a professor at Chalmers University of Technology, Gothenburg, Sweden. Kildal was an expert in the design of feed horns and antenna diffraction effects and a former student of Hagfors. He had performed some of the design work on the EISCAT antennas for his doctoral thesis. When Kildal worked for the NAIC for two months during the summer of 1984, he joined NAIC line feed designer Lynn A. Baker. Baker and Kildal devised a practical Gregorian design to correctly illuminate the primary reflector.32
 Designing and installing the Gregorian reflector also changed the mechanical stress on the suspended platform. In order to work on the mechanical engineering aspects of the project, Hagfors asked Paul Stetson, an antenna builder formerly with Lincoln Laboratory, to come out of retirement. Stetson joined the NAIC in February 1984.33
As a test of the Gregorian feed concept, the NAIC at its own expense constructed and installed a so-called mini-gregorian antenna which was to illuminate a 107-meter (350-ft) diameter area of the reflector. Also, the ground screen underwent preliminary design, and another study determined that the dish surface could be adjusted to be operational up to 8 GHz (3.75 cm).34
In 1984, as these design studies were underway, the NAIC submitted a preliminary proposal to the National Science Foundation for Phase 1, the ground screen. The NAIC submitted the Phase 2 preliminary proposal in 1985 for the Gregorian reflector system, the new radar transmitter, ancillary receivers, and data processing equipment. The NAIC then entered into negotiations with both the NSF and NASA, the two NAIC funding agencies. The House subcommittee that handled NSF appropriations was well aware of the upgrade project. Jerome Bob Traxler (D-Mich.), the chairperson of the House subcommittee, Harry Block, the NSF director, and Dick Mallow, the subcommittee's chief of staff, visited Arecibo several times.35
The key to selling the project to the scientific community, which ultimately reviewed all NSF proposals, was the building of consensus, a standard strategy among American scientists. The NSF proposals were supposed to stand on their own merit. Whether those reviews were good or bad was critical to the success of the upgrade project. The keystone of consensus-building was a workshop held at Cornell University 13-15 October 1986. The NSF proposal for Phase 1 was already under review, when the workshop took place. Talks highlighted the kinds of scientific experiments one could do with the upgraded telescope, whether in atmospheric research or in radio astronomy. Steve Ostro, Don Campbell, and Irwin Shapiro pitched the possibilities for radar astronomy.
Ostro largely proposed research on mainbelt and near-Earth approaching asteroids, passing quickly over other solar system objects, such as the moons of Mars, Jupiter, and Saturn. Don Campbell emphasized exploration of the terrestrial planets and comets. The major impact of the upgrading, he and Shapiro acknowledged, would be on the observation of asteroids.36 The scientific repercussion of the Arecibo upgrade for radar astronomy would be to sustain the observatory as the major research instrument and to make asteroid studies the predominant area of research.
The NSF sent the NAIC upgrade proposals out for review. The reviews aided the NSF in prioritizing its spending. Where the project stood within the NSF's own priority list of projects also was subject to input from the Division of Astronomy, primarily, and from the  Division of Atmospheric Sciences. Within NASA, the planetary program decided funding priorities. In 1988, following the Cornell workshop, the NAIC submitted the main proposal for the Gregorian system and radar transmitter. Numerous discussions, presentations, committee meetings, and reviews followed. Also providing input was the Bahcall Committee, the successor to the Whitford Panel.37
The Bahcall Committee, named for its chair John N. Bahcall, Princeton Institute for Advanced Study, and formally known as the Astronomy and Astrophysics Survey Committee, was a group of 15 astronomers and astrophysicists commissioned in 1989 by the National Academy of Sciences to survey their fields and to recommend new ground and space programs for the coming decade. To carry out the actual work, the Committee established 15 advisory panels to represent different subdisciplines, and those panels submitted their reports in June and July 1990.38
Radar astronomy came under the general umbrella of the Planetary Astronomy Panel, chaired by David Morrison, NASA Ames Research Center, chair, and Donald Hunten, University of Arizona, vice chair. Among the 22 planetary scientists constituting the panel was one radar astronomer, Steve Ostro. The Planetary Astronomy Panel recommended several facilities as "critically important" for planetary astronomy in the 1990s. Prioritized according to their cost (small, medium, large) within the categories "space-based" and "ground-based," the most important small ground facility for planetary astronomy was the Arecibo upgrade.39
The upgrade was never regarded as a huge project. The total estimated price tag of the upgrade, around $23 million spread out over four years, placed it in the "small" category; even the medium-sized proposed facilities cost substantially more. The relatively small total amount underwent further diminution in such a way that the project was never big enough to be a separate line item within the budget of the Office of Management and Budget. Both NASA and the NSF split the total cost, which underwent further division within each agency, so that the total amount per year was never a huge sum for each agency or for each agency program.
Geoff Briggs, director of the Division of Solar System Exploration within the NASA Office of Space Science, chaired discussions about the project with the NAIC, NASA, and the NSF. According to Don Campbell, "Briggs somewhat arbitrarily just took it on himself to break up who was going to pay for what right there."40
The allocation of the costs of what was already considered a small, low-cost project was a strategy in tune with the budgetary times. NASA would pay 100 percent of the ground screen and the one-megawatt radar transmitter costs, but the money came from the budgets of three different divisions. The Division of Solar System Exploration paid for the ground screen; the Office of Space Communications paid for the transmitter; and the Division of Biological Sciences, the source of SETI (Search for Extra-Terrestrial Intelligence) funding, contributed partially to the Gregorian reflector. The NSF paid for the remainder, with the Division of Astronomical Sciences paying for some specific equipment. The distribution of individual program contributions split the cost evenly between the two agencies and became the basis for the memorandum of understanding between NASA and the NSF that covered the upgrade.41
The Arecibo upgrade, when completed, promises entirely new research capabilities that will open up a new set of targets to be explored and new problems to be solved. Another upgrade, though not intended to provide new radar capability, created a research instrument that never existed before. That was the Voyager upgrade. It involved improvement of the GSSR, as well as the Very Large Array (VLA), a radio telescope located in New Mexico. For the VLA upgrade, NASA installed low-noise X-band receivers on each of the 27 VLA antennas. When radar astronomers linked the Goldstone radar and the VLA in a bistatic mode, they created a radar with an extraordinary capacity for exploring the solar system.
The upgrade of the VLA for the Voyager mission originated in the need to communicate with the spacecraft at unprecedented distances. During Voyager's encounter with Neptune, its X-band radio signal would be less than one-tenth as strong as from Jupiter and less than one-half as strong as from Uranus. In addition to the enlargement of the DSN 64-meter antennas to 70 meters in diameter, the Neptune encounter required assistance from the Parkes telescope in Australia and the VLA. Through the radio astronomy technique of arraying, and the installation of low-noise receivers on each VLA dish, the echoes received from the VLA were combined with those received at the Goldstone 70-meter and 34-meter dishes to provide a data rate more than double that which would have been available with Goldstone's antennas alone.42
The idea of using the VLA as a receiver in a bistatic radar system was not new; Ed Lilley had suggested some two decades earlier a bistatic radar consisting of the VLA and the NEROC transmitter for carrying out planetary radar mapping.43 Moreover, the VLA management already had thought of the possibility of a Goldstone-VLA bistatic radar years earlier, when they were looking for a broader foundation of support for a facility strictly dedicated to radio astronomy. They, therefore, were receptive to the suggestion of Nick Renzetti (JPL) that joint Goldstone-VLA radar experiments be conducted, provided the proposed experiments first would undergo the normal review process.44
As the Goldstone and VLA upgrades were underway, Caltech professor Dewey Muhleman became interested in the possibilities opened up by a Goldstone-VLA bistatic radar. After abandoning a career in radar astronomy in 1966 as professor of planetary science at Caltech, Muhleman switched to the study of radio emissions from the planets. Muhleman thought the Goldstone-VLA radar an excellent tool for exploring Saturn's barely explored and poorly understood moon, Titan. Scientists knew nothing about Titan's surface, because like the surface of Venus, it is hidden by an opaque cloud cover.45
Despite, or perhaps because of, this lack of knowledge, scientists speculated on the nature of the satellite's surface. According to conventional wisdom, Titan's surface was an ocean of ethane and methane, which would have almost no reflecting surface at radar wavelengths.46 In 1980, Voyager 1 flew past Titan and provided fresh facts about the moon's surface temperature (about 94° Kelvin) and surface pressure (around 1,500 millibars). Voyager found an atmosphere composed mainly of nitrogen and trace amounts of  hydrocarbons and nitriles, including ethane, methane, and acetylene. But Voyager revealed nothing about the moon's surface features.47
Titan's surface remained hidden from the view of radar astronomers, too. In February 1979, using the Arecibo S-band radar, Don Campbell, Gordon Pettengill, and Steve Ostro unsuccessfully attempted to detect Titan. Later, in 1987 and 1992, Dick Goldstein and Ray Jurgens also failed to receive echoes from Titan using the Goldstone Mars Station alone.48 The bistatic Goldstone-VLA radar, however, promised an extra measure of sensitivity.
Muhleman hoped to find land masses and challenge the ethane ocean model. He already had conducted a radio study of Titan, but that research had yielded ambiguous results. Muhleman teamed up with JPL radar astronomer Marty Slade, who oversaw operation of the Goldstone half of the bistatic radar. Muhleman's graduate students, Bryan Butler and Arie Grossman, participated in the experiments, too. In order to test the system, Muhleman, Slade, and Butler attempted a known target, the rings of Saturn, in the spring of 1988. The success encouraged them to attempt Titan.49
Muhleman, Butler, and Slade first observed Titan on the nights of 3, 4, 5, and 6 June 1989 with the VLA in the so-called C configuration, in which the maximum separation among the 27 25-meter (82-ft) telescopes was about three km. The echoes were marginal, although those obtained on 4 June were strong, and the detection of 5 June was "quite certain." "The data," they concluded, "appear to favor a real variation in surface properties but more observations are required."50
The backscatter from Titan was highly diffuse, similar to that from the Galilean satellites of Jupiter. The diffuse backscatter, they believed, was a strong argument against an ethane ocean being the reflecting medium. A liquid body without floating scatterers would be a specular not a diffuse reflector. Instead, the radar echoes from Titan suggested an icy surface similar to that of Europa, Ganymede, or Callisto. The experiment, however, did not rule out entirely the existence of liquid hydrocarbons on Titan's surface that might exist in the form of small lakes.
Muhleman, Slade, and Butler attempted Titan again in August 1992 and in the summer of 1993.51 From these fresh echoes, they concluded that Titan does not always keep the same hemisphere towards Saturn, as had previously been believed. In addition, one region very bright to the radar consistently appeared 15 hours earlier than expected, suggesting that its rotational period was 49 minutes shorter than its orbital period of 15.945 Earth days.
More importantly, variations in radar reflectivity gave the first indications of surface conditions on Titan. Results from instruments on the Voyager spacecraft in the 1980s suggested that there might be a global ocean of liquid ethane. However, Muhleman, Slade, and Butler reported that only a few patches of liquid will be found by the European-built Huygens probe scheduled to land on Titan early in the next century after a journey  aboard the Cassini spacecraft. The moon's surface seems to be covered mainly by icy continents, perhaps coated in tars of other hydrocarbons.
The results of Muhleman's radar research on Titan were of enormous interest to Dennis L. Matson, Cassini project scientist, and others involved in the planning of the Cassini mission. In 1989, NASA was preparing the Cassini Announcement of Opportunity for release on 1 December 1989. A major experiment on Cassini, as then planned, was a radar instrument to be built by JPL. The nature of Titan's surface was a major parameter in the design of any radar system for the Cassini mission.
If an ocean of ethane and methane really covered Titan, the radar would have to be designed to anticipate the special scattering conditions that such a surface would create. The Goldstone-VLA radar data, then, would be useful in targeting the Huygens probe, and the targeting decisions had to be made before the launch of the Cassini spacecraft itself.52 As Nick Renzetti characterized the situation: "So why put a $20 million radar on Cassini and get zilch? That really stirred the community for the last three years."53
The radar results from Titan were revealing but puzzling. The radar study of Titan also highlighted the continuing mission-oriented nature of radar astronomy. The same was true of Mars radar research. Although Muhleman intended to use the Goldstone-VLA bistatic radar primarily to study Titan, equally startling results came from its application to Mercury and Mars. The Goldstone-VLA system allowed radar astronomers to solve problems previously unsolved or solved unsatisfactorily. The Goldstone-VLA work added to a long tradition of studying Mars topography that began, as we saw in an earlier chapter, before Viking went to Mars, and continued in support of the Viking mission. Most of the Mars radar topography work done in the 1970s, in fact, related directly to Viking.
The exploration of Martian topography and radar reflectivity from the 1970s into the 1980s had yielded some rather interesting results. The studies done for Viking had revealed high roughness (large rms slopes) and sharp roughness transitions in the area around the Tharsis volcanoes and their associated lava flows. Tharsis itself was found to have a low overall reflectivity. The most unusual and controversial development was the claim by Stan Zisk and Peter Mouginis-Mark, from their analysis of Goldstone Mars data from 1971 and 1973, that the Solis Lacus region showed seasonal variations in its radar reflectivity which might indicate the presence of near-surface liquid water.54
The Tharsis and Syrtis Major regions were of special radar interest. Syrtis Major was a classical radar dark spot on Mars. From topographical data, George Downs showed that the Tharsis bulge was lower than originally thought. Geologists used the radar data to show that Tharsis had been tectonically inactive since the occurrence of the last major lava flows. Topographical data for the south Tharsis region suggested that it was an ancient impact basin. Interpretation of the radar studies of Downs and Simpson (at Arecibo) of the Syrtis Major area by USGS geologist Gerry Schaber indicated that it was a low-relief  shield volcano, rather than the impact basin it had always been believed to be, because it was not very heavily cratered.55
John Harmon arrived shortly after the installation of the Arecibo S-band radar as a Research Associate, after graduating from the University of California at San Diego with a doctoral thesis on solar winds. John Harmon began a series of studies of Mars topography and scattering, initially under the direction of Don Campbell, and drew a the first topographic profile of Syrtis Major. Starting in February 1980, Harmon and Steve Ostro undertook a study of Tharsis and the surrounding area using both the Arecibo S-band and the Goldstone X-band radars and taking data in both senses of circular polarization, in order to compare polarization ratios at both S-band and X-band.
While the initial focus in 1980 had been on the Tharsis region, the 1982 observations took in a broader area and revealed correlations between maximum depolarization and the volcanic regions Tharsis and Elysium, while the heavily cratered upland terrain yielded relatively low depolarization. This led to the suggestion by Harmon and Ostro, and confirmed independently by radar astronomer Tommy Thompson and USGS Menlo Park geologist Henry J. Moore, who used Goldstone data, that most of the strong sources of diffuse and depolarized backscatter on Mars were rough-surfaced lava flows.56
Such was the state of radar studies of Martian topography and scattering, when Muhleman, Butler, and Slade began looking at Mars with the Goldstone-VLA bistatic radar in 1988. The proximity of Mars, in contrast to the great distance to Titan, allowed them to construct full-disk images of the planet. During the 1988 Mars opposition, moreover, the Earth and Mars were closer than they had been for 17 years.
These images were not the product of radar range-Doppler techniques, but of standard VLA radio astronomy imaging software. The array and its software avoided the problem of north-south ambiguity that typically plagued planetary range-Doppler mapping; the VLA radio imaging software, which Muhleman regularly used in his planetary radio astronomy research, created unambiguous images. In this bistatic imaging mode, the Goldstone radar illuminated the target with a continuous-wave signal whose frequency was adjusted to remove the Doppler shift. When the VLA aimed at a target, the signal came from all over the planet, as though the target were a natural emitter of radio waves. Then the powerful imaging software of the VLA processed these echoes.
Muhleman, Butler, and Slade observed Mars twice during the opposition of 1988 and three times during the opposition of 1992-1993. They obtained surface resolutions of 80 km at the subradar point. The Mars observations differed from those of Titan, because for Mars the VLA A-array (36-km maximum spacings) was used. The transmitted signal to  Mars was circularly polarized and both opposite circular and same circular echoes were received and mapped. As anticipated, the opposite circular echoes were dominated by the so-called specular (or phase-coherent) reflections.
Muhleman and Butler found regions with anomalously high radar cross sections on Mars, particularly around the three Tharsis volcanoes and Olympus Mons. These, Muhleman recalled, "just lit up like a Christmas tree." In contrast, the region west of Tharsis, extending over 2,000 km in the East-West direction and 500 km across at its widest point, displayed no cross section distinguishable from the noise in either polarization. "We didn't believe that result. We've never seen that on any real surface," Muhleman explained.57
Muhleman dubbed the area "Stealth," because it was invisible to the radar. Photographs do not indicate the nature of the Stealth region. Muhleman interpreted the lack of radar echo as arising from a deposit of ash or pumice spewed from the bordering Tharsis volcanoes and carried by winds blowing off the Tharsis ridge. He estimated that the Stealth material would have a density of less than about 0.5 grams per cubic centimeter, be free of rocks larger than one centimeter across, and have a depth of at least five, if not ten, meters.
Equally surprising was the radar signature of the residual southern polar ice cap. The 1988 observations were made in the southern hemisphere around -24° latitude in late spring, so the seasonal carbon dioxide ice cap had sublimated away and exposed the residual southern polar ice cap. That area had the highest radar cross section of any other area observed on the planet in 1988. Furthermore, the residual ice cap exhibited strong circular polarization inversion. Thus, unexpectedly, part of one of the terrestrial planets displayed radar characteristics more typical of the Galilean satellites.
When Muhleman, Butler, and Slade looked at the VLA images, they "instantly saw that the brightest thing on the planet was the South pole, which turned out to be the residual South polar ice cap," Muhleman recalled. "The amazing thing to us was that this ice was so reflecting, so bright, and its size was exactly the residual polar cap."58 Also amazing was the fact that Dick Simpson and Len Tyler had failed to notice any unusual scattering properties from the North pole in data from a bistatic radar experiment conducted from the Viking spacecraft.59
Butler, Muhleman, and Slade again looked at Mars with the Goldstone-VLA radar during the 1992-1993 opposition, when the planet's North pole was visible from Earth. It was early northern spring on Mars, and much of the seasonal carbon dioxide polar ice cap was present. They were anxious to study the northern polar ice cap, but the ice was invisible to the radar. In stark contrast to the southern pole, no regions with enhanced radar cross sections appeared. "We still haven't figured that out," Muhleman admitted. "It's totally a mystery why we didn't find the residual North polar ice cap."60
The high radar cross section and polarization inversion of the Martian South polar ice cap were confirmed by observations made at the Arecibo Observatory during the 1988 opposition by John Harmon, Marty Slade, and R. Scott Hudson. Hudson was a Caltech graduate student working on a doctoral degree in electrical engineering and had chosen aircraft radar imaging as his dissertation topic. Like those made by Harmon and Ostro in  1980 and 1982, these were monostatic, dual-polarization continuous-wave observations made with the Arecibo S-band and the Goldstone X-band radars.
After obtaining promising results from a comparison of the 1988 data at both wavelengths, an additional set of observations were made at S-band and X-band during the 1990 opposition. Despite scheduling difficulties and the demands of competing types of radar observations (ranging observations for altimetry and mapping were also made at the two facilities), a good continuous-wave data set for S/X-band comparison was obtained in 1990.
The Arecibo data confirmed the existence of Stealth. Using an algorithm developed by Scott Hudson and the Doppler spectra taken in the unexpected sense of polarization, they produced depolarized reflectivity maps that showed clearly the anomalously high radar reflectivity and polarization inversion of the residual South polar icecap. Hudson's algorithm allowed the investigators to use only Doppler spectra, without range measurements, to create a two-dimensional map of the Martian disk largely free of north-south ambiguity.61
Hudson's imaging technique was necessary in order to overcome the planet's overspread nature. In comparison to Venus, Mars rotates rapidly on its axis and causes radar echoes from the limb (beyond the subradar area) to disperse broadly. The echo delay corresponding to the radius of Mars is 22.6 microseconds, which is much greater than the maximum interval of 0.725 microseconds needed to preserve complete spectral information over the band of frequencies present in the echo. As a result, when the computer samples signals, echoes from different ranges contaminate each other and become indistinguishable. Such radar targets are called "overspread."
Arecibo scientists also had a technique for overcoming the overspread problem, but they were not motivated to apply it until the Goldstone-VLA results became known. Harmon explained: "Dewey Muhleman, with his VLA experiment, spurred us on to try and do better. I really hadn't been thinking about the overspreading problem. I probably should have; I should have been trying to figure out ways to get around it."62
Overspreading was a problem that ionospheric scientists had been dealing with for years, because the ionosphere is an extremely overspread target. Michael P. Sulzer, an ionosphericist at the Arecibo Observatory, solved the problem for the ionosphere by using non-repeating codes. Although Don Campbell at one time had asked Sulzer to think about applying the technique to Mars, no progress had been made until Harmon told Sulzer he was interested in trying the non-repeating code technique.
Harmon then worked with Sulzer and Phil Perillat, who wrote the modified data-taking program. Normally, when a continuous-wave radar sends out a signal, the signal carries a code with a finite number of elements, and the code repeats at a regular interval. Harmon and Sulzer tested the non-repeating code, called alternately the "random code" or "coded long pulse" technique, and it worked the first time. Then Harmon wrote programs to do the data analysis.
Harmon and Sulzer made their first random-code observations on 18 nights during the Mars opposition of September-December 1990 and created range-Doppler maps. Those maps, like all range-Doppler maps, included a north-south ambiguity around the Doppler equator. However, from eyeball comparisons with maps obtained early and late in the opposition, Harmon was able to resolve much of the ambiguity.
 The random-code maps confirmed the observations made with the Goldstone-VLA radar and revealed new information about the Elysium region, which Harmon had spent a long time studying in previous observations of Mars. Through those and subsequent observations made during the 1992-1993 opposition, he discovered strong depolarized radar echoes from the Elysium/Amazonis outflow channel complex. He interpreted the region, which was very young by Martian standards, as having lava flows that appeared to have partially filled pre-existing channels cut by flowing water.63
The strange radar signature exhibited by the southern residual polar ice cap of Mars, reminiscent of the radar characteristics of the icy Galilean satellites of Jupiter, did not prepare Muhleman, Butler, and Slade for the surprising discovery of ice on Mercury. Mercury was simply too hot to support even the smallest ice deposit. Previous radar observations of Mercury had focused on scattering and topography and had not detected ice.
Analysis of Mercury data taken between 1963 and 1965 at Goldstone, Haystack, and Arecibo showed the planet to have a radar roughness "very similar" to that of the Moon. Dick Goldstein, from radar observations of Mercury made in 1969, started characterizing Mercury's topography. His work was the most detailed radar study of Mercury's surface prior to the Mariner 10 encounters and provided the first strong evidence for the existence of craters on the surface. Dick Ingalls at Haystack and Don Campbell at Arecibo also found altitude variations on Mercury's surface from 1971 observations. Some of the earliest topographic radar studies of Mercury were carried out at Haystack by Bill Smith and Dick Ingalls.64
These early radar studies of Mercury were not linked to any specific NASA mission, but not because of any radar shortcomings. NASA made no meaningful effort to study Mercury until Mariner 10 flew by and photographed that planet in 1974-1975. The Mariner 10 photographs revealed a heavily cratered, lunar-like surface, as predicted by radar. Although Mariner 10 photographed over half of Mercury's surface during its flyby mission, it did not photograph any of the side not then exposed to the Sun's light and yielded only limited topographic information. Moreover, its flyby geometry prevented Mariner 10 from examining either pole directly.65
These gaps in Mercury coverage motivated a program of observations at Arecibo and Goldstone. John Harmon and Don Campbell, working in collaboration with Brown University geologists D. L. Bindschadler and James W. Head, carried out a campaign of  S-band radar observations of Mercury at Arecibo from 1978 to 1984. They measured Mercury's topography over much of the equatorial zone (between 12° North and 5° South latitude), an area not imaged by Mariner 10, and concluded that radar depths for large craters supported previous indications from photographs that Mercury's craters were shallower than lunar craters of the same size.66 At the same time, Ray Jurgens, using the Goldstone S-band radar, started an ongoing series of Mercury observations to study the planet's topography and to correlate radar measurements with Mariner 10 visual images, in collaboration with geologists Gerald G. Schaber (USGS Flagstaff) and P. E. Clark (JPL).67
Such was the state of radar research on Mercury, when Muhleman, Butler, and Slade began their observations with the Goldstone-VLA bistatic radar during the inferior conjunction of August 1991. Although they made further observations during the inferior conjunctions of November 1992 and February 1994, the 1992 effort failed because of transmitter problems, and the 1994 data yet remains to be reduced.68 The key results, then, were those from the 1991 observations. They did nothing less than revolutionize our knowledge of Mercury in a way that radar had not done since the discovery of the planet's 59-day spin rate by radar astronomers Gordon Pettengill and Rolf Dyce in 1965.
During the first Goldstone-VLA observation of Mercury on 8 August 1991, Ray Jurgens coordinated activities at the Goldstone X-band transmitter, while Marty Slade and Bryan Butler awaited the echoes at the VLA, which was operating in the so-called A-array, the most widely spaced configuration. During the 10 hours of observation, the VLA received in both senses of circular polarization. At the time of these observations, Mercury was at inferior conjunction and presented the hemisphere not photographed by Mariner 10, roughly between 180° to 360°, to the radar. As a result, the subradar point was far enough North to see over the North pole and into areas believed to be permanently shadowed from the Sun.
When Muhleman, Butler, and Slade looked at their results, they were astonished; they had found ice near Mercury's North pole. What signalled the presence of ice was the abnormal radar signature of the spot, which was unusually bright and showed a ratio of same circular to opposite circular polarization greater than unity, that is, a circular polarization inversion. This was the same type of radar signature displayed by Jupiter's Galilean moons. Muhleman recalled: "We instantly looked at the first image and saw this white spot on the North pole. We said, 'My God! Are we going to find an ice cap on every planet we look at?' This is crazy!" Marty Slade remembered looking at the bright spot and reacting: "It's not possible that could be ice! It's too hot!"69
Muhleman, Butler, and Slade again observed Mercury with the Goldstone-VLA radar two weeks later on 23 August 1991. This time, they transmitted both right-handed (RCP) and left-handed circular (LCP) polarization, and they received in both senses of polarization for either sense, so that they could make all four correlations of the two polarizations (RCP to LCP, RCP to RCP, LCP to RCP, and LCP to LCP). Mercury as seen from Earth had rotated 101°. The subradar point was around 353° and the ice near the northern polar  still stood out brightly and exhibited polarization inversion. The researchers now knew that this was no fluke.70
Surprised by their own results, Muhleman, Slade, and Butler announced their results in two separate talks given on 6 November 1991 at the meeting of the AAS Division for Planetary Science, held in Palo Alto, California.71 The scientific community greeted the news of their discovery with a fair amount of skepticism.72 Prior to the launch of Mariner 10, few had suggested the presence of ice on Mercury, and then for the wrong reasons. Some drawings of Mercury showed a white spot visible at the northern pole, and in 1974, on the eve of Mariner 10's first reconnaissance of Mercury, an atmospheric scientist had proposed that ice could have accumulated in the small planet's polar regions, perhaps in permanently shaded regions.73
The evidence for the presence of ice near Mercury's northern pole was based on an analogy between the radar signatures of known icy targets, the Galilean moons of Jupiter, and those found on Mercury. But more convincing evidence was needed, because Mariner 10 had documented that planet's intense surface heat. The landscape was a parched wasteland of impact craters and volcanic plains, where midday temperatures soared to 700° K, hot enough to melt lead. At the same time, though, Mariner 10's ultraviolet spectrometer had identified traces of hydrogen and oxygen in the tenuous atmosphere of Mercury. Project scientists had considered them to be remnants of the comets and asteroids that periodically collide with the planet.74
While such collisions would explain the existence of water on Mercury, an explanation for the existence of a permanent water ice deposit on the planet came from a consideration of the geometry of Mercury's orbit. An impact crater could provide an area of permanent shade, provided that the geometry was just right. Mercury spins on its axis and rotates around the Sun in such a way that its equator always lies in the same plane as the Sun. As a result, neither pole ever sees more than a sliver of the Sun's disk above the horizon. On the other hand, the plane of Mercury's orbit about the Sun is inclined by seven degrees relative to that of the Earth, so that Earth-based radars can see into impact craters that are never directly illuminated by the Sun.
David A. Paige and Stephen Wood of UCLA recomputed the thermal environment for Mercury's surface and concluded that the interior slopes of impact craters within five degrees of the poles would be cold enough to keep the loss of water ice through sublimation at essentially zero. Other planetary scientists also began to argue for the existence of ice in craters on Mercury, and they suggested that craters on the Moon might also contain ice. As early as 1961, Kenneth Watson, Bruce C. Murray, and Harrison Brown had proposed that ice might exist in permanently shadowed craters near the lunar poles, but  to date no lunar probe, not even the Clementine orbiter, has found any ice on the Moon.75 A radar search at Arecibo also proved unsuccessful.
Nick Stacy, a graduate student working on a thesis in radar astronomy under Don Campbell, looked for ice on the Moon with the Arecibo radar. Earlier, starting in 1982, Don Campbell and Peter Ford had carried out high-resolution range-Doppler imaging of the Moon and found no evidence of ice, but they were not looking for it. Ford and Campbell brought the resolution of their images down from 300 to 150 meters, using the Higuillales antenna in a bistatic mode with the big dish. Stacy reduced the resolution to 20 meters and aimed at the lunar poles. Unfortunately, the radar could not see far enough into the polar craters and detected no ice, although Stacy found some unusual scattering properties around a number of lunar craters.76
Although the discovery of lunar crater ice remained elusive, John Harmon and Marty Slade at the Arecibo Observatory confirmed the existence of ice on Mercury. They imaged Mercury using the non-repeating code technique developed by Harmon and Mike Sulzer in order to overcome overspreading on Mars. These Arecibo images, according to David Paige, left "little room for doubt" about the presence of ice on Mercury.77
Soon after observing Mercury on 8 August 1991 with the Goldstone-VLA radar, Marty Slade travelled to the Arecibo Observatory to collaborate with Harmon on a different set of Mercury observations. They acquired their initial data prior to 8 August 1991, on 28 separate dates during the periods 28 March to 21 April 1991, 31 July to 29 August 1991, and 14 to 29 March 1992. During the spring 1991 observations, the subradar point of the Arecibo telescope subtended an area in the southern hemisphere of Mercury, while the summer 1991 observations covered a portion of the northern hemisphere, as the Goldstone-VLA had. The March 1992 data added to that already observed in the southern hemisphere.
When Slade arrived at Arecibo, his first time at the observatory, Harmon had not yet analyzed the spring 1991 data; he had been too busy studying Mars data. Slade suggested to Harmon that they analyze the Mercury data and look for the icy radar signature near the North pole, which he, Muhleman, and Butler had just found with the Goldstone-VLA radar. According to Harmon, Slade said, "We think it's the pole; we're not sure." The Arecibo data confirmed the Goldstone-VLA discovery. There was no question of priority; Muhleman, Butler, and Slade discovered the ice on Mercury first, with the Goldstone-VLA radar.
Harmon also examined the data collected from the southern hemisphere of Mercury in March-April 1991. "I saw a feature coming from what I figured probably had to be the South pole, because the latitude was about five degrees South [sic]," Harmon related. "I was pretty convinced it was coming from the South."78 To confirm that the South pole was the source of the icy radar signature and not an artefact of north-south ambiguity, which would have shown a portion of the northern polar echo at the South pole, Harmon and Slade observed Mercury again in March 1992, when the subradar point was again in the southern hemisphere. The polar ice feature was seen again, confirming the presence of ice at the planet's South pole.79
 Next, Harmon and Slade proceeded to fit the radar results to photographic data from Mariner 10. Showing a correlation between a known crater and the radar ice would be persuasive confirmation of the discovery. Matching the northern polar radar ice location with a crater was hard; no Mariner 10 photographs were available for the entire region. Furthermore, the North polar radar anomaly was too large to fit within a single crater. The image, instead, appeared to consist of a number of crater-size (15-60 km in diameter) bright spots. Harmon and Slade plotted those features on a locating map created by NASA and the USGS and assigned letter labels to those features that lay in the photographed hemisphere and to three prominent features in the unphotographed hemisphere. Many of the radar spots (8 out of 20) appeared to correspond to impact craters. Correlating the southern polar radar image with topography was simpler. The radar spot was entirely inside a crater called Chao Meng-Fu.80
The Goldstone-VLA and Arecibo images of Mercury once again highlighted how planetary radar astronomy often solves problems left unsolved or unsatisfactorily solved by optical techniques. The discovery of ice near Mercury's North and South poles, moreover, has inspired the European Space Agency to mount a major "keystone" mission to Mercury in search of polar ice, as well as a more modest-sized NASA Discovery flight.81
Radar astronomers also sought signs of anomalous radar signatures on other terrestrial planets. Muhleman, Butler, and Slade turned the Goldstone-VLA radar on Venus twice, 18 and 25 February 1990, receiving both senses of polarization in order to detect any peculiar polarization inversion, and made two maps. The maps had several striking features. Surprisingly, Alpha Regio had a high unexpected (depolarized or SC) reflectivity on both maps and contained the second highest reflectivity values after Maxwell. On the second day's map, the point of highest reflectivity was in the Aphrodite region and was not visible in the previous map. On both maps, many very small areas, only a few pixels across, also had large unexpected (depolarized or SC) reflectivities, and some of them corresponded to mapped elevated areas such as Gula Mons, Sif Mons, and Bell Regio. Muhleman, Butler, and Slade concluded that a correlation existed between unexpected (depolarized or SC) reflectivities and elevation. Further bistatic observations of Venus in the spring of 1993 furnished fuel for another Muhleman graduate student, Albert Haldeman, to begin doctoral research, while Slade and Ray Jurgens also found highly reflective areas on Venus using just the Goldstone radar.82
Throughout the 1980s and into the 1990s, the number of asteroids discovered and the number of publications dealing with asteroids grew at an unprecedented rate, at first as a result of the Palomar Planet-Crossing Asteroid Survey studies initiated in the 1970s, then as the number of asteroid researchers swelled. In 1932, an astronomer discovered the first Earth-crossing asteroid, 1862 Apollo. By 1994, about 200 Earth-crossing asteroids were known, more than half of which had been discovered in the previous seven years; yet  the undiscovered population is huge. In the decade 1975-1985 alone, the total number of catalogued asteroids rose from 2,000 to more than 3,200.83
The field, as measured by the expanding literature, was undergoing the kind of swift growth that is typical of Big Science. Asteroid astronomy became a new theoretical framework with problems that radar astronomers sought to solve. Radar found its niche within asteroid astronomy because it could solve problems that other observational techniques could not do, namely, the creation of more accurate and reliable ephemerides and the imaging of asteroids.
The focus of asteroid research was on near-Earth asteroids, although main belt objects remained of interest, too. Near-Earth asteroids, like meteorites, are thought to come primarily from mainbelt asteroids (Table 8). A large population of asteroids also cross the orbits of Earth and Mars. The term near-Earth asteroid usually means any asteroid that can come close to the Earth, whether or not it crosses the orbit of the Earth. Eros, for example, crosses the orbit of Mars, but it is not an Earth-crossing asteroid and does not come near the Earth. Almost all of the near-Earth asteroids detected so far by radar are Earth-crossers.
 The more interesting near-Earth asteroids also were better radar targets than main belt asteroids, because now and then they come closer to the Earth. With targets as small as asteroids, some only a kilometer or two in diameter, the distance to the target is critical to radar observations. The number of asteroids observed by radar astronomers grew rapidly during the 1980s because of the availability of radars with sufficient power and sensitivity to detect and study them. Another key factor in the growth of radar asteroid studies was the decision of one radar astronomer, Steve Ostro, to begin studying asteroids almost exclusively. Quickly, his efforts dominated the asteroid study started at Arecibo and Goldstone in the 1970s.
Before beginning this intense study of asteroids, Ostro had been making radar observations of the Galilean moons and the rings of Saturn. In March 1979, about the time of Voyager's encounter with Jupiter, Ostro attended the third Tucson asteroid conference organized by Tom Gehrels. There, Ray Jurgens and Gordon Pettengill delivered a joint paper on radar observations of asteroids. The conference, especially the talks that placed the science of meteoritics and asteroid science in context with each other, gave Ostro the asteroid bug. He saw how the study of asteroids was essential to understanding the origin and evolution of the solar system. He also realized that radar was potentially the primary post-discovery technique for observing asteroids, and that asteroids, unlike planets and their moons, constitute a huge and diverse population.84
Later in 1979, his MIT dissertation completed, Ostro took a teaching position at Cornell University and began preparing a campaign of asteroid observations at Arecibo. The following year, he submitted his first NASA proposal for support of asteroid research. Echoing the work of Jurgens a few years earlier, Ostro laid out those asteroid opportunities that would become available over the forthcoming decade at Arecibo, as well as the kinds of information he expected from his experiments. As targets, Ostro proposed three main-belt asteroids (Iris in September 1980, Psyche in November 1980, and Vesta in February 1981) and two Earth-crossing asteroids (1862 Apollo in November 1980 and 1915 Quetzalcoatl in March 1981). He planned to detect echoes from each target, estimate echo strength, and measure polarization, spectral bandwidth, and Doppler shift. From those four quantities, Ostro proposed to estimate asteroid size and rotation, place constraints on the composition and structure of asteroid surfaces, and improve knowledge of their orbital parameters.85
Over the following years, the estimation of asteroid physical properties and the determination and refinement of their orbits remained fundamental aspects of Ostro's radar studies of asteroids. He systematically took range and Doppler data on all asteroids, as well as polarization measurements (receiving in both the expected and unexpected senses) in order to best estimate their surface roughness and structure. From measurements of the surface's reflectivity came estimates of the bulk density of the surface, its porosity, and relative metallic composition. With each observation, Ostro tried to contribute to scientific knowledge about asteroids.
Ostro also studied mainbelt asteroids. "Virtually every experiment gave an interesting result, and each radar signature was different," Ostro recalled. "Every single experiment was lucrative."86 By 1992, Ostro had observed 28 near-Earth and 36 mainbelt asteroids. Between 1980 and 1985 alone, he made dual-polarization observations of 20 mainbelt asteroids at Arecibo. These objects had low circular polarization ratios (the ratio of unexpected to expected echo power) ranging from about 0.00 to 0.40. The lowest  value, 0.05 ± 0.02 for the asteroid 2 Pallas, required that nearly all the echo arise from single-reflection backscattering from very smooth surface elements.
"It became clear," Ostro explained, "that the mainbelt asteroids had a dispersion of reflectivities and polarization ratios. This was evidence for diversity in surface structure and in surface bulk density."87 The data collected helped to characterize asteroid surfaces at scales between several centimeters and several kilometers and furnished constraints on surface bulk density and metal concentration, beyond those constraints obtained by optical methods.
The metallic composition of the asteroids was an interesting question relating to possible meteoritic analogues. The radar observations suggested wide variations in metal abundance, porosity, and decimeter-scale roughness on mainbelt asteroid surfaces, underscoring the diversity of the asteroid population already evident from visible and infrared wavelength studies. Although the radar signatures of mainbelt asteroids required substantial surface roughness at some scale much larger than a meter, Ostro could not discern the precise scale of this structure, much less the actual morphologies of surface features. Similarly, the radar albedos bolstered the hypothesis that metal concentrations on asteroids run the gamut. Serious questions remain, however, about detailed mineralogies, meteoritic associations, and evolutionary histories.88
"Each of the near-Earth asteroids is interesting in its own way," Ostro pointed out, "and still some interesting mysteries remain."89 Echoes from the near-Earth asteroid 1986 DA showed it to be significantly more reflective than other radar-detected asteroids. This result supported the hypothesis that 1986 DA was a piece of nickel-iron metal derived from the interior of a much larger object that melted, differentiated, and cooled, and subsequently was disrupted in a catastrophic collision. This two-kilometer-sized asteroid appeared smooth at centimeter to meter scales but extremely irregular at 10- to 100-meter scales. It might be (or have been part of) the parent body of some iron meteorites. The composition of asteroids thus bears directly on the question of their relationship to meteorites, as well as the relationship between near-Earth and mainbelt asteroids.90
Starting in 1983, Steve Ostro began observing echo spectra with unusual shapes, including some spectra with double peaks (called bimodal). The first asteroid to show a bimodal spectra was 2201 Oljato, observed during 12-17 June 1983 at Arecibo. Asteroid astronomers had been discussing binary asteroids and contact-binary asteroids for a long time, but no evidence of their existence was at hand. 216 Kleopatra, a large mainbelt asteroid, exhibited a strong bimodal echo spectrum. "That almost definitely is a contact binary," Ostro explained. "But almost definitely is not definitely."91
Proof of the existence of binary and contact-binary asteroids eventually came from radar data. Finding that proof was a problem left unsolved by optical and other research techniques. To the telescope, the biggest asteroid looks like a little dot, its shape indiscernible.92 Radar succeeded in solving that problem through the development of new imaging and modeling techniques. The key to developing an appropriate technique, though, was to avoid simplistic models. Too, it was important that the asteroid approach Earth close enough to provide the Arecibo and Goldstone radars a sufficiently strong echo to resolve the target.
 Ray Jurgens developed the first modelling technique for describing asteroid shapes in the 1970s. He applied it to spectral data from Eros. Steve Ostro applied Jurgens' triaxial ellipsoid model to his 1980 and earlier 1972 Toro data and derived a rough description of the asteroid.93 Similarly, when he applied Jurgens' model to the Earth-crossing asteroid 2100 Ra-Shalom in 1981, Ostro found it to have a somewhat irregular shape.94
With researchers at Cornell, Ostro developed a different modelling technique, one that synthesized echo spectra acquired at different rotational phases of the asteroid into a convex envelope, called the hull, which represented the asteroid's silhouette as viewed from a pole. After he fit the hull model to Jurgens' Eros data, Ostro modeled the Earth-crossing asteroids 1627 Ivar and 1986 DA, observed in July 1985 and April 1986 at Arecibo. Interestingly, the hull estimates indicated that 1986 DA's hull was "extremely irregular, highly nonconvex, and possibly bifurcated."95
The case of 1986 DA suggested that any asteroid model had to accommodate the possibility that the target might not be convex; both Jurgens' triaxial ellipsoid and the hull models were inadequate. Ostro also attempted to image asteroids with range-Doppler mapping techniques, beginning with 1627 Ivar in 1985.
Range-Doppler mapping revealed a bimodal distribution of echo power, suggesting that the target was not convex. All previous images had relied exclusively on Doppler spectra data; these were the first range-resolved images of an asteroid. Nonetheless, they failed to define the asteroid's global shape.96
The next opportunity to attempt range-Doppler imaging came in 1988, with the close approach to Earth of the small asteroid 1980 PA. The technicians at Arecibo improved the telescope's data acquisition software and hardware, in order to improve resolution of the asteroid; the resolution of the Goldstone radar on the same target was still not fine enough. The Goldstone radar did not achieve the limit needed for radar asteroid observations until 1986, when the Voyager upgrades were completed.97
Ostro again attempted range-Doppler images, this time of 1989 PB, later known as 4769 Castalia. On 9 August 1989, Eleanor Helin discovered the object on photographic plates taken at Palomar Observatory. Orbital calculations two days later showed that the asteroid would pass through the Arecibo Observatory's declination window during 19-22 August and that at closest approach, Castalia would be only 0.027 astronomical units from Earth. These were ideal conditions for imaging the asteroid at Arecibo, though not so at Goldstone. Communications with Voyager 2, which was making its closest approach to Neptune, occupied the Goldstone 70-meter antenna; it was unavailable for use as a radar telescope until 30 August, when some observations of Castalia took place after closest approach. "At Goldstone," Ostro recalled, "everything was a disaster. We had an eight-hour track, and we got about 20 minutes of data."98
Getting time on the Arecibo antenna on such short notice (10 days after detection) was not a problem; Ostro already had time to observe Victoria, a mainbelt asteroid, which had shown a hint of a double-peak spectral structure in 1982. After doing a few runs on...
...Victoria, Ostro spent the rest of the time on Castalia. "We saw CW [continuous-wave] echoes instantly," Ostro remembered. "A few of them from the first day looked strongly bifurcated." This was the first echo signature that said "This is a contact binary." Although he had never claimed discovery of bifurcated asteroids in print, Ostro had seen the idiosyncratic radar signatures several times before.99
From the Doppler and range data, Ostro created 64 images of the asteroid with an average of two dozen pixels each. Each image was bifurcated and showed a bimodal distribution of echo power. Reading the sequence of images from left to right from top to bottom, one can see the asteroid rotate.
When Ostro presented the images at the AAS Division for Planetary Science meeting a few months later, they attracted a dramatic intensity of attention; it was no less than the first time that anyone had resolved the shape of an asteroid from Earth. "This was a major breakthrough, definitely a major breakthrough," Ostro reflected.100
 About a year after the imaging of Castalia, Scott Hudson started working on a mathematical modelling technique to reconstruct the asteroid's shape in three dimensions. While still a Caltech graduate student, Hudson had worked with Ostro in developing a technique for creating planetary Doppler images free of north-south ambiguity. Hudson devised a complex mathematical model with 169 parameters in order to capture the shape of asteroid Castalia. The resultant three-dimensional model showed indisputably that the asteroid was bifurcated into two distinct, irregular, kilometer-sized lobes.101
Modelling three-dimensional asteroid shapes from radar data provided further evidence for the existence of asteroids with exotic shapes upon the approach of 1989 AC (later known as 4179 Toutatis), discovered in January 1989. Because of its extremely close approach to Earth, 9.4 lunar distances on 8 December 1992, Toutatis showed high promise as a candidate for imaging. Ostro proposed a Toutatis experiment to the NAIC and to Nick Renzetti, explaining how extraordinary the opportunity was and urging that Goldstone also make observations.
Ostro planned to use both telescopes to take data in both senses of polarizations and to create range-Doppler images of several thousand pixels, considerably more resolution than had been achieved ever before. The high resolution was possible at Arecibo and Goldstone because of incremental improvements made in the data acquisition hardware and software over the preceding years. Ostro obtained continuous-wave echoes at Goldstone on 27 November, then range-Doppler images daily from 2-18 December 1992 and at Arecibo each day from 13-19 December. In addition to the routine monostatic observations, Ostro's team took advantage of new antennas recently made available. They observed Toutatis bistatically with the DSS-14 transmitting and a new 34-meter beam-waveguide antenna (DSS-13) 21 km away receiving, and on one day they received with both DSS-14 and DSS-13 to acquire interferometric data. On yet another day, they collected data with the Goldstone-VLA radar.
Preliminary analysis of the data showed Toutatis to have an unusually slow rotation rate and a maximum dimension of no less than 3.5 km. Interestingly, too, Toutatis appeared to consist of two irregularly-shaped components in close contact. The images provided a first glimpse of craters on an Earth-crossing asteroid, as well. The asteroid's roughness, as measured by the circular polarization ratio, indicated a considerable degree of general roughness at centimeter-to-decimeter scales, supporting the belief that Toutatis had undergone a complex collisional history.
Since then, Scott Hudson has elaborated his model to recreate three-dimensional asteroid shapes. With Toutatis, he was dealing with over 1,000 parameters. The application of Hudson's reconstruction technique to the Toutatis images was complicated by the asteroid's rotation. Unlike all other targets detected by radar, Toutatis was in a tumbling rotational state.
At the 1994 AAS Division for Planetary Science meeting in November 1994, Ostro and Hudson presented several movies of Castalia, including one in which the asteroid was portrayed as it might be viewed in space, complete with fictional optical illumination. The use of the older Castalia data was on purpose; it suggested the potential rewards of using higher resolution data. In addition to Castalia and Toutatis, Ostro and Hudson began working in 1994 on three-dimensional modeling of 1620 Geographos, an asteroid which the ill-fated Department of Defense's Clementine spacecraft was scheduled to observe during a flyby mission. Although a computer malfunction prevented the Clementine encounter, Ostro captured a detailed sequence of Geographos images at Goldstone only days before the scheduled flyby. Subsequent modeling of the data has yielded an impressive simulation of an asteroid flyby.102
The state-of-the-art imaging and modeling of Castalia, Toutatis, and Geographos are feats that only a spacecraft flying by an asteroid could match. Although no probe deliberately set out to photograph an asteroid, Galileo, on its voyage to Jupiter, sent back the first spacecraft images of an asteroid, mainbelt object 243 Gaspra, on 29 October 1991 from a distance of about 16,200 km. Interestingly, Galileo also discovered that mainbelt asteroid 951 Ida had an orbiting satellite, recently named Dactyl.103
While the Toutatis images in themselves are spectacular witnesses to the ability of radar astronomy to solve problems left unsolved by other techniques, radar astronomy has achieved an equally great degree of success in another problem-solving area, asteroid orbits. Determining asteroid orbits with better degrees of accuracy and predictive reliability gained higher attention as scientists and the general public came to perceive asteroids as an ultimate threat to human civilization and to life itself on Earth. The perception grew out of the work of nuclear physicist and Nobel laureate Luis Alvarez, who first proposed that an asteroid was responsible for the extinction of the dinosaurs some 65 million years ago. Since then, evidence supporting the theory has accumulated, though not without arguments and evidence questioning the theory.
In a seminal paper published in Nature, Clark Chapman and David Morrison argued that the probability of a kilometer-size asteroid hitting Earth in the next century was 1 in 5,000. The collision would have a global effect, regardless of the impact site, because the dust blasted into the stratosphere would end agriculture for several years. Billions of people would starve to death.104
In order to detect a potentially civilization- and life-threatening asteroid, the scientific community proposed Spaceguard, a network of six optical telescopes dedicated to detecting asteroids. The name Spaceguard came from the book Rendezvous with Rama, in which its author Arthur C. Clarke envisioned an asteroid striking Earth in northern Italy in the year 2077. In response to the impact's devastation, the nations of Earth formed Project Spaceguard.
Creating a real Spaceguard has not been so straightforward. After asteroid 1989 FC came very close to the Earth in 1989, the American Institute of Aeronautics and Astronautics recommended to the House Committee on Science, Space, and Technology that it sponsor studies of asteroid detection and defense. Congress then commissioned NASA in 1990 to write reports on those subjects.  NASA already had considered asteroid detection in a 1981 workshop held in Colorado, but 10 years later it acted in response to a Congressional mandate. The NASA 1991 workshop brought together 24 asteroid scientists from around the world, including radar astronomer Steve Ostro.105
Although Congress has not yet funded Spaceguard, a battle over how to defend the planet against a "killer asteroid" rages. The recent collision of Comet Shoemaker-Levy with Jupiter has driven home the point that the planets, Earth included, are susceptible to potentially life threatening impacts from comets and asteroids. The Spaceguard proposal came along just as the Department of Defense was seeking post-Cold War applications of its nuclear arsenal. The deflection of a menacing asteroid or comet with a series of nuclear explosions is, in the words of Carl Sagan and Steve Ostro, "a double-edged sword," which if wielded by the wrong hands could "introduce a new category of danger that dwarfs that posed by the objects themselves." They pointed out that a series of nuclear explosions capable of thwarting a dangerous asteroid is also capable of diverting a benign asteroid toward Earth.106
Regardless of the means used to defend Earth against asteroid hazards, radar is suited to play a vital role in identifying potentially hazardous objects. Radar is the essential tool for astrometry (position and movement); it can determine asteroid orbits with greater accuracy and reliability than any other method. After the detection of an asteroid and the determination of its orbit, astronomers extrapolate the orbit into the future. Without radar precision measurement, the uncertainty of that extrapolation increases strikingly. The role of radar in Spaceguard, consequently, is as the primary, post-discovery ground-based technique for refining asteroid orbits.
After Steve Ostro's experiences with the errors in the ephemerides provided for 1986 DA and 1986 JK, he and fellow JPL employees Don Yeomans and Paul Chodas, who were in charge of calculating ephemerides for space missions, including those for a potential future asteroid flyby mission, assessed the extent to which radar observations could improve the accuracy of near-Earth asteroid ephemerides. They wanted to know how useful radar ranging was for refining the orbits of Earth-crossing asteroids. Could radar improve the extrapolation of asteroid orbits into the future?
They studied four asteroids with different histories of optical and radar observations, 1627 Ivar, 1986 DA, 1986 JK, and 1982 DB. The radar data provided only a modest absolute improvement for Ivar, which had a long history of optical astrometric data, but rather dramatic reductions in the future ephemeris uncertainties of asteroids having only short optical-data histories. Those improvements were impressive ones, to three orders of magnitude.
Ray Jurgens, who had been observing asteroids at Goldstone since the 1970s, wrote a proposal to fund asteroid emphemeris work at JPL and persuaded Don Yeomans and Paul Chodas to help in the analysis of asteroid ephemerides. As Jurgens became overwhelmed by research and the rebuilding of the Goldstone radar, Steve Ostro took up the tasks of strengthening JPL's asteroid ephemeris program and advocating software tools and other measures for improving Goldstone's capability of detecting asteroids and improving the accuracy of asteroid orbit predictions.107
 The astrometric and imaging capabilities of radar soon will combine to reformulate the IAU circular that announces the discovery and orbit of a new asteroid. For newly spotted asteroids, Ostro has a vision of the kind of IAU circular that might be available before the end of the century. After astronomers discover and track an asteroid optically for a few nights and the orbit is at least crudely known, an IAU circular announces the object's existence. A few days later, the Arecibo or Goldstone radar observes the asteroid and takes range-Doppler data, refines the orbit, and images the object. The ephemeris is updated immediately. Streamlined software transforms the image data into a three-dimensional model of the asteroid, then produces a video simulation of the Sun-illuminated asteroid. This process yields a computer file that becomes the first post-discovery IAU circular: a finely-resolved video image of the object, as if made by a flyby spacecraft within a few days of discovery.108 Here was the future of asteroid radar research and, to a dramatic degree, the future of planetary radar astronomy, as well.
1. Jurgens 23/5/94. In addition to the computer and other hardware problems, small cracks appeared in the pedestal of the Goldstone Mars Station. The repair involved raising the 3,000-ton structure and replacing a large portion of the pedestal concrete. The antenna did not return to service until June 1984, after being down a year for repairs. JPL Annual Report, 1983, p. 26, and ibid., 1984, p. 26, JPLA.
2. Memorandum, Standish to R. Green, 10 May 1979, Jurgens materials.
3. Downs 4/10/94.
4. Henry J. Moore to Arden L. Albee, 2 July 1979, Jurgens materials; Memorandum, William H. Bayley to Murray, 4 February 1980, 91/7/89-13, JPLA; Various documents in "NASA Correspondence, 1980-1981," JPLPLC.
5. Jurgens 23/5/94; Planetary Radar Working Group mailing list, Jurgens materials; Memorandum, Carl W. Johnson to Murray, 27 October 1980, 99/8/89-13, JPLA.
6. Jurgens 23/5/94; Downs 4/10/94; C. H. Terhune, Jr., to B. I. Edelson and R. E. Smylie, 20 September 1982, "Chron 1982, #2," JPLPLC.
7. Renzetti 16/4/92; Renzetti 17/4/92.
8. Ostro 18/5/94; GSSR Min. 6/12/1984.
9. Thompson 29/11/94; Slade 24/5/94; GSSR Min. 6/12/1984 and 31/3/1988.
10. GSSR Min. 29/12/1986.
11. GSSR Min. 22/1/1987 and 26/2/1987; NAIC QR Q1/1986, 19.
12. GSSR Min. 26/2/1987.
13. Memorandum, Muhleman to Edward C. Posner, 28 October 1986, Ostro materials.
14. GSSR Min. 3/12/1987, 14/1/1988, 18/2/1988, and 28/4/1988.
15. Ostro 18/5/94; GSSR Min. 14/1/1988 and 18/2/1988.
16. JPL Annual Report, 1973-1974, p. 15; ibid., 1984, p. 13; ibid., 1987, p. 41; and ibid., 1988, p. 28, JPLA.
17. Ostro 18/5/94.
18. GSSR Min. 22/1/1987, 18/6/1987, 23/7/1987, 24/9/1987, 18/2/1988, 31/3/1988, and 26/4/1990; Renzetti, Thompson, and Slade, "Relative Planetary Radar Sensitivities: Arecibo and Goldstone," TDA Progress Report 42-94 (Pasadena: JPL, April-June 1988), pp. 287-293; Arthur J. Freiley, Bruce L. Conroy, Daniel J. Hoppe, and Alaudin M. Bhanji, "Design Concepts of a 1-MW CW X-Band Transmit/Receive System for Planetary Radar," IEEE Transactions on Microwave Theory and Techniques 40 (1992): 1047-1055.
19. GSSR Min. 6/2/1992.
20. Memorandum, Ostro to Elachi, 29 August 1990; Memorandum, David Hills to Dick Mathison, 1 October 1990; Memorandum, Ostro to Larry N. Dumas, 15 October 1990, Ostro materials.
21. Pettengill to Elachi, 22 August 1991, and attachments, Ostro materials. The members of the Goldstone Planetary Radar Science Review Committee were Gordon H. Pettengill, MIT; Michael J. S. Belton, Kitt Peak National Observatory; Donald B. Campbell, NAIC; Clark R. Chapman, Planetary Science Institute; Tor Hagfors, NAIC; Bruce W. Hapke, University of Pittsburgh; Randolph L. Kirk, USGS; David Morrison, NASA Ames Research Center; and F. Peter Schloerb, University of Massachusetts.
22. Muhleman 27/5/94.
23. Pettengill to Elachi, 22 August 1991, and attachments, Ostro materials.
24. Muhleman 27/5/94.
25. Hagfors, "The Arecibo Gregorian Upgrading," in Joseph H. Taylor and Michael M. Davis, eds., Scientific Benefits of an Upgraded Arecibo Telescope (Arecibo: NAIC, 1987), p. 4, and Ostro, "Benefits of an Upgraded Arecibo Observatory for Radar Observations of Asteroids and Natural Satellites," in ibid., p. 233.
26. Campbell 9/12/93.
27. Kay, A Line Source Feed, passim, and Pierluissi, A Theoretical Study of Gregorian Radio Telescopes, passim.
28. Hagfors, "The Arecibo Gregorian Upgrading," p. 3; Per-Simon Kildal, Lynn A. Baker, and Hagfors, "The Arecibo Upgrading: Electrical Design and Expected Performance of the Dual-Reflector Feed System," Proceedings of the IEEE 82 (1994): 714.
29. NAIC QR Q3/1982, 19; Campbell 7/12/93; Campbell 9/12/93.
30. Lovell, The Jodrell Bank Telescope, pp. 270-271.
31. Campbell 9/12/93; Hagfors, "The Arecibo Gregorian Upgrading," p. 3.
32. Campbell 9/12/93; NAIC QR Q2/1984, 14; Hagfors, "The Arecibo Gregorian Upgrading," p. 3; Kildal, Baker, and Hagfors, p. 714.
33. NAIC QR Q2/1984, 14; Campbell 9/12/93; Hagfors, "The Arecibo Gregorian Upgrading," p. 3.
34. NAIC QR Q2/1984, 14, and Q3/1984, 15; Hagfors, "The Arecibo Gregorian Upgrading," p. 4; Kildal, Baker, and Hagfors, pp. 717-718 & 722.
35. Campbell 9/12/93; Dickman 2/12/92.
36. Ostro, "Benefits of an Upgraded Arecibo," pp. 233-239; Campbell, "Prospects for Radar Observations of Comets and the Terrestrial Planets," in Taylor and Davis, pp. 243-248; Shapiro, "Radar Tests of Gravitational Theories and Other Exotica," in ibid., pp. 225-232.
37. Campbell 9/12/93; Kildal, Baker, and Hagfors, p. 715.
38. John Bahcall, "Preface," in National Research Council, The Decade of Discovery in Astronomy and Astrophysics (Washington: National Academy Press, 1991), pp. ix-xi.
39. National Research Council, Working Papers: Astronomy and Astrophysics Panel Reports (Washington: National Academy Press, 1991), pp. X-1-X-20.
40. Campbell 9/12/93.
41. Dickman 2/12/92; Campbell 9/12/93.
42. Murray to Morton S. Roberts, 25 February 1982, "Chron 1982 #1," and Memorandum, Associate Administrator for Space Tracking and Data Systems to Deputy Director, JPL, 28 February 1983, "NASA Correspondence, 1983, pt. #1," JPLPLC; JPL Annual Report, 1984, p. 13, and ibid., 1987, p. 41, JPLA.
43. Memorandum, Lilley to CAMROC Project Office Members, 14 June 1966, 18/1/AC 135, MITA.
44. Renzetti 17/4/92.
45. Muhleman 8/4/93.
46. The ethane-methane ocean model of Titan was developed by Jonathan I. Lunine, David J. Stevenson, and Yuk L. Yung. See, for example, Lunine, Stevenson, and Yung, "Ethane Ocean on Titan," Science 222 (1983): 1229-1230.
47. Muhleman, Arie W. Grossman, Bryan J. Butler, and Slade, "Radar Reflectivity of Titan," Science 248 (1990): 975-980.
48. NAIC QR Q1/1979, 9; Campbell 8/12/93; Goldstein and Jurgens, "DSN Observations of Titan," in Posner, ed. The Telecommunications and Data Acquisition Report: Progress Report, Jan. - Mar. 1992 (Pasadena: JPL, 1992), pp. 377-379.
49. Muhleman, G. Berge, and D. Rudy, "Microwave Emission from Titan and the Galilean Satellites," Bulletin of the American Astronomical Society 16 (1984): 686; JPL Annual Report, 1988, p. 29, JPLA.
50. Muhleman, Grossman, Butler, and Slade, "Radar Reflectivity of Titan," Science 248 (1990): 975-980, quote on p. 979.
51. Muhleman, Grossman, Slade, and Butler, "Titan's Radar Reflectivity and Rotation," Bulletin of the American Astronomical Society 25 (1993): 1099; Butler, Muhleman, and Slade, "Results from 1992 and 1993 VLA/Goldstone 3.5 cm Radar Results," ibid., p. 1040; GSSR Min. 19/2/1993.
52. Muhleman to W. E. Giberson, 10 February 1989, and Memorandum, D. L. Matson to Dumas, 8 January 1991, Renzetti materials.
53. Renzetti 17/4/92.
54. Harmon, "Radar Observations of Mars and Mercury: History and Progress," paper read at the Thirtieth Anniversary Celebration of Planetary Radar Astronomy, 3 October 1991, Caltech; Zisk and P. J. Mouginis-Mark, "Anomalous Region on Mars: Implications for Near-Surface Liquid Water," Nature 288 (1980): 735-738. See also Aaron P. Zent, Fraser P. Fanale, and Roth, "Possible Martian Brines: Radar Observations and Models," Journal of Geophysical Research vol. 95, no. B9 (1990): 14,531-14,542.
55. Downs, Mouginis-Mark, Zisk, and Thompson, "New Radar-Derived Topography for the Northern Hemisphere of Mars," Journal of Geophysical Research 87 (1982): 9747-9754; Mouginis-Mark, Zisk, and Downs, "Ancient and Modern Slopes in the Tharsis Region of Mars," Nature 297 (1982): 546-550; Simpson, Tyler, Harmon, and Alan R. Peterfreund, "Radar Measurement of Small-Scale Surface Texture: Syrtis Major," Icarus 49 (1982): 258-283; Schaber, "Syrtis Major: A Low-relief Volcanic Shield," Journal of Geophysical Research 87 (1982): 9852-9866; Roth, Downs, Saunders, and Schubert, "Radar Altimetry of South Tharsis, Mars," Icarus 42 (1980): 287-316; R. A. Craddock, R. Greeley, and P. R. Christensen, "Evidence for an Ancient Impact Basin in Daedalia Planum, Mars," Journal of Geophysical research 95 (1990): 10,729-10,741; Downs, R. Green, and Reichley, "Radar Studies of the Martian Surface at Centimeter Wavelengths: The 1975 Opposition," Icarus 33 (1978): 441-453; Roth, Saunders, Downs, and Schubert, "Radar Altimetry of Large Martian Craters," Icarus 79 (1989): 289-310.
56. Harmon 15/3/94; Harmon, Campbell, and Ostro, "Dual-Polarization Radar Observations of Mars: Tharsis and Environs," Icarus 52 (1982): 171-187; Harmon and Ostro, "Mars: Dual-Polarization Radar Observations with Extended Coverage," Icarus 62 (1985): 110-128; Thompson and Henry J. Moore, "A Model for Depolarized Radar Echoes from Mars," Proceedings of the Lunar Planetary Science Conference 19th (1989): 409-422; Moore and Thompson, "A Radar-Echo Model of Mars," Proceedings of the Lunar Planetary Science Conference 21 (1991): 457-472. Later radar mapping supported these observations: Muhleman, Butler, Grossman, Slade, and Jurgens, "Radar Images of Mars," Science 253 (1991): 1508-1513; Harmon, Michael P. Sulzer, Phillip J. Perillat, and Chandler, "Mars Radar Mapping: Strong Backscatter from the Elysium Basin and Outflow Channel," Icarus 95 (1992): 153-156.
57. Muhleman 27/5/94.
58. Muhleman 27/5/94.
59. Muhleman 27/5/94; Slade 24/5/94; Muhleman, Butler, Grossman, and Slade, "Radar Images of Mars," Science 253 (1991): 1508-1513; Butler, "3.5-cm Radar Investigation of Mars and Mercury: Planetological Implications," Ph.D. diss., California Institute of Technology, 9 May 1994; Simpson and Tyler, "Viking Bistatic Radar Experiment: Summary of First-Order Results Emphasizing North Polar Data," Icarus 46 (1981): 361-389.
60. Muhleman 27/5/94; Butler, "3.5-cm Radar Investigation;" Butler, Muhleman, and Slade, "Martian Polar Regions: 3.5 cm Radar Images," Lunar and Planetary Science Conference 25 (1994): 211-212; Butler, Muhleman, and Slade, "The Polar Regions of Mars: 3.5 cm Radar Images," Icarus submitted in May 1994.
61. Hudson, telephone conversation, 21 November 1994; Hudson and Ostro, "Doppler-Radar Imaging of Spherical Planetary Surfaces," Journal of Geophysical Research 95 (1990): 10,947-10,963; Harmon, Slade, and Hudson, "Mars Radar Scattering: Arecibo/Goldstone Results at 12.6- and 3.5-cm Wavelengths," Icarus 98 (1992): 240-253.
62. Harmon 15/3/94.
63. Harmon 15/3/94; NAIC QR Q2/1990, 7; Q4/1990, 7-8; and Q1/1991, 7; Q1/1993, 9; Harmon, Sulzer, and Perillat, "Mars Radar Mapping: Strong Depolarized Echoes from the Elysium/Amazonis Outflow Channel Complex," Lunar and Planetary Science Conference 22 (1991): 513.
64. Muhleman, "Radar Scattering from Venus and Mercury at 12.5 cm," Journal of Research of the National Bureau of Standards, Section D: Radio Science 69D (1965): 1630-1631; Evans, Brockelman, Henry, Hyde, Kraft, W. A. Reid, and W. W. Smith, "Radio Echo Observations of Venus and Mercury at 23 cm Wavelength," The Astronomical Journal 70 (1965): 486-501; Pettengill, Dyce, and Campbell, "Radar Measurements at 70 cm of Venus and Mercury," The Astronomical Journal 72 (1967): 330-337; Goldstein, "Mercury: Surface Features Observed during Radar Studies," Science 168 (1970): 467-469; Goldstein, "Radio and Radar Studies of Venus and Mercury," Radio Science 5 (1970): 391-395; Goldstein, "Radar Observations of Mercury," The Astronomical Journal 76 (1971): 1152-1154; Goldstein, "Review of Surface and Atmosphere Studies of Venus and Mercury," Icarus 17 (1972): 571-575; Zohar and Goldstein, "Surface Features on Mercury," The Astronomical Journal 79 (1974): 85-91; Smith, Ingalls, Shapiro, and Ash, "Surface-Height Variations on Venus and Mercury," Radio Science 5 (1970): 411-423; Ingalls and Rainville, "Radar Measurements of Mercury: Topography and Scattering Characteristics at 3.8 cm," The Astronomical Journal 77 (1972): 185-190.
65. Murray, Michael J. S. Belton, G. Edward Danielson, Merton E. Davies, Donald E. Gault, Hapke, Brian O'Leary, Robert G. Strom, Verner Suomi, and Newell Trask, "Mercury's Surface: Preliminary Description and Interpretation from Mariner 10 Pictures," Science 185 (1974): 169-179.
66. Harmon, Campbell, Bindschadler, Head, and Shapiro, "Radar Altimetry of Mercury: A Preliminary Analysis," Journal of Geophysical Research 91 (1986): 385-401.
67. See, for example, P. E. Clark, M. E. Strobell, Schaber, and Jurgens, "Some New Radar-Derived Topographic Profiles of Mercury," Bulletin of the American Astronomical Society 16 (1984): 668; Clark, Jurgens, and M. Kobrick, "Analyses of Radar-Derived Topography and Scattering Properties of Mercury's Equatorial Region," Bulletin of the American Astronomical Society 17 (1985): 712; and Clark, M. A. Leake, Slade, Jurgens, Robinett, and C. Franck, "Scattering and Altimetry Measurements from Goldstone Radar Observations of Mercury in 1987," Bulletin of the American Astronomical Society 19 (1987): 863.
68. Butler, "3.5-cm Radar Investigation," preface.
69. Muhleman 27/5/94; Slade 24/5/94.
70. Slade, Butler, Muhleman, "Mercury Radar Imaging: Evidence for Polar Ice," 258 Science (23 October 1992): 635-640; Butler, Muhleman, and Slade, "Mercury: Full-Disk Radar Images and the Detection and Stability of Ice at the North Pole," Journal of Geophysical Research vol. 98, no. E8 (1993): 15,003-15,023.
71. Slade, Butler, and Muhleman, "Mercury Goldstone-VLA Radar: Part I," Bulletin of the American Astronomical Society 23 (1991): 1197, and Butler, Muhleman, Slade, and Jurgens, "Mercury Goldstone-VLA Radar: Part II," Ibid., p. 1200.
72. David A. Paige, "Chance for Snowballs in Hell," Nature 369 (1994): 182; Chapman, "Ice Right Under the Sun," Nature 354 (1991): 504-505; J. Kelley Beatty, "Mercury's Cool Surprise," Sky & Telescope 83 (January 1992): 35-36.
73. Richard Baum, "Radar Bright, Ice Bright: V. A. Firsoff and Ice Caps on Mercury," Journal of the British Astronomical Association 103 (1993): 126 & 139; Firsoff, "Could Mercury have Ice Caps?" The Observatory 91 (1971): 85-87; and G. E. Hunt, "There is no Evidence for Ice Caps on Mercury," The Observatory 92 (1972): 16; Beatty, "Mercury's Cool Surprise," Sky & Telescope 83 (1992): 35-36; Gary E. Thomas, "Mercury: Does its Atmosphere Contain Water?" Science 183 (1974): 1197-1198.
74. Beatty, p. 36; Chapman, "Ice," p. 505; Chapman, Planets of Rock and Ice: From Mercury to the Moons of Saturn (New York: Scribner, 1982); and Faith Vilas, Chapman, and Matthews, eds., Mercury (Tucson: University of Arizona Press, 1988).
75. Simpson 10/5/94; Paige, Stephen E. Wood, and Ashwin R. Vaasavada, "The Thermal stability of Water Ice at the Poles of Mercury," Science 258 (1992): 643-646; Andrew P. Ingersoll, Tomas Svitek, and Murray, "Stability of Polar Frosts in Spherical Bowl-Shaped Craters on the Moon, Mercury, and Mars," Icarus 100 (1992): 40-47; Kenneth Watson, Murray, and Harrison Brown, "The Behavior of Volatiles on the Lunar Surface," Journal of Geophysical Research 66 (1961): 3033-3045.
76. Ford 3/10/94; Campbell 10/3/93; Campbell 8/12/93; Stacy, "High-Resolution Synthetic Aperture Radar Observations of the Moon," Ph.D. diss., Cornell University, May 1993.
77. Paige, "Chance for Snowballs in Hell," Nature 369 (1994): 182.
78. Harmon 15/3/94.
79. Harmon 15/3/94; Harmon and Slade, "Radar Mapping of Mercury: Full-Disk Images and Polar Anomalies," Science 258 (1992): 640-642; Harmon and Slade, "An S-band Radar Anomaly at the North Pole of Mercury," Bulletin of the American Astronomical Society 23 (1991): 1121.
80. Harmon, Slade, Vélez, Andy Crespo, M. J. Dryer, and J. M. Johnson, "Radar Mapping of Mercury's Polar Anomalies," Nature 369 (1994): 213-215; Harmon and Slade, "Radar Mapping of Mercury: Full-Disk Images and Polar Anomalies," Science 258 (1992): 640-643.
81. Muhleman 27/5/94; Paige, "Snowballs," p. 182.
82. Slade 24/5/94; K. A. Tryka, Muhleman, Butler, Berge, Slade, and Grossman, "Correlation of Multiple Reflections from the Venus Surface with Topography," Lunar Planetary Science 22 (1991): 1417; Jurgens, Slade, and Saunders, "Evidence for Highly Reflecting Materials on the Surface and Subsurface of Venus," Science 240 (1988): 1021-1023; Butler, "3.5-cm Radar Investigation," passim; Slade 24/5/94; and information provided by Bryan J. Butler.
83. Ostro, Campbell, and Shapiro, "Mainbelt Asteroids: Dual-Polarization Radar Observations," Science 229 (1985): 442.
84. Ostro 25/5/94; Pettengill and Jurgens, "Radar Observations of Asteroids," in Gehrels and Matthews, pp. 206-211.
85. Ostro 25/5/94; Ostro, "Radar Investigations of Asteroids," proposal submitted to NASA in June 1980 for support 1 November 1980 through 31 October 1981, Ostro materials.
86. Ostro 25/5/94.
87. Ostro 25/5/94.
88. Ostro, Campbell, and Shapiro, "Mainbelt Asteroids: Dual-Polarization Radar Observations," Science 229 (1985): 442-446.
89. Ostro 25/5/94.
90. Ostro, Campbell, Chandler, Hine, Hudson, Rosema, and Shapiro, "Asteroid 1096 DA: Radar Evidence for a Metallic Composition," Science 252 (1991): 1399-1404.
91. Ostro 25/5/94.
92. See, for example, the discussion in W. I. McLaughlin, "Radar Tracking of Asteroids," Spaceflight 34 (1992): 167-169.
93. Ostro 25/5/94; Ostro, Campbell, and Shapiro, "Radar Observations of Asteroid 1685 Toro," The Astronomical Journal 88 (1983): 565-576.
94. Ostro, Alan W. Harris, Campbell, Shapiro, and James W. Young, "Radar and Photoelectric Observations of Asteroid 2100 Ra-Shalom," Icarus 60 (1984): 391-403.
95. Ostro, Robert Connelly, and Leila Belkora, "Asteroid Shapes from Radar Echo Spectra: A New Theoretical Approach," Icarus 73 (1988): 15-24; Ostro, Rosema, and Jurgens, "The Shape of Eros," Icarus 84 (1990): 334-351; Ostro, Campbell, Chandler, Hine, Hudson, Rosema, and Shapiro, "Asteroid 1096 DA: Radar Evidence for a Metallic Composition," Science 252 (1991): 1399-1404, esp. pp. 1400-1401.
96. Ostro 25/5/94; Ostro, Campbell, Hine, Shapiro, Chandler, C. L. Werner, and Rosema, "Radar Images of Asteroid 1627 Ivar," The Astronomical Journal 99 (1990): 2012-2018.
97. Ostro 25/5/94.
98. Ostro 25/5/94.
99. Ostro 25/5/94.
100. Ostro 25/5/94; Ostro, Chandler, Hine, Rosema, Shapiro, and Yeomans, "Radar Images of Asteroid 1989 PB," Science 248 (1990): 1523-1528.
101. Hudson, telephone conversation, 21 November 1994; Ostro 25/5/94; Hudson and Ostro, "Shape of Asteroid 4769 Castalia (1989 PB) from Inversion of Radar Images," Science 263 (1994): 940-943; Hudson and Ostro, "Doppler-Radar Imaging of Spherical Planetary Surfaces," Journal of Geophysical Research 95 (1990): 10,947-10,963.
102. Ostro 25/5/94; Hudson, telephone conversation, 21 November 1994; Ostro, Jurgens, Rosema, R. Winkler, D. Howard, R. Rose, Slade, Yeomans, Campbell, Perillat, Chandler, Shapiro, Hudson, P. Palmer, and I. de Pater, "Radar Imaging of Asteroid 4179 Toutatis," Bulletin of the American Astronomical Society 25 (1993): 1126.
103. The Spaceguard Survey: Report of the NASA International Near-Earth-Object Detection Workshop (Pasadena: JPL, 25 January 1992), p. 19; NASA Press Release 94-158, 20 September 1994, Renzetti materials.
104. Chapman and Morrison, "Impacts on the Earth by Asteroids and Comets: Assessing the Hazard," Nature 367 (1994): 33-40.
105. The Spaceguard Survey: Report of the NASA International Near-Earth-Object Detection Workshop (Pasadena: JPL, 25 January 1992), pp. 1-3 & 49-52; Cunningham, pp. 113-116 & 141.
106. Sagan and Ostro, "Dangers of Asteroid Deflection," Nature 368 (1994): 501.
107. Ostro 25/5/94; Yeomans, Ostro, and Paul W. Chodas, "Radar Astrometry of Near-Earth Asteroids," The Astronomical Journal 94 (1987): 189-200; Ostro, "The Role of Ground-Based Radar in Near-Earth Object Hazard Identification and Mitigation," in Hazards Due to Comets and Asteroids, in press, p. 9. For a summary of asteroid radar astrometry, see Ostro, Campbell, Chandler, Shapiro, Hine, Vélez, Jurgens, Rosema, Winkler, and Yeomans, "Asteroid Radar Astrometry," The Astronomical Journal 102 (1991): 1490-1502; and Yeomans, Chodas, M. S. Keesey, Ostro, Chandler, and Shapiro, "Asteroid and Comet Orbits using Radar Data," The Astronomical Journal 103 (1992): 303-317.
108. Ostro 25/5/94.