Atmospheric Science at NASA: A History

• 8 Missions to Planet Earth: Architectural Warfare
• Chapter
• Johns Hopkins University Press
• pp. 243-275
• View Citation
• Additional Information

CHAPTER EIGHT

Missions to Planet Earth: Architectural Warfare

The Earth Observing System (EOS) was subjected to cuts every year through 1995. This, in the words of a harsh National Academy of Sciences assessment, resulted in management turbulence.¹ Seemingly unending reviews, redefinitions, rescopings, and rebaselinings consumed time, squandered project resources, and demoralized staff. By the time the system definition finally stabilized in 1996, the project had been shrunk from $17 billion to $7 billion through the beginning of the new century. The resulting delays pushed the first launch into 1999, twenty-one years after the original National Climate Program Act had gotten planning for EOS started.

EOS’s fiscal troubles stemmed from a number of causes. The sudden collapse of the Soviet Union in 1991 undermined the American space program overall, as it had always been primarily a tool of Cold War propaganda.² Space Station Freedom suffered the same descoping, and in fact survived an effort to terminate it entirely in 1993 by a single vote in the Senate. Large flagship science projects outside NASA, like the Superconducting Supercollider, were attacked.³ The early 1990s also witnessed an unusual bidding war over the federal deficit, with the two political parties trying to cut areas of the budget that largely benefited the other’s principal constituencies. While this party competition eventually produced the first balanced budget in a generation, the result for NASA was catastrophic. The agency’s budget declined 18 percent between 1991 and 1999, when it became obvious that further cuts would require terminating major agency functions.⁴ Further, the Republican Revolution of 1994 placed anti-environmental leaders firmly in charge of Congress, making EOS an obvious target.⁵ Finally, NASA administrator Daniel Goldin, appointed in early 1992, believed that everything the agency did could be done far more cheaply, and therefore did not resist the cuts. He had also directly attacked EOS’s “Battlestar Galactica” approach in a private White House briefing.⁶ He welcomed the fiscal discipline being imposed as a way to force his preferred “faster-better-cheaper” approach to space science on the agency.⁷

In the process of shrinking EOS, NASA leaders abandoned the large satellite approach. They also abandoned the idea of maintaining the primary EOS measurements for fifteen years. In the early years of the downsizing, the range of scientific questions was narrowed, but expanded again later as the savings from the smaller satellite approach became clearer. NASA also began to try to migrate some of the key measurements to a new, joint air force–National Oceanic and Atmospheric Administration (NOAA)–NASA operational weather satellite program in hopes of maintaining continuity of measurements; finally, it sought to make the climate observing system international as it had once tried to make the weather satellite system international. It found international partners for some missions, and in some cases dropped planned measurements entirely when other countries promised to take them over. And yet, by the first years of the twenty-first century, NASA had fully abandoned Shelby Tilford’s original conception of an integrated, long-term, well-calibrated set of Earth observations. Instead, the nation’s strategy was back where it had been in 1980, and the successor to EOS, a new generation of meteorological satellites, seemed to be firmly on the twenty-years-to-space track.

DEMISE OF THE POLAR PLATFORMS

The revolt against EOS had started even before the program had achieved White House approval in 1989. There were three major streams of criticism developing even as Shelby Tilford and Dixon Butler tried to shepherd EOS through the approval process. A vocal contingent within the scientific community opposed EOS because of its inflexibility, slow pace, and gigantism. Second, from a technological direction, particularly the clandestine world of the Strategic Defense Initiative (SDI), came criticism of EOS for its use of “old” early 1980s technology instead of new and untried concepts developed in (mostly) military labs. Third, at $17 billion, there was immediate criticism of the program’s cost. Many scientists believed that it would swallow all the resources available for Earth science research for the next two decades, and that therefore scientists who had not been part of EOS in its formative phase would be shut out of NASA funding effectively forever. These three lines of criticism converged in a demand for a smaller, less expensive, and more flexible system architecture.

As it stood in 1990, EOS consisted of a set of four Hubble Space Telescope–sized polar orbiting platforms, two built by NASA and one each provided by the European Space Agency and the National Space Development Agency of Japan (NASDA). These were to be launched between 1996 and 1998, and repeated three times each to achieve the desired fifteen-year time span. The first NASA platform was assigned instruments for surface studies, atmospheric temperature soundings, and ocean and land surface altimetry. This platform also contained instruments intended to replace those on the NOAA series of operational weather satellites. The second NASA payload carried a very large synthetic aperture radar and a suite of atmospheric chemistry and physics instruments. These included instruments aimed at studies of the thermosphere and mesosphere, and of the Earth’s magnetic field. The European and Japanese platforms were similarly equipped. In its original guise, EOS was meant to envelop the full range of the Earth sciences. It was not focused on any particular scientific question despite its origins in the early climate observing system planning exercises of the late 1970s. It also contained no provision for smaller missions, for missions requiring non-polar orbits, or for missions responsive to new scientific questions that might arise over the fifteen-year period. This basic system architecture, not the science EOS was supposed to do, was the initial point of conflict over EOS.

The Space Studies Board of the National Research Council (NRC) had issued the first formal statement of scientific discontent over EOS in a 1988 study, Strategy for Earth Explorers in the Global Earth Sciences. It had been prepared at the request of NASA’s Space and Earth Science Advisory Committee, which was composed primarily of university-based researchers. This group, whose chairman was geodicist Byron Tapley of the University of Texas, sought to supplement EOS with a series of smaller satellites that would take less time and money to develop. Tapley’s group believed that smaller satellites provided a number of benefits. One was that the large polar platforms could not make some of the desired measurements. Structurally, large platforms would be too flexible for use in precision altimetry measurements, for example, which were necessary for oceanographic and cryospheric research. Another was that large platforms carried large risk. Loss of a small, inexpensive satellite in the event of a launch vehicle accident, or from a spacecraft failure, was much easier to overcome financially than loss of one of the large platforms. Finally, smaller satellites also offered greater programmatic flexibility. New scientific demands could be accommodated more easily if NASA funded a line of small satellites in addition to the large platforms. Hence, Tapley’s committee argued for a line of Earth Explorers within the Mission to Planet Earth that would be similar in nature to the Explorer series of satellites used by NASA’s physics and astronomy community. They sought sufficient funding for two small missions per year or one medium mission every three years.⁸

Artists’ conception of the EOS large platform. This version carries a large radar, the panel in the lower left. Courtesy NASA/Caltech/JPL.

Tapley’s committee reflected a growing unhappiness within the scientific community over the sheer length of time NASA was taking to fly new Earth science missions. Much of the work done in American science was done by graduate students and postdoctoral fellows, who had short time horizons (four to six years) and could not participate effectively in missions taking ten to twenty years to develop and launch. This undermined space-based science in general by drying up the pool of future talent. The possibility that EOS could take decades was made real for the committee by the Upper Atmosphere Research Satellite (UARS), which by 1988 had been in development since 1978 and, because of the Challenger accident, would not fly for several more years. It had thus far taken ten years to develop this critical group of allegedly high-priority measurements, and as a result, the scientific limelight had been stolen by the ground- and aircraft-based instruments NASA and NOAA had deployed in the National Ozone Expedition (NOZE) and the Airborne Antarctic Ozone Experiment (AAOE). EOS, planning for which could also be traced back to 1978, seemed to be taking an even slower road to orbit, and since it would be the only Earth science program for the next twenty years, the committee found its large platform approach deeply troubling.⁹ EOS seemed to be an example of Big Engineering getting in the way of scientific research.

During 1990, the Goddard Institute for Space Science’s (GISS) James Hansen began advocating a set of small satellites called CLIMSATs (Climate Satellites) that were also about the size of the proposed Earth Explorers. His CLIMSAT proposal involved two small satellites, each with three instruments: improved versions of the Langley Stratospheric Aerosol and Gas Experiment (SAGE) and Earth Radiation Budget Experiment (ERBE) instruments and an Earth Observing Scanning Polarimeter (EOSP), to measure cloud radiative properties. EOSP was a derivative of instruments previously sent to Venus and Jupiter. All three had been selected for EOS during the 1988 announcement of opportunity process, and represented relatively mature technologies with previous flight experience. Hansen argued that their technical maturity meant that they could provide critical measurements sooner than would the large EOS platforms.

Hansen’s scientific motivation for the CLIMSAT proposal was a 1989 article by influential radiative transfer specialists, who had argued that water vapor feedback would enhance the carbon dioxide greenhouse enough to make it detectable by an ERBE-like instrument within the next couple of decades.¹⁰ Hence, they had concluded, ERBE or something very much like it, needed to be flown for the next several decades. Hansen added SAGE to monitor both ozone, which was also a greenhouse gas, and aerosols, which tended to cool the Earth. The scanning polarimeter was to study the interaction of aerosols and clouds. These three instruments, Hansen contended, were the ones necessary to detect anthropogenic climate change, and needed to be flown as quickly as possible to establish a continuous record. He also believed that they would be inexpensive enough to be reflown and maintained in orbit for long-term monitoring; the hugely expensive EOS constellation would almost certainly not be.¹¹

Hansen’s proposed CLIMSAT system was aimed at studying the changing climate. EOS, however, had evolved for process studies, not climate monitoring. Hence, CLIMSAT would not fulfill EOS’s other science goals, such as better forecasting of El Niño. EOS’s perceived ability to contribute to better regional prediction had been an important reason for its approval by the White House, so CLIMSAT was not really an alternative to EOS. It was complementary. But it was badly timed. Hansen’s public advocacy of it while NASA was attempting to get EOS through Congress tended to make it appear as a smaller, cheaper competitor to EOS. Hansen himself argued that CLIMSAT was complementary to EOS, and should be funded via an expanded Earth Probes program. But it nonetheless caused difficulties for Lennard Fisk, Shelby Tilford, and Dixon Butler, who had to defend EOS’s mission anew.¹²

Yet a third line of criticism emerged from the National Space Council. Veterans of the Defense Department’s SDI contended that it had developed a variety of microsat technologies that could make EOS’s measurements earlier and much less expensively. As ballistic missile defense was being conceived in the late 1980s, dozens of low-altitude, infrared scanning, micro-satellites called Brilliant Eyes were supposed to provide guidance information to ground-based interception missiles. These became the source of a third challenge to EOS, layered on top of the Earth Probes and the CLIMSATS.¹³

The collapse of East Germany in 1989 and the subsequent end of the Cold War, finally, led to reexamination of the federal budget overall, and NASA’s in particular. NASA’s role as a vehicle for technocratic competition and propaganda victories left the agency without its primary political function. Space Station Freedom was immediately threatened with termination; it survived but was repeatedly descoped until (temporarily) stabilizing in 1994 as the International Space Station. EOS was also subjected to budget reductions almost immediately. At hearings on EOS in April 1991, Senator Albert Gore, Jr., of Tennessee, warned Lennard Fisk, NASA’s associate administrator for space science, that the fiscal year 1992 budget then being debated in the Senate would not fully fund EOS; the conference committee report that emerged placed a run-out cap on EOS of $11 billion through fiscal year 2000.¹⁴

The conference committee report also required NASA to empanel an independent review of EOS. NASA Administrator Richard M. Truly asked Edward Frieman, who had succeeded William Nierenberg as director of the Scripps Institution of Oceanography in 1986, to assemble a panel to help the agency restructure EOS. This group became known as the EOS Engineering Review Committee. Frieman was invited to Washington at the end of April to meet with senior administration officials and get their views on EOS; his interviews included Shelby Tilford; oceanographer W. Stanley Wilson, now EOS program scientist; NASA Administrator Truly; Mark Albrecht, the Space Council’s director; and Vice President Dan Quayle. He reported to his committee members that “there seems to be strong support within the Administration for EOS.”

But he also found three common concerns in all of his interviews: EOS was too expensive, it did not make use of newer small satellite technology, and “[it] is perceived by some as being too distant to help solve the critical near-term global change policy concerns.”¹⁵ As a result, Frieman recalled later, there was great hostility at the Space Council and at the White House toward NASA’s approach to EOS. In a letter to Frieman in mid-May, Representative George E. Brown, chairman of the House Committee on Science, Space, and Technology, put it slightly differently: “despite my own conviction and that of many of my colleagues that increased investments will be needed to better define future environmental policy options, major scientific undertakings such as EOS, for which a clear technical consensus does not yet exist, is difficult to sustain in this budgetary environment [sic].”¹⁶

Frieman’s committee recommended several approaches to bringing down EOS’s cost. The first of these was reducing the program’s scope. As it existed in 1990, the EOS concept included sensors aimed at all possible Earth science disciplines, from solid Earth geophysics to Sun-Earth interaction. Frieman’s committee advocated descoping EOS back to the Global Habitability initiative’s focus on climate change.¹⁷ It made this recommendation based upon the scientific priorities laid out by the Intergovernmental Panel on Climate Change (IPCC), whose 1990 report had called for an increased emphasis on “the various climate-related processes, particularly those associated with clouds, oceans, and the carbon cycle.”¹⁸ The IPCC’s three priorities, in turn, became one basis of EOS’s reconstruction.

The IPCC had chosen to emphasize clouds, oceans, and the carbon cycle because these were the sources of the largest uncertainties in predicting the rate of future warming. Ocean modeling was in its infancy compared to atmospheric modeling, with substantial model development not starting until the early 1980s. Ocean remote sensing was also in its infancy. Seasat A’s ninety-nine-day mission had been the first dedicated oceanographic flight, and to date the only one. NASA, the European Space Agency, and the Japanese space agency all had ocean-sensing instruments planned for flight during the 1990s, but with the exception of a modified version of the Coastal Zone Color scanner, these were all essentially first-generation instruments. They would require time to be fully understood, just as had the atmosphere sensing instruments. And because the oceans served as Earth’s primary heat storage system, improving ocean circulation data and ocean circulation models was fundamental to gaining better climate forecasting.

The IPCC’s focus on clouds grew out of the Langley Research Center’s ERBE. In January 1989, the project’s science team had published a paper in Science that weighed in on a long-standing controversy over whether clouds produced a net warming of the Earth by absorbing outgoing infrared energy, or whether they produced a net cooling by reflecting away incoming sunlight. The ERBE data clearly showed that clouds produced a net cooling effect, which the team chose to define in terms of forcing. Cloud reflectance of incoming sunlight was shortwave forcing, while cloud absorption of outgoing infrared, or longwave, radiation was longwave forcing. Defining their terms in this fashion allowed them to separate and analyze independently the two different effects clouds had.¹⁹

The ERBE data also allowed the team to assess how cloud forcings varied with latitude. Shortwave forcing was highest at high latitudes, while in the tropics shortwave and longwave forcings nearly cancelled each other out, as shown in Plate 5. And the magnitudes of the individual shortwave and longwave forcings were about 10 times the radiative forcing that would be produced by a doubling of carbon dioxide concentration. Because of this, the authors pointed out, a shift of mid-latitude cloudiness patterns toward the equator during the last ice age would have dramatically reinforced cooling and enhanced glaciation.²⁰ An expansion of tropical cloudiness patterns out of the tropics would, alternatively, have a warming effect. Clouds, their data made clear, were the dominant factor in regulation of the Earth’s climate.

The ERBE team also found from a brief survey of six existing global climate models that they had predicted this variation in cloud forcing qualitatively, but they had not done so quantitatively. The six models had shown wide variation in their analyses of cloud forcings, with shortwave forcing, for example, differing from the ERBE data by as much as 50 percent. Because the cloud effect on climate was so much more powerful than a carbon dioxide doubling, the inconsistent treatment of clouds by leading climate models introduced a large measure of uncertainty to forecasts of future climate.

Writing for Physics Today later that year, ERBE team members Veerabhadran Ramanathan, Bruce Barkstrom, and Edwin Harrison pointed out that the scientific community did not know why the cooling effect should be dominant.²¹ The physics of clouds were not well understood, and therefore the models could not be programmed to generate them from physical principles. Instead, clouds were incorporated into the models via parameterization, which allowed their known effects to be described within the models but largely prevented cloudiness patterns from changing as the model climates changed. This significantly impaired the models’ ability to forecast climate. This view was reinforced the following year, when Robert Cess published a paper surveying fourteen climate models, finding that while they relatively accurately represented clear-sky radiative transfer through the atmosphere, they diverged considerably in their treatment of cloud impacts.²² Hence, the IPCC had made cloud forcings their top research priority in their 1990 assessment.

Taking the independent IPCC’s report as its cue, Frieman’s committee had argued that EOS be refocused on climate change, and particularly on cloud and aerosol forcings and ocean measurements. They also argued for a greater internationalization of EOS to reduce the cost to the United States. While NASA had gotten agreements for European and Japanese polar platforms in the late 1980s, it had not attempted to create an integrated international global observing system. Instead, it had pursued international efforts somewhat piecemeal. It had gotten French support for TOPEX (Ocean Topography Experiment)/Poseidon without making arrangements for a successor; gained a space for a NASA-built scatterometer on the first Japanese environmental research satellite, ADEOS, scheduled for 1996; and forged a partnership with Japan for the Tropical Rainfall Measuring Mission. Foreign experimenters also had a number of instruments chosen for EOS via the standard announcement of opportunity process. But NASA had not tried to negotiate an overall division of labor among the other spacefaring governments over who would support which measurements over the two decades EOS was supposed to run.

Finally, Frieman’s reengineering committee recommended a substantial restructuring of EOS. EOS had been planned around large polar platforms that required the Titan 4 launch vehicle, which was the most expensive by far (except, of course, for the Shuttle). Delinking EOS from the Shuttle program had not changed that, partly because Tilford, Butler, and their planners believed that simultaneity of measurements was scientifically necessary, and therefore the instruments needed to be on the same platforms. This policy had also been based on the lack of a mid-sized launch vehicle, with the only rockets available for the Vandenberg Air Force Base site (the only site available for polar launches) being the small Delta and the large Titan. However, for its own reasons, the U.S. Air Force had decided in 1991 to improve its Vandenberg facility to take the mid-sized Atlas rocket, and this had been the opening wedge for Frieman to recommend repackaging EOS onto satellites sized for Atlases. Atlas-sized satellites would permit some, but not all, of the desired EOS sensors to be clustered together for simultaneity.

But Frieman had also been under a great deal of pressure to shrink EOS further, by splitting more of the sensors off onto small, single-instrument payloads. After accepting the responsibility of preparing his review, Frieman had been brought to Washington to interview senior members of the administration about EOS; after his return, he had told his panel members that this subject came up in several different meetings.²³ Advocates of this approach, many of whom had experience in classified programs related to Reagan’s SDI, argued that cost was an exponential function of complexity. A five-instrument satellite would cost far more than five one-instrument satellites. Hence, one could achieve large cost reductions by building more, simpler satellites.

These advocates also argued that the simultaneity claim of EOS’s supporters was overblown. Satellites could, they believed, be maneuvered to fly in formation close enough together so that their measurements would be essentially simultaneous. Formation-flying had not been demonstrated in any program known to the Space Studies Board of the National Academy of Sciences as late as 1995; nonetheless, it became part of Frieman’s recommendations. As a potential future technology, formation-flying offered a great deal of promise; a major cost of satellite building was the cost of trying to integrate many different instruments onto the same satellite bus. Limiting satellites to one or two instruments could permit many more instruments to be flown for the same amount of money—if the formation-flying idea worked out. Finally, the smaller satellite approach promised to produce some research results sooner, which would help mollify scientists and policymakers who were becoming outspoken in their criticism of the slowness of EOS, while also responding to the scientific advocates of the Earth Probes concept.²⁴

In February 1992, the Senate subcommittee on Science, Technology, and Space reviewed the reengineered EOS. As NASA’s Lennard Fisk presented it, EOS now consisted of six payloads: three Atlas-sized satellites, one Delta-class satellite, and two sized for a new small launch vehicle intended to replace the discontinued Scout rocket. The three Atlas packages were named EOS-AM, a polar, “morning”-orbit satellite instrumented primarily for surface studies; EOS-PM, a polar, “afternoon”-orbit satellite instrumented for weather and cloud studies; and EOS-CHEM, a polar, “afternoon”-orbit chemistry satellite to succeed UARS. The Delta payload was EOS-ALT, an altimetry satellite for ocean circulation and ice sheet studies, while the two smallest were EOS-COLOR and EOS-AERO, for ocean color and aerosols, respectively. These were to be launched between June 1998 and 2002. They would then be repeated to achieve the required fifteen-year coverage, so there would ultimately be three EOS-AM flights, three EOS-PM flights, and so on.

Fisk also explained that to reduce a gap in coverage of the Earth’s radiation budget, a very high priority for the IPCC, a Clouds and the Earth’s Radiant Energy System (CERES) instrument would be added to the Tropical Rainfall Measuring Mission, scheduled for launch in 1997, while a Stratospheric Aerosol and Gas Experiment (SAGE) III would be added to a “flight of opportunity,” a Russian Meteor weather satellite, as things would turn out.²⁵ Finally, in the process of repackaging EOS, NASA also shrank it, deleting instruments aimed at geomagnetism, upper atmosphere research, and solid Earth geophysics. The largest instruments, a space-based large-aperture lidar and synthetic aperture radar, were also dropped, as they could not be accommodated on the much smaller Atlas-sized satellites.

This reengineering of EOS did not silence public criticism by the scientific community. There were several points of contention, including the continued slow pace of the program, its lack of competition, and its apparent consumption of all available funding for Earth science missions for the foreseeable future, but one will suffice for discussion: launch order. The AM package, to be launched first, was primarily aimed at land and ocean surface studies, and was not directly relevant to the central question of detecting anthropogenic climate change. The PM package contained instruments aimed at that question. Climate models almost universally predicted that as the troposphere warmed, the tropopause, the boundary between the troposphere and the stratosphere, would change its average altitude, while the stratosphere would cool significantly. Due to the large errors and poor calibration records of the 1970s era satellite temperature sounders still in use, a definitive detection of this effect could not be made using their data. Radiosondes had similar problems, with inherent biases that had to be identified and corrected.²⁶ As the launch order controversy played out in the first half of the 1990s, it was clear to the climate research community that instruments capable of measuring tropospheric temperature with much better calibration capabilities were vital to the question of anthropogenic climate change.

EOS as of 1992. From Ghassem Asrar and David Jon Dokken (eds.), EOS Reference Handbook, NASA Earth Science Support Office, 1993, p. 10.

The Jet Propulsion Laboratory’s (JPL) Atmospheric Infrared Sounder (AIRS) instrument, which was expected to produce improved vertical resolution and reduced error compared to the 1970s generation of temperature sounders on the operational weather satellites, was supposed to be able to measure the predicted changes. The AIRS principal investigator, Moustafa Chahine, intended to overcome the cloud-clearance problem infrared instruments had by pairing AIRS with the Advanced Microwave Sounding Unit (AMSU) and using the microwave data in his temperature retrieval algorithms. Similarly, the atmosphere was expected to become wetter as it warmed, and a PM instrument to be provided by NOAA initially, and ultimately by Brazil, called the Measurements of Humidity Sounder, could determine whether this was happening. The humidity sounder’s data was also to be used to improve AIRS’s retrievals, and the three instruments effectively formed a single package aimed at the classical meteorological values of temperature and moisture. They also, of course, had direct climate relevance. These three instruments were to be flown with the Moderate Resolution Imaging Spectroradiometer (MODIS), which would provide cloud-top temperatures and altitudes, also likely to change in a warming world; the Clouds and the Earth’s Radiant Energy System (CERES), whose radiation budget measurements nearly everyone agreed were of the highest scientific priority and directly relevant to anthropogenic climate change; and a new microwave imager. Because this PM package was the central climate satellite, many members of the Frieman’s engineering review panel thought that it should be the first launch.²⁷

The counterargument that NASA officials had made was that the AM satellite, whose contractor, General Electric, had originally been assigned to build the large polar platforms, was already under construction and delaying it would only raise overall costs without providing any benefits. Further, its most expensive instrument, ASTER, a high-resolution land imager, was being provided by Japan and thus the AM satellite, despite having more instruments, was less expensive to NASA. It was also less technologically challenging. The AIRS instrument, EOSPM’s centerpiece, was the most difficult U.S. instrument, and none of NASA’s leaders believed that it could be accelerated successfully. Finally, the PM satellite contract award had not been made yet. Nothing would be gained by trying to rearrange the flights except raising costs. As of late 1991, it was already too late to reprioritize.²⁸

While NASA’s decision not to change the launch order stood, it did not sit well outside the agency. In June 1992, Pierre Morel, by now head of the planning staff for the World Climate Research Program (WCRP), argued that scientific priority still clearly belonged to the PM satellite. In early 1993, Science reporter Gary Taubes reported that many members of Frieman’s committee did not think NASA had taken their recommendations seriously, and while EOS was now much smaller, it was still not focused on the correct set of questions.²⁹ Hence the larger scientific community continued criticizing EOS for its misplaced priorities and lack of a tight focus on detection of anthropogenic climate change; here the mid-1980s strategy of trying to produce long-term, continuous, and comprehensive datasets for use by all parts of the Earth science community was working against EOS. Instead, with budgets shrinking, the community sought the opposite: an observing system aimed at a narrowly drawn, specific set of questions.

RESCOPING AND REBASELINING EOS

Shortly after the Senate review of the reconfigured EOS, NASA received a new administrator, Daniel S. Goldin. Vice President Dan Quayle had had visions of building his own political future on a vastly expanded space program aimed at Mars colonization, and he had found Administrator Truly to be unsupportive. Truly was too wedded to the high-cost Space Shuttle to undertake reforms Quayle thought necessary. Hence, Quayle had attempted to arrange a bypass around Truly by issuing a directive specifying that NASA’s assistant administrator for exploration communicate directly with the Space Council, which Quayle chaired, instead of going through Truly’s office.³⁰ When this workaround was exposed, it was not seen favorably either in Congress or in the White House. Truly was then removed. After a prolonged search, Quayle chose Goldin. Goldin was an advocate of what he called “faster-better-cheaper,” an effort to do more small missions with less money.³¹

Goldin then embarked on what was known as the “Red team/Blue team” study of all the agency’s programs. In essence, all of the agency’s programs were reviewed by two teams, a Blue team drawn from the program in question and a Red team composed of people not associated with it. In EOS’s case, the Blue team was led by Goddard Space Flight Center’s Chris Scolese, while the Red team was chaired by JPL’s John Casani. The Blue team reviewed its own program and then defended it before the Red team; the Red team did an independent review and critiqued the Blue team. In EOS’s case, Goldin also required the teams to find $3 billion in additional savings. This initiated what insiders called the rescoping of EOS to separate it from the previous years’ reengineering, and which was not entirely complete before the EOS budget was cut again—a $750 million slice—this time by Congress. This cut provoked a rebaselining.³² It also led to the departure of Shelby Tilford.

This time, there was no new outside review committee; instead, EOS’s reconstruction was carried out largely internally, with Ed Frieman and some of his committee members serving as informal consultants. The first rescoping stage of the cuts was carried out by forcing all but one of the EOS payloads onto Delta-class rockets or smaller; to protect the 1998 launch schedule, EOS-AM remained an Atlas-class payload. Several instruments were cut from the EOS-PM and EOSChem satellites, and other instruments were descoped. But the most significant cuts in the rescoping exercise were made to the data system, the Earth Observing System Data and Information System (EOSDIS). This forced, or enabled, depending on one’s point of view, a fundamental redirection of the system.

EOSDIS had originally been conceived in accordance with a centralized data processing model that derived from large-scale Defense Department computing projects, such as the SAGE air defense system.³³ This model was based upon the high cost and large size of high-powered computers, in company with the very limited supply of programming talent for them during the early years of digital computing. This was known as the MITRE model due to its historical linkage with the MITRE Corporation, which had been the system engineer for SAGE and other projects like it. For EOSDIS, this centralized model meant a single, large-scale installation would develop the retrieval algorithms for the instruments, process, and then validate the datasets before providing them to offsite Distributed Active Archive Centers (DAACs).³⁴ EOSDIS would also provide satellite command and control.

By 1993, as the rescoping was taking place, criticisms of this approach had appeared. From inside the agency, a number of experienced investigators opposed the centralized model. Langley engineer Bruce Barkstrom, for example, who had been the architect for the ERBE and CERES data processing systems, believed that the programmers doing algorithm development had to be very familiar with an instrument and the data it produced to get useful results; because of the historical evolution of NASA’s various Earth science components, that expertise was not all located in one place. Langley had specialized in aerosols and atmospheric radiation, while JPL had specialized in physical oceanography and stratospheric chemistry. It made little sense to redevelop that expertise elsewhere. Further, he argued that the centralized model, like the original EOS large platform model, was unnecessarily expensive. In software development, the interfaces between modules of various programs had to be carefully controlled to ensure they remained compatible. Increasing the number of programs a given computer system had to run increased the number of interfaces exponentially, not linearly, and hence cost was also an exponential function of complexity. Finally, computing power had become dramatically less expensive, and programming expertise more widespread, in the decades since the formulation of the MITRE model. Centralization of computing resources was no longer necessary. In his view, a distributed system built around the extant DAACs made much more sense.³⁵

Similar arguments appeared in an NRC review of EOSDIS that was presented to NASA in September 1993. Shelby Tilford had requested the review late in 1991, and NRC chairman Bruce Alberts had asked Charles Zraket of Harvard University to chair the new study. Zraket’s group argued that EOSDIS’s centralized design did not have the flexibility necessary to achieve the EOS’s desired results. It was “simply an automated data distribution system” that would provide a set of standardized products to users. Researchers would not be “able to combine data from different sensors, alter the nature of the products to meet new scientific needs, or revise the algorithms used to process data for different purposes.”³⁶ The ability to do this sort of interdisciplinary investigation was one of EOS’s major selling points; Zraket’s group did not believe EOSDIS would support it effectively. Further, the growing availability of computing power made a distributed architecture both possible and desirable. This new distributed architecture model descended from the Advanced Research Projects Agency’s (ARPA) ARPANET, which by 1993 had become publicly available as the Internet.³⁷ This had demonstrated that a distributed computer network could change with extraordinary speed as users and user demands changed.

Zraket’s panel therefore recommended redesigning EOSDIS around the extant DAACs, and adding new ones if additional subject area expertise was necessary. The DAACs would be the system’s interface with EOS’s user communities, while an EOSDIS Core System would provide the network infrastructure the archive centers would require. Their vision was of EOSDIS evolving into “UserDIS,” with science users able to access datasets from multiple sources and integrate them into new products. This meant, in turn, relying on the “entrepreneurial spirit of the DAACS and other interested organizations.”³⁸ But the resulting data system would be much more flexible, able to adapt to new uses and new user-generated products as EOS itself evolved.

The redesign cost money, however, which, like the actual budget cuts, had to be paid for in capability. Dixon Butler recalls that the only really firm cost figures available were attached to processing capacity, and this was where most of the cuts were taken. EOSDIS’s processing capacity was reduced by more than half. The number of data products was reduced even further, from 809 to 128.³⁹ To a degree, the reduction in data products would be made up by researchers who could be expected to use the available datasets to create new ones. User innovation was, after all, the hoped-for outcome of the conversion of EOSDIS to a decentralized network. Lost, however, was the ability to process the huge amounts of data in real time. Instead, data releases would be made months after receipt. While this violated EOS’s original intent, it was not perceived as a great loss at the working level. Many researchers had seen the real-time goal as unrealistic, because of the time necessary to evaluate the quality of the data. Very few of NASA’s missions had succeeded at producing immediate, high-quality data outputs right after launch, a fact that had made the research community skeptical of Butler’s real-time goal.

Tilford, who was the architect of the original EOS concept of long-term, comprehensive Earth observations, by late 1992 had begun to fight the redirection of EOS toward smaller, less capable satellites and a less-than-real time data system. Tilford did not believe Goldin’s goal of shrinking all satellites down to single-instrument payloads would permit the achievement of simultaneity in related measurements; while Tilford had not liked the large platform approach, he considered its antithesis, the micro-satellite approach, to be at least equally bad. Hence, Tilford’s and Goldin’s goals were not the same.

Goldin started looking for a replacement for Tilford in mid-1993. Ed Frieman recommended one of his former students, Charles Kennel, who was an astrophysicist at UCLA. He had recently completed service on an NRC panel that had drafted a decadal survey of the field, which set priorities for astrophysical research for the 1990s and early 2000s. This had strongly supported the use of small satellites just as the 1988 Earth science report had, and of course helped make Kennel an appealing choice. Kennel flew to Washington to meet Goldin for an interview at his apartment in the Watergate hotel. Kennel remembers that this went long into the night. Goldin explained the political problems that EOS faced, both from the scientific community’s criticism of EOS and from congressional opponents of climate research, and that he was looking for an associate administrator for it who was not connected to the Earth sciences at all in order to immunize the program from criticism. Reflecting the ongoing scientific criticism of EOS, Goldin made clear that he wanted science to be in charge of the Mission to Planet Earth. He also wanted EOS refashioned to use smaller satellites, made more flexible, and finally, kept within a very restrictive budget.⁴⁰

Kennel accepted the position, intending to stay two years before returning home to California. Goldin announced the choice on 6 January 1994.⁴¹ This produced some difficulties for Kennel, as Tilford was well-known and popular in Washington science circles. Tilford was widely regarded as the developer of the concept of integrated Earth observations and had built one of the most successful research programs in the government. Criticism of the decision therefore erupted, channeled toward Kennel’s lack of experience in the Earth sciences. He was, after all, to be responsible for the largest single Earth science program ever carried out. This was resolved when he was allowed to choose Robert C. Harriss, who had left Langley Research Center for the University of New Hampshire in 1988, to be the director of the Mission to Planet Earth’s Science Division. This made Harriss responsible for decisions about what science proposals were funded. Kennel and Harriss came aboard in January 1994, initiating the next phase in EOS’s transformation.

RESHAPING EOS

Kennel joined NASA at an odd moment in NASA history. The 1992 election had put William Jefferson Clinton, a centrist Democrat, in the White House with former Senator Albert Gore, Jr., as his vice president. Gore was a self-proclaimed environmentalist and had written a book titled Earth in the Balance about a litany of environmental crises.⁴² His ascension had seemed to promise stability for EOS and a revitalized set of environmental policies. That isn’t what happened. Gore had not supported EOS’s original large platform approach, preferring a more rapid deployment of smaller satellites.⁴³ In his own interview with Kennel, he had made clear that EOS had to shrink while its science had to be protected. Gore also appointed NASA’s Robert Watson to head the environmental section of the Office of Science and Technology Policy, and Watson reiterated that theme. The presence of an environmentalist vice president did not signal safety for EOS, particularly as Gore was directly involved in the international negotiations over greenhouse gas emissions reductions that descended from the 1992 Framework Convention on Climate Change. Instead, EOS remained under pressure from the administration until the Republican Party took over the House of Representatives in the 1994 election, when its association with the activist vice president made it a target of political retaliation.

Kennel set in motion an initiative to redefine EOS shortly after arriving in Washington that resulted in the reshaping plan. Its core objective was to separate the definition of the system from its hardware configuration and instead define it by the science it would do. Kennel did not believe that it was possible to fly three identical sets of instruments, as had been EOS’s original design. Components went out of production, and no two instruments ever built were truly identical. Instead, what made the data from instruments comparable was the establishment of a calibration record, preferably by flying an old instrument and its replacement side by side to intercalibrate, although there were other possible approaches. This meant that new technology could be infused between instrument generations. It also meant that hardware no longer needed to be the defining factor in the system.

Within a week after arriving in Washington, Kennel was approached by Michael Luther, who was in charge of mission development for the Mission to Planet Earth, and asked whether he would support a proposal for a line of small satellites to complement the larger observatory missions. He agreed, and this became the Earth System Science Pathfinder (ESSP) series of missions. These missions were to be chosen by the competitive announcement of opportunity process. This permitted nearly anyone to propose new missions, with mission prioritization and selection done by a panel of NASA and non-NASA researchers. Kennel hoped that the smaller, less expensive ESSP missions would allow him to maintain, or perhaps expand, the science content of EOS while adapting to the imposed budget cuts. Since one major goal of the ESSP program was to speed up the process of getting new missions into space, the first announcement of opportunity was scheduled for early 1996.

Kennel also formed a science team to examine EOS again. The team was chaired by Michael King, the EOS project scientist at Goddard Space Flight Center and one of the MODIS instrument principal investigators. King had also been a prominent advocate of redefining EOS by its science, not its system architecture, and his team constructed a set of twenty-four measurements that were to represent EOS. These are given in Table 8.1. For the atmosphere, these were clouds and radiation, precipitation, chemistry, aerosols and volcanic effects, and atmospheric structure. Other measurements included surface temperatures, ocean circulation, ice and snow coverage, soil moisture, and solar irradiance.⁴⁴ The task of his science panel was to find a way to maintain this set of measurements past 2000, when EOS was likely to be held within an annual cost cap of $1 billion, and to suggest ways to migrate some of the measurements to the next-generation operational weather satellites.

The next-generation weather satellite program was called NPOESS: National Polar Orbiting Environmental Satellite System. In 1993, Vice President Gore had initiated a “national performance review” aimed at streamlining the federal government. One of the resulting recommendations was merging of the separate military and civilian weather satellite programs, and in May 1994 President Clinton had issued a Presidential Decision Directive ordering their convergence. The directive ordered NASA, NOAA, and the Defense Department to form an Integrated Project Office (IPO) to manage the program. NASA’s primary responsibility within the IPO was the provision of new technology. Since several of the EOS instruments had been intended as testbeds for the next-generation weather satellites, the formation of the NPOESS project presented an opportunity to migrate these to an operational status. Moustafa Chahine’s AIRS was one obvious candidate. Another was MODIS, which contained heritage channels from the Advanced Very High Resolution Radiometer (AVHRR) imagers on the current weather satellites but with better resolution and calibration capability. A third was the AMSU. These instruments were on the EOS-PM payload, and thus migrating these and some other relevant instruments to the post-2004 NPOESS constellation would eliminate the need for NASA to replace the EOS-PM mission, and therefore reduce its post-2004 financial needs.⁴⁵ NPOESS was not a complete solution to the climate observing system problem, which required oceanic, cryospheric, and chemical measurements as well, but it was a start.

TABLE 8.1.
EOS Measurements

The Republican victory in the fall 1994 congressional elections gave Kennel still another challenge. The minority Republicans on the science committees became the majority members, and opponents of climate science took over the committee chairs in January 1995. The new Republican leadership immediately targeted the U.S. Global Climate Research Program, and NASA’s Mission to Planet Earth, for termination. This anti-environmental revolt had been brewing since the early Reagan administration, when the so-called “wise use” movement began. This had sought the rollback of land-use restrictions in the American West, where most of the nation’s public lands were located.⁴⁶ Later in the decade, and well into the early 1990s, conservative propagandists had attacked the anthropogenic ozone depletion hypothesis in the public arena, as discussed in chapter 6. In this highly charged context, the new chairman of the House Science committee, Robert S. Walker, relied on claims by members of the George C. Marshall Institute that solar irradiance changes would be responsible for whatever warming might happen in calling for a cut of $2.7 billion from the Mission to Planet Earth 1996–2000 budgets. He also scheduled hearings designed to showcase the arguments of fringe scientist-contrarians on both ozone depletion and climate change.⁴⁷

But both Bob Harriss and Ed Frieman recall that Walker had other concerns beyond the nakedly partisan. The EOS concept had been based on the collection of long-term datasets whose analysis would provide EOS’s scientific return. This meant spending billions of dollars for an unpredictable return in the distant future. EOS had not been structured to provide shorter-term results that would help to justify it. It also had not been primarily aimed at developing applications that might have economic benefit or aid in the provision of some public service. There were no shorter-term results to help convince congressional leaders that the public’s money was being well-spent. Harriss had already set out to change this, carving out a small amount of money to establish an applications section within his science division and trying to recruit scientists interested in developing near-term results. Yet this was controversial within the EOS science community, which thought an applications emphasis diluted the Earth System Science concept, and in any case the strategy was just getting started.⁴⁸

For several reasons, then, Congressman Walker wrote to the National Academy of Sciences president, Bruce Alberts, in early April and requested another review of EOS and of the larger U.S. Global Change Research Program by its Board of Sustainable Development. The chairman of the board was Ed Frieman, who had chaired the reengineering study of four years before. Frieman accepted the charter, and his review took place in July 1995.⁴⁹

The EOS principal investigators met to prepare for the questioning they would face, in Santa Fe, New Mexico, at the end of June. Kennel explained the situation to the group, pointing out that the proposed $2.7 billion cut would allow completion of the first mid-sized satellite, AM 1, plus Landsat 7 and the Tropical Rainfall Measuring Mission, but would not permit completion of the remaining constellation. The basic concept of an Earth observation system would die from a cut this size. The only bright spot was that the Senate science committee was not inclined to go along with the House’s cut, and there was therefore a good chance that it would not happen. But NASA and the EOS investigators needed to convince Frieman that EOS could not tolerate any further cutting, and that it was finally in a sustainable form aimed at appropriate scientific objectives.

Frieman’s Board on Sustainable Development met to investigate EOS at the Scripps Institution of Oceanography on 19–28 July 1995. Kennel, Robert Harriss, Michael King, Joe McNeal, Ghassem Asrar, and Claire Parkinson, the EOS-PM project scientist, presented the results of the reshaping initiative to the board. The reshape plan kept the first three intermediate EOS missions, AM, PM, and Chem, as multi-instrument flights while using smaller satellites to fly the altimetry and aerosols missions. New technology developed under EOS would allow future flights of the primary EOS instruments to be lighter and thus less expensive, and so there would be no need to fly three identical sets of 1980s era instruments. Further, the ESSPs, they argued, would permit expanding the range of scientific investigations while also allowing new investigators to participate in the program. Finally, they argued that construction of the AM and PM satellites was far enough along that changing them would not result in any savings. Instead, future savings would have to come from technology infusion and migration of some measurements to NPOESS.

The NASA group also argued that EOS was well-focused on a thoroughly reviewed body of scientific questions. Harriss reminded the board members that EOS and the U.S. Global Change Research Program had been formulated together and were strongly linked. NASA scientists and EOS investigators had been involved in the National Research Council’s planning for the U.S. Global Change Research Program, and EOS was the primary observational tool for the program. In their presentation, Harriss and Asrar gave thirteen areas in which EOS measurements contributed to important climate change questions, beginning with the cloud feedback problem. This they described as the largest source of uncertainty in climate model predictions, echoing the IPCC.⁵⁰

After the NASA presentations, the board members engaged in a somewhat acrimonious debate over their evaluation that continued through the report-writing and reviewing phase. Two principal points of conflict were over a perception some members of the board had that the Mission to Planet Earth was still too focused on remote sensing to the detriment of in situ measurements, and that the reformulated EOSDIS did not go far enough in decentralizing data processing and analysis. The first complaint descended from the second-class status suborbital science often seemed to have at NASA, a problem reinforced by Administrator Goldin’s announced intent to make NASA an orbit-only agency by 2000. This goal obviously jeopardized the in-atmosphere research programs that had allowed the agency to respond quickly to the Antarctic ozone hole and that was also the source of many of its new instrument concepts for future space-borne use. EOSDIS, finally, came under criticism for not placing enough control over data products into the hands of the investigators; it still appeared to be tied too much to centralized Big Engineering concepts.

After the panel settled its differences, it produced a report that called for implementing the AM, PM, Landsat, and Tropical Rainfall Measuring missions as planned and modifying the chemistry mission’s tropospheric instruments to focus on ozone and its precursors. It also called for expansion of in situ, process, and modeling studies in Mission to Planet Earth, and advocated re-reconfiguring EOSDIS to transfer responsibility for data product generation to a “federation of partners,” which could be composed of universities, research corporations, or other entities.⁵¹ Finally, they explicitly argued that the 1995 reshape exercise Kennel had initiated had achieved all the savings possible at that point in the satellite infrastructure. Further cuts would merely produce delays, result in loss of data continuity, or eliminate the technology development that offered the primary potential for future cost reductions.⁵²

Frieman’s group also sought to deflect the political critics of climate research by mounting a strong defense of the science. Their first bullet point in the executive summary stated “science is the fundamental basis for the USGCRP [U.S. Global Change Research Program] and its component projects, and that fundamental basis is scientifically sound” (emphasis in original). To reinforce their point, they included a series of science working group reports on atmospheric chemistry, ecosystems, decadal to centennial scale climate processes, and seasonal to interannual climate processes. These provided details on the scientific needs in each area. The decadal to centennial scale climate report, for example, written by Eric J. Barron, an oceanographer at Pennsylvania State University, argued that the past decade of research had shown that the Earth’s climate was variable on short as well as long timescales, and that human activities had the potential to cause further changes. He listed a series of specific accomplishments, including ERBE contributions to improving climate models and improved ability to assess the impact of volcanic eruptions. The same research had demonstrated that the Earth’s climate processes were highly complex, and the U.S. Global Change Research Program needed to establish solid understandings of both natural climate variability and anthropogenic forcings.⁵³

Ed Frieman then took the draft report to Washington and briefed it to the congressional committees. It was favorably received in the Senate, where Harriss had been actively working with western state senators interested in possible land-use and wildfire detection capabilities offered by EOS, but Walker, while reportedly impressed by the report, still attempted to impose the first installment of his desired cut, $323.9 million.⁵⁴ Walker and some of his political allies then staged what journalist Ross Gelbspan has called a “book burning” in late 1995, producing a forum in which the political critics of ozone and climate science were allowed to proclaim against Robert Watson, Daniel Albritton, and climate researcher Jerry Mahlman, among others, in advance of the release of the IPCC’s 1995 assessment.⁵⁵

The IPCC’s report was thus immediately controversial when it was released. It contained the first consensus statement by a major scientific group that human influence on climate had been detected. Their phrasing in the summary for policymakers, that “the balance of evidence suggests a discernible human influence on global climate,” was weak, but had been carefully chosen to reflect the limited current knowledge and of ongoing controversies over, for example, why tropospheric warming was clear in the radiosonde record but not in the data provided by David Staelin’s Microwave Sounding Unit aboard the NOAA series of polar orbiters.⁵⁶ Too, the summary for policymakers was subjected to a line-by-line vetting and approval process at a plenary session of the scientific working group, which included members from nations whose major export product was oil, and thus had to be as uncontroversial as possible while still getting the point across.

Nonetheless, the IPCC report was attacked shortly after its publication in 1996, this time by the fossil fuel industry–funded group Global Climate Coalition and by the Marshall Institute, whose scientists launched attacks in the Wall Street Journal on Benjamin Santer, the lead author of the assessment’s chapter on “Detection of Climate Change and Attribution of Causes.”⁵⁷ These groups accused Santer of having altered the text to deemphasize uncertainties on the basis of rumors that they never sourced, at least in public; this caused a furious exchange of electronic mail messages between William Nierenberg, one of the authors of the Marshall Institute letter, and Thomas Wigley of the University of East Anglia in Britain, co-author of the chapter, and provoked the Executive Council of the American Meteorological Society to publish an open letter in support of Santer in the Bulletin of the American Meteorological Society. They also published the complete text of the letters sent to the Wall Street Journal, and indicated how the Journal had edited them.⁵⁸

The purpose of these public attacks on the scientists involved in the IPCC process was to generate public doubt over climate science, to stifle a growing international drive to devise a new treaty that would mandate carbon dioxide emission reductions and perhaps limitations on other greenhouse gases as well.⁵⁹ They were not directly aimed at EOS. However, because EOS was part of the national climate science program it was affected by these political controversies as well. In this context of overt, extremely partisan attacks on climate science, the U.S. Senate largely blocked the House drive to eliminate EOS, agreeing to a cut of only $91 million.

For the next several years, the system architecture for EOS stabilized. A $150 million cut in the fiscal year 1997 budget affected the launch dates for the PM and chemistry missions, but did not affect the core measurements beyond the resulting delay. Kennel then found money to implement the ESSP missions in fiscal year 1996 despite the cuts by assuming that the larger EOS satellites would not need to be replaced at the end of their official five-year life spans. He explained later that by the mid-1990s most spacecraft far outlived their expected terms. The UARS was already a year past its designed life, for example, with all but its cryogen-limited instruments still operating, while the ERBE was five years past its intended life span in 1995, with the SAGE II instrument still fully functional and the ERBE instrument partly functional. Hence he had some confidence that the satellites would outlive their design lives quite significantly.⁶⁰ The advantage of assuming longer-than-expected life spans was that it meant the replacement satellites did not have to be built as quickly, freeing resources for other uses.

From the first ESSP announcement of opportunity, NASA selected two missions for development, a Vegetation Canopy Lidar proposal to measure forest canopy density, and the Gravity Recovery and Climate Experiment (GRACE) to make improved measurements of the Earth’s gravity field. These were cost-capped at $60 million and $86 million, respectively, and were scheduled for launch in 2000 and 2001. Each mission represented a science area that had been left out of the post-1991 EOS design. GRACE, for example, was chosen because its measurements would be beneficial to the investigators on the EOS altimetry missions, improving the quality of their ocean circulation measurements. It could also play a role in ice sheet studies. A satellite pair, finally, it would demonstrate formation-flying, with high-precision spacing between the two spacecraft provided by laser ranging.⁶¹

Kennel returned to UCLA as the university executive vice chancellor at the end of 1996, and his deputy for engineering, William F. Townsend, succeeded him as acting associate administrator for Mission to Planet Earth. Goldin tried for about a year to find a university scientist to replace Kennel; discovering that no one wanted to move to Washington to take the highly controversial job, he appointed Ghassem Asrar associate administrator, and Townsend became deputy director of Goddard Space Flight Center. Harriss left soon after, moving to Texas A&M for a few years before finally settling at the National Center for Atmospheric Research (NCAR) in Colorado.

As it stood when Kennel left, EOS consisted of a set of twenty-four measurements, which were packaged into the (unchanged) AM payload, a reduced six-instrument PM payload, a still smaller four-instrument Chem payload, Landsat 7, and a set of smaller satellites: IceSat, a laser altimetry satellite for ice sheet measurement; Jason 1, a replacement for the highly successful TOPEX-POSEIDON ocean altimetry satellite; and SAGE III, which was to go into space on a Russian weather satellite. Ocean surface wind measurements were to be made by SeaWinds, a scatterometer derived from SeaSat A, on the Japanese satellite ADEOS 2, and QuikScat, a rapid replacement for the failed ADEOS 1. Finally, negotiations for an integrated, global climate observation system patterned after the international weather satellite system were just beginning. These would take more than a decade to bear fruit.

TRIANA AND THE POLITICS OF CLIMATE SCIENCE

One morning in March 1998, Vice President Gore woke up with a vision of a satellite in the L-1 position between Earth and Sun. At L-1, the gravity of the Earth and Sun exactly balance, permitting a satellite to remain there indefinitely.⁶² His satellite would continuously stream an image of the sunlight Earth onto the Internet, and Gore had suggested the idea to NASA administrator Goldin. Gore thought such a satellite would have both scientific and “mystical” value. Considering the proposal a challenge to demonstrate the agency’s newfound flexibility, Goldin had then had the newly renamed Earth Science Enterprise issue an announcement of opportunity for a science mission to use L-1 for Earth observation. This mission became “Triana” to its supporters, “GoreSat” to its detractors, and a metallic symbol of the intense politicization of Earth science at the turn of the century.⁶³

NASA received nine proposals for the mission in response to the announcement, and chose one from the Scripps Institution of Oceanography. Francisco Valero had proposed a satellite with three instruments: a narrowband spectrometer with channels chosen to replicate MODIS and Total Ozone Mapping Spectrometer (TOMS) channels for aerosol, cloud, and ozone study; a set of broadband radiometers for albedo and radiation budget studies; and a solar plasma instrument. The advantage of these choices was that they would provide the same data as MODIS, TOMS, and CERES, but for the entire sunlight side of the Earth simultaneously. From their positions in low Earth orbit, these instruments saw swaths of the Earth, which then had to be stitched together mathematically to arrive at a full “Earth” of data. This introduced errors that data from simultaneous viewing would not have, and thus the L-1 satellite’s data could be used to check and correct the data produced by the other satellites. Further, the Moon would occasionally occlude the satellite’s view of Earth, and since the satellite would only see the same part of the unchanging lunar surface, it made an excellent calibration target—several, but not all, EOS instruments already used it for that purpose.⁶⁴ On the strength of its benefits to the other EOS satellites, the Scripps proposal had been accepted.

“GoreSat” was immediately challenged by congressional Republicans led by Dave Weldon of Florida, who inserted an amendment to the fiscal year 2000 budget canceling it. This passed the House but not the Senate; the resulting conference committee report barred NASA from spending money on Triana until the National Academy of Sciences passed judgment on its scientific merits. It also barred NASA from launching the satellite prior to 1 January 2001, after the 2000 presidential election, exposing the political relevance of the mission: Gore was the expected Democratic contender in the race to succeed President Clinton. On 14 October 1999 Ghassem Asrar wrote to Bruce Alberts, president of the National Research Council, requesting the study; on 3 March 2000, the review panel, chaired by James J. Duderstadt of the University of Michigan, responded with a letter report. This was positive, although not without significant technical caveats, and the satellite development restarted.⁶⁵

This was not the last criticism of Triana, however. In a letter to Science, science policy scholar Roger Pielke, Jr. and former science director for Mission to Planet Earth Robert C. Harriss criticized NRC for failing to adequately carry out its tasking. The review panel had specifically refused to examine Triana’s probability of success, or, more important, its priority relative to other possible Earth science missions. NASA typically based the research topics proposed in an announcement of opportunity on the relevant National Academy of Science decadal survey to ensure a given mission had the support of the scientific community. In Triana’s case, it had not done so. The Earth science survey of 1988 had contained no suggestion that L-1 might be good for Earth viewing (although L-1 was already in use by several Sun-viewing missions). While this was possibly an oversight, or lack of imagination on the Earth science community’s part, the fact that Triana had skipped the usual scientific vetting and prioritization without drawing condemnation from the National Academies was very troubling to Pielke and Harriss. It was not at all clear to these two that Triana would have been rated by the community above a synthetic aperture radar mission, for example, which evidence from the European and Canadian Synthetic Aperture Radar (SAR) satellites, as well as JPL’s DC-8 based SAR, suggested could provide earthquake warnings and measure the flow velocities of large glaciers. Hence, they criticized NRC for not carrying out what they perceived as its responsibilities to provide adequate advice to Congress.⁶⁶

Instruments, platforms, and expected launch dates for EOS as of 1999. From Michael D. King and Reynold Greenstone, eds., 1999 EOS Reference Handbook, p. 20. Courtesy NASA.

NASA’s bypassing of the policy process in Triana’s case left it open to attack by opponents of climate research; similarly, it left NRC and its parent organization, the National Academy of Science, open to attack. Further, their refusal to criticize Triana, EOS, or the larger agenda of climate research was taken as evidence of partisan intent by members of the American political right. By this time, bizarre theories had circulated in right-wing circles that the National Academies and the editors of the major science journals, including the British journal Nature, were members of a pro-Gore conspiracy.⁶⁷ Hence despite Kennel’s and Harriss’s efforts in the mid-1990s to make sure science drove EOS, by 2000 it was clear to right-wing political activists and the politicians they communicated with that precisely the opposite was the case. They believed politics drove EOS and climate science.

TRANSITIONS

During the Triana conflict of 1999, the newly renamed Earth Science Enterprise’s leadership also formalized the first part of its transition strategy to move some of the EOS measurements to the NPOESS. The NPOESS project office had selected a somewhat different set of sensors for its weather satellites than had been chosen for EOS-PM, the satellite containing the primary EOS weather measurements, and hence the EOS-PM instruments would not serve as the prototypes for NPOESS. Further, the differences meant that the NPOESS instruments would not have had their scientific qualities and calibration capabilities demonstrated by EOS, or their algorithms validated. In a sense, they represented starting over, redoing for new instruments what was already being done for EOS. However, the NPOESS project office was not a scientific research organization, and it was also not required to produce a climate-quality, as opposed to a weather-quality, dataset. To fulfill the demands of climate science, Ghassem Asrar and Bill Townsend formulated one more new, medium-sized EOS mission, inelegantly named the NPOESS Preparatory Project (NPP).⁶⁸

The year before, the National Academy of Science’s Committee on Global Change Research had acknowledged the fiscal improbability of maintaining the EOS constellation indefinitely and had called for a single additional medium mission to serve as a bridge between the EOS-PM mission and the first NPOESS launch in 2009 to ensure data continuity.⁶⁹ The NPP satellite would carry copies of the instruments chosen for NPOESS, and assuming everything went according to schedule, it would overlap the life spans of the EOS-PM satellite and the first NPOESS satellite. This overlap would permit cross-calibration between the EOS-PM and NPP satellite, then the NPP and NPOESS satellite, effectively moving the calibration forward in time. This was considered essential by the climate science community to ensure comparability of measurements made by the three satellites. Further, the NPP’s science teams would generate the data reduction algorithms for the NPOESS sensors and they would also define and produce the climate-quality data products for the new sensors. Indeed, this was one of the central weaknesses of the NPOESS transition, as far as the climate science community was concerned. It was not within NPOESS’s managerial responsibility to support climate-quality data production and archiving, let alone research using the data. NPP, of course, was a project limited in time span, not a permanent research organization designed to sustain climate research for the foreseeable future. NPP and NPOESS were therefore at best partial solutions to the problem of how to maintain climate-quality data production in a post-EOS era.⁷⁰

The National Academy of Science’s 1998 study had also recommended that EOS be restructured into smaller, more focused missions along the Earth Probe and ESSP lines.⁷¹ A series of planning workshops took place during 1998 and 1999 to map out a new direction for NASA’s Earth science research. The strategy that emerged was to use NPOESS to maintain key long-term measurements and use smaller Earth Probe and ESSPs to develop new measurement capabilities that might be migrated to NPOESS, or maybe NOAA, or perhaps some as-yetundefined climate agency later. These might include carbon cycle measurements or soil moisture measurements; the new strategy also called for new technology investment to reduce the cost of maintaining the observation system in the future. It left unclear, however, how the new measurements would be transitioned to operational use or to what agency they would be sent. By 2000, therefore, NASA no longer planned to replace the four EOS mid-sized satellites when they reached the end of their lives. The budget cuts, congressional hostility to climate science, and scientific hostility to the EOS approach left the agency without the political support necessary to sustain the program through its intended fifteen-year span.

EOS-AM was finally launched in 1999, two decades after planning for a climate observing system had started. It was renamed Terra. This very long lead time for space hardware limited the agency’s ability to respond to important policy questions. As the global warming controversy heated up during the 1990s, NASA was not in a position to respond with new global observations; it could not even accelerate measurements that it had in progress. To the larger scientific community, it had been clear that the PM mission had contained the highest-priority instruments for detection of climate change, but having started down a different path, the agency was effectively trapped by its prior history of decisions. This fact had left EOS vulnerable; as Representative George Brown had pointed out in his letter, the clear lack of consensus, either technological or scientific, of how EOS should be structured made it difficult to defend even when the basic concept of a comprehensive observing system was politically favored. EOS thus shrank, first to a set of four medium satellites plus a flurry of small satellites, then to one that was essentially Vern Suomi’s concept of 1980—a climate observing system based upon the weather satellites plus research satellites.

When Shelby Tilford had launched the EOS effort in the early 1980s, the Reagan administration had reverted to the grandiose space engineering dreams of the early space race, encouraged by unsubstantiated, and unrealized, claims of inexpensive, reliable, and routine space access via the Space Shuttle. This had bound the project to the larger Space Station Freedom initiative, which itself did not survive without severe descoping. The EOS architecture had originally been a means of promoting the use of humans in space. It had not been chosen as the most efficient means to accomplish a given set of scientific goals. Nor had it been constrained within a specified budget. Instead, “no arbitrary constraints” had been applied to the project during its earliest phases, leading to an architecture that became politically unsustainable as soon as the Cold War demand for space spectacles vanished. Big Engineering was EOS’s Achilles heel.

Further, because the initial EOS conception was so expensive and long-term, it was widely seen in the scientific community as locking in funding to a select group of researchers and locking out everyone else. To many of the scientists who were not selected at the 1988 announcement of opportunity, there would be no more opportunities within their working lives. This violated scientific norms of fairness and competition and left no means of introducing graduate students and new researchers to space science, creating resentment that further undermined EOS. There were also sound technical reasons for having alternatives to the giant platforms. The polar platforms’ Sun-synchronous orbit was not useful for certain science missions, such as ocean topography, and the lack of alternatives within the original architecture meant no means of carrying out these other missions. NRC’s expressed demand for smaller, competed missions—eventually the ESSPs—reflected these frustrations.

Finally, the fact that environmentally relevant science had become an issue associated with only one political party during the 1980s began to harm NASA’s atmospheric science programs. The brief 1970s, when both parties had supported environmental improvement while disagreeing over regulatory methodology, was long over. Instead, environmentalism, and any branch of science that touched on the natural or human environment, became merely another partisan issue. NASA’s decision to make itself the lead agency for atmospheric and climate sciences in the late 1970s had brought the wrath of the new majority party down on it in the 1990s. This would only get worse as EOS began to fly.