Reconsidering the Rules for Space Security

Plausible Prospects

Back to table of contents
Authors
Nancy Gallagher and John David Steinbruner
Project
Reconsidering the Rules of Space

Could dominance in space be achieved if adequate resources were provided and effectively managed? American space enthusiasts are fond of citing the Apollo program as evidence that the United States can mobilize the economic resources to overcome major technical challenges on a tight timeline when its leaders have a bold and inspirational vision. Some basic physical laws and technical facts impose unavoidable constraints on space operations, however. They give reason to question whether U.S. military space dominance could be achieved with any plausible multiple of the current effort, regardless of how well it might be managed.

The requirements of launching, maneuvering, and operating satellites in space impose greater burdens and more constraints than are encountered in other environments. In order to stay aloft, satellites to be inserted into LEO (100–2,500 km) need a velocity of 7–8 km/sec—thirty times faster than a passenger plane—when they reach their target altitude. Those to be placed in geosynchronous orbit (36,000 km) often are first launched into a LEO parking orbit, then maneuvered into their operational position, where they travel at about 3 km/sec. The energy necessary to impart these velocities is substantial—to launch a satellite into LEO, approximately 45 tons of propellant are needed for every ton of payload. Fuel requirements for each subsequent maneuver increase exponentially with the amount of velocity change (∆V) needed to modify the satellite’s orbital altitude. Low ∆V maneuvers such as changing orbital altitude might use a mass of propellant equal to 10 percent of the satellite’s mass, whereas high ∆V maneuvers such as changing a satellite’s orbital plane could require a mass of propellant many times greater than the mass of the satellite itself.172 These fuel requirements create practical limits on satellite weight and maneuverability.

Although ballistic missiles and launch rockets are similar, the extra velocity required to put a satellite in orbit makes that job harder in some respects than hitting a target on Earth or in space. As a general rule, a ballistic missile that can deliver a payload at maximum range R could loft that payload to an altitude R/2 and would need significantly more thrust to get that payload up to orbital speed. Modern rockets typically can put into LEO only satellites whose weight is a small fraction of the rocket’s total mass at liftoff and can put less than half that much satellite mass into GEO. Some short-range missiles, such as the Scud B (range, 300-km), could disperse a cloud of small pellets at lower LEO altitudes where they could cause significant damage to speeding satellites, but the Scud would not have enough velocity for its pay- load to remain in orbit. If a developing country wanted to convert an intermediate-range missile into a KE ASAT, it would face severe technical challenges in building a kill vehicle that could home in on the satellite and guide itself precisely enough at high speed to intercept the satellite.173 Countries that want to parlay their missile programs into indigenous space-launch pro- grams—such as North Korea’s unsuccessful attempt to launch a LEO communications satellite in 1998 and India’s efforts to develop cryogenic rockets that could put heavy satellites into GEO—will necessarily also extend their missile capabilities even if that is not their primary objective.174

Space activities are intrinsically expensive because of the specialized components required to operate in space, the high costs of launch, and the many uncertainties involved. Available estimates of the initial cost of a satellite can range from $15 to 20 million or more for a small satellite, to $100 million for a typical commercial satellite, to billions of dollars for a sophisticated spy satellite.175 The figure most commonly used to represent the magnitude of launch costs to LEO is $10,000 per pound or $20,000 per kilogram of payload (the satellite plus the fuel needed for maneuvers), with satellite weight ranging from around ten kilograms for nanosatellites up to 4,000 kilograms for the MILSTAR communications satellite. The generic per-kilogram launch figure is highly misleading, though, because launchers rarely fly fully loaded, and the specifics of orbital altitude and inclination also affect launch costs. Actual per-kilogram costs for commercial non-GSO launches throughout the 1990s ranged from roughly $10,000 to $55,000 and higher, while per-kilogram GSO launch costs were around $25,000 by the end of the decade.176 Little public information has been provided on actual commercial launch costs since 2000, and no comparable data are available for military launch costs, which could be significantly higher due to lack of competition or lower due to hidden government subsidies. Any use of space involving heavy satellites, large constellations, or significant maneuvering would include launch costs at least in the high tens to hundreds of millions of dollars.177 Finally, insurance for launch and in-orbit operations currently adds about another hundred million to the cost of an average commercial satellite—a figure that could easily rise much higher if debris, space traffic management problems, or space warfare increased the risks associated with space operations.178

Exorbitant costs have long been considered a major impediment to the realization of transformative space ambitions, whether they involve the widespread commercialization of space envisioned at the end of the Cold War, the total U.S. military dominance of space currently sought by SPACECOM, or the colonization of space that inspires some futurists. Yet, decades of effort to dramatically reduce launch costs have produced remarkably little change.179 The two highest profile U.S. government efforts to develop lower-cost launch options, the Space Shuttle and the EELV programs, used wildly optimistic assumptions to project huge cost savings that were not achieved.180

Many theories have been proposed for why launch costs have remained so high and whether they could be reduced enough to initiate a virtuous cycle of increasing use of space at decreasing cost. The most likely reasons, though, involve basic characteristics of space activity that would be difficult to change. Aerospace analyst Peter Taylor reviewed 19 explanations and concluded that the “principle proximate cause” is the “lack of intact abort capability;” that is, space flight is much more expensive than air flight because most problems cannot be fixed after takeoff. Because space vehicles are complex systems designed on the technical edge to maximize performance-to-weight ratios, multiple redundant subsystems and a huge “standing army” of experts are used to make sure that nothing goes wrong. Yet, these safeguards add new design challenges, increase complexity, and create new potential reliability problems.181

John London offered a similar technical explanation compounded by a public goods problem created by the size of initial investment needed for substantial technological innovation in space. Making extensive use of advanced technology to reduce significantly the recurring costs of spaceflight would require huge development costs that are hard to justify without unrealistic assumptions about future rates of use and recurring costs. When governments consider funding extremely expensive development projects, strong pressures are exerted to use existing technologies and personnel, so recurring costs end up being as high as they were before.182 Private industry might be more innovative and cost conscious, but it is less able and willing to invest heavily in high-risk development efforts without a guaranteed market for launch services, and it is more likely to operate in ways that increase reliability concerns.183

SPACECOM supporters understand that high launch costs—and related issues such as length of time to launch a satellite—pose serious challenges for their vision of space dominance. They propose to develop “operationally responsive spacelift” (ORS) to provide “orders of magnitude reduction in cost, significant improvements in responsiveness and greater reliability” so that they could quickly replace damaged satellites, meet short-term specialized ISR needs, and afford to deploy much larger constellations of satellites than are currently practical.184 They hope to accomplish this, however, with low initial government investments. In the acquisition plan that AFSPC deemed “unexecutable” (see Figure 4), the “space operations vehicle” budget category balloons after 2009. Therefore, AFSPC wants to defer any significant spending on new launch vehicle development until 2020 or later, and is instead awarding small contracts for preliminary ORS work and hoping that entrepreneurs will absorb most of the up-front development costs, a strategy that appears likely to be derailed by the problem London identified.

Many companies submitted ORS concept proposals to DARPA in 2003, but the two that remain in the design competition are a long way from meeting the ORS goals, so the one ORS launch that has occurred used one of Orbital Science’s Minotaur rockets.185 AirLaunch LLC has completed successful test drops from Air Force C-17 cargo planes but has yet to demonstrate that it could put satellites into orbit without expensive modifications to the planes, a problem that led to the cancellation of a prior DARPA air-launch project called RASCAL. This approach could, in theory, be an attractive way to launch lightweight ISR satellites, but it could not be used for heavier communications or early-warning satellites. Moreover, achieving the responsiveness goal (launch on a few days’ notice) could require dedicated aircraft on stand- by, which would significantly raise overall per-launch costs unless the number of ORS air-launches is unexpectedly large.186

The other remaining ORS contender, the SpaceX Corporation, claims that it will be able to launch small satellites for around $7 million and larger satellites for $27 to $78 million. These price projections are questionable, though, because the first test of SpaceX’s small Falcon 1 rocket failed in March 2006 and, after several postponements, the second test flight failed to reach its intended orbit.187 Elon Musk, the owner of SpaceX, has invested $100 mil- lion of his own money and hopes to recoup his investment by developing another rocket (Falcon 9) that can compete with Boeing and Lockheed Martin for the more lucrative heavy-launch government contracts. Given the high cost and value of its satellites, the U.S. government wants to see a track record of 98 percent reliability for any commercial rocket that would compete with the EELV. Few customers who are not mandated to buy American launch services are likely to risk an expensive satellite on an unproven rocket when Russia already offers reliable GEO launches in the $70 million range.188 Despite Musk’s deep pockets and record of success in other high-tech ventures, making long-term space policy decisions based on the assumption that he, or anyone else, will finally succeed in reducing launch costs by a factor of ten any time soon is probably not prudent.189

The physics of space also have important implications for the technical requirements and costs associated with different types of space operations. We have already seen why it is much easier and less expensive to use space for purposes that involve collecting and transmitting information over long distances compared with purposes that involve transporting large amounts of mass from the Earth into space (e.g., space-based global strike weapons) or significant maneuvering in space (e.g., military space planes or inspector satellites). The physics of space also affect choice of orbital altitude for different types of applications, numbers of satellites needed for episodic or continual coverage, and the relative difficulty of conducting and disrupting space activities. Some of these considerations apply to all types of space operations, while others vary depending on the orbital altitude that is best suited for a particular type of operation.

All satellite systems have a number of components that must smoothly function together for effective operation. These include the satellite itself, the ground station used to control it, and the up- and downlinks used for communication between the satellite and its control station as well as other receivers on the ground. The handful of incidents most commonly cited as real-world examples of space warfare include cases where ground-based jammers were used to overpower the GPS signals being sent down to ground-based receivers and cases where ground-based jammers were used to prevent satellite transponders from receiving signals being sent up for broadcast back down. A hostile state or terrorist group would need relatively little technical sophistication to attack a ground station or conduct some types of electronic interference. Standard military measures can be used against these kinds of low-tech terrestrial threats, albeit at additional expense.

Satellites, however, are intrinsically more vulnerable than terrestrial systems for performing similar functions. Because satellites move at high speeds, accidental or deliberate collisions with even tiny objects can have very damaging results. Satellites naturally move along a predictable path, and most can be tracked by amateur astronomers, so secrecy is not a reliable source of protection. Other means of passive protection commonly applied to planes, tanks, submarines, and ships, such as evasive maneuver or hardening, not only increase satellite and launch costs but also involve performance penalties and major practical constraints. Developing so-called bodyguard satellites is not be a reliable solution because of the difficulty of doing enough real-world testing to have confidence they would work and because they would be unlikely to provide persistent protection from a determined adversary. Finally, repairing a satellite in orbit is practically impossible, and replacing the more valuable types of satellites could easily take years and cost hundreds of millions of dollars.

The number and placement of satellites in a constellation depends on their function, the territory to be covered, and the desired frequency of coverage. Continuous coverage of the entire Earth except the polar regions can be achieved with just three GEO satellites. This orbit is uniquely valuable for broadcast services and for communications systems that support users from widely different, nonpredetermined locations. Orbital physics limit the number of satellites that can be stationed in GEO, however, creating controversies over the allocation of scarce orbital slots and radio frequency spectrum both among different space-faring countries and between military and nonmilitary users.190

To defray the high costs of launching a satellite into GEO, commercial operators need to carry enough transponders to serve many different customers. If one of these communications satellites malfunctions, the consequences can be far-reaching. For example, a 1998 anomaly with a processor on PanAmSat Corporation’s Galaxy IV satellite disabled most pagers in the United States for several days and prevented a major oil company’s customers from paying for services at the pump.191 Multitransponder satellites also pose a practical problem for counterspace operations because efforts to deny commercial satellite communications services to adversaries could also affect friendly and neutral users, which would violate international law even during wartime.

Some space applications, such as mobile telephone service, space-based missile defense, or high-resolution imagery, are best done by satellites in LEO.192 The rapid speed with which LEO satellites move relative to the Earth means that the lower the orbit, the more satellites are needed to ensure that at least one is in position at any given time. The Iridium mobile phone system got its name because the constellation design required 77 satellites in 665 km polar orbits to provide anytime, anywhere coverage without excessive transmission delays or power requirements.193 An American Physical Society(APS) study group calculated that at least 1,600 space-based interceptors stationed much closer to Earth (300 km) would be required to stop a single liquid-fueled intercontinental ballistic missile launched from Iran.194 Although LEO satellites for some applications can be smaller, lighter, and less expensive than GEO satellites, the number needed to avoid absentee problems makes the total cost of a constellation quite substantial. Each Iridium satellite weighed about half a metric ton and was worth $45 million.195 The APS study group calculated that a 1,600-interceptor system would require a total mass in orbit of at least 2,000 metric tones, necessitating at least a five- to ten-fold increase in total current U.S. annual launch capacity just to deploy this particular space system.196

Satellites in LEO are close enough to Earth that they would be vulnerable to a variety of ASAT attacks if legal and normative protections disappeared. Lasers can be used to temporarily dazzle or permanently blind optical sensors on remote imaging satellites. Moreover, any satellite in LEO could be damaged or destroyed using a missile that was much less capable than the rocket used to launch that satellite. A country with short-range missile capabilities could use an indiscriminate ASAT method to drive up the general cost and difficulty of operations in LEO, for example, by releasing a cloud of debris or detonating a nuclear explosion, but attackers would need sophisticated tracking and guidance skills to destroy specific satellites.197 U.S. military satellites are somewhat better able to avoid or withstand these types of attacks than commercial or civilian satellites are, so if deliberate interference with satellites becomes more common, the softer targets are more likely to suffer.

The combination of satellite vulnerability and the high absentee ratio in LEO poses particular problems for space-based missile defense, because an adversary could create a hole in the constellation by destroying a few interceptors (or inducing them to fire in self-defense or at a decoy missile), then launching through the hole the next time it passed over a launch site. Satellite absenteeism also exacerbates the cost-effectiveness problem with missile defense, because designing a space-based interceptor system that could stop two missiles launched simultaneously from the same location would require twice as many satellites as a system designed to intercept only a single launch at a time. Satellite vulnerability and absenteeism would also affect an offensive application of the interceptor system—that is, preventing other countries from launching objectionable satellites. But they would pose less acute problems because a missed intercept would result in a satellite in orbit that might be disabled or destroyed by other means before it could fulfill its threatening mission. Still, the physics of space make total space control essentially impossible; the physics also favor offense over defense in highly destabilizing ways.

These same physical principles place practical limits on improvements in U.S. space-based intelligence capabilities. As best as can be determined from the public record, the NRO currently operates up to three spy satellites in each of three categories: the Keyhole series of optical satellites, the Lacrosse/Onyx series of radar satellites, and the Misty series of stealth satellites.198 The optical satellites already have extremely high resolution (reportedly about ten centimeters), while the radar satellites can collect lower-resolution images even at night and in cloudy weather. The satellites need to be close to the Earth to take high-quality pictures, but this means that they can view only a narrow swath of the Earth, that they rapidly move over a given ground-track, and that they are not in position to see the same location again for several days. With only a small number of satellites in orbit, these systems are well suited for certain strategic purposes, such as early warning of troop movements, arms control verification, or episodic observation of other targets of interest, but they do not work well for some desired tactical purposes, such as tracking moving targets, keeping suspect sites under continuous surveillance, or providing warfighters with total battlespace awareness.

Increasing the number of advanced imaging satellites would reduce revisit time over high-value targets and expand the total amount of ground area that could be observed in a given time period. Given the difficulties in the NRO’s FIA program, the stopgap approach has been to pay industry to launch a new generation of commercial high-resolution satellites and to allow their imagery with better than 0.5-meter resolution to be sold only to the U.S. government. DigitalGlobe launched its first WorldView-1 satellite in 2007, and GeoEye (formerly Orbimage and Space Imaging) plans to launch its first satellite in early 2008. These advanced satellites will provide some improvements over the current generation of commercial imagery satellites, such as the ability to differentiate between different types of large military vehicles or to identify the location of an observed object with an accuracy of a few meters. But commercial firms are unlikely to launch many of these higher resolution satellites because they cost much more yet collect less imagery and the best data can be sold only to one customer.199

Achieving qualitative breakthroughs in the U.S. military’s ability to identify, understand, and address emerging security challenges would require a much more extensive program. Because satellites cannot see inside buildings, efforts to dramatically improve the utility of space-based imagery for finding and neutralizing chemical or biological agents would most likely involve taking much more frequent pictures throughout the construction of anything that might one day become a suspect site, then frequently checking for external signs of suspicious activity. The notion of an “unblinking eye in the sky” scanning the entire global for evidence of suspicious activity that requires closer scrutiny would also require vastly expanded capabilities. If satellites with one-meter resolution were used and could image both day and night, then roughly 200 satellites would be required for a six hour revisit time, assuming that every spot on the Earth would be imaged at least once every six hours. As many as 1,200 satellites would be needed to be able to image every spot on the Earth at least once an hour. Hundreds of terabytes (1012) of raw data would be collected on the six-hour schedule, while petabytes (1015) would be collected on the one-hour schedule, creating downlink bandwidth bottlenecks and requiring ten- to fifty-fold increases over current U.S. imagery data processing and storage capabilities.200 If a mix of U.S. and foreign government and commercial imagery satellites were used, lack of common standards would create potential compatibility problems. As the number of different sources of imagery data increase, integrating the information into a single coherent picture or measuring changes at the same location over time becomes more and more difficult. Finally, mountains of archived and fresh satellite data would be of little value without a comparable investment in highly skilled imagery analysts, a perennial problem in the intelligence community.201

Another transformational goal for space-based intelligence would involve using radar satellites to find, track, and target moving objects such as mobile missile launchers, especially in places where U.S. aircraft cannot easily operate. Current plans call for the Air Force, the NRO, and the NGA to jointly develop a constellation of synthetic aperture radar (SAR) satellites. First launch is projected for 2016, but significant technical hurdles remain. Differentiating between stationary and moving objects is much more difficult from space than with airborne radar because from the perspective of a satellite in a 1,000 km orbit fixed objects on the Earth’s surface are rotating at 15,000 miles per hour and mobile targets are moving only tens of miles faster. Other challenges include developing a large phased-array radar that could survive launch and deployment in space and finding a practical way to meet high power requirements. These technical challenges do not seem insurmountable, but the cost of deploying enough satellites to achieve the unique benefits of a military space radar system might well be. Although the program is still in its earliest stages, soaring cost estimates and budget constraints have already caused the Air Force to reduce the planned number of satellites from twenty-two to eight and to scale back promises about system capabilities.202

The CBO used information from unclassified studies of previous space radar concepts to assess three architectures using five, nine, and twenty-one satellites with 40-square-meter radar arrays and one architecture comprising nine satellites with 100-square-meter radar arrays. The CBO analysts deter- mined that for life-cycle costs ranging from $25 billion to $90 billion, a space radar system could increase the availability of high-resolution SAR imagery and shorten response time but could not provide continuous SAR coverage of a given region. Even at the theoretical optimal limit for signal-processing algorithms, the less expensive architectures would be able to detect targets moving at or below 20 miles per hour less than 30 percent of the time, while the detection probability for the 21-satellite constellation would be about 60 percent. Perhaps the most valuable capability attributed to space radar by its proponents, the ability to continually track a mobile missile launcher or other moving target until it could be destroyed, would require at least four or five times more satellites than are currently under consideration, with a corresponding multiplication of costs.203

Some applications are best done using a medium number of satellites in medium earth orbits (MEO). Because satellite-based navigation requires simultaneous signals from at least four locations, the GPS, GLONASS, and Galileo systems are designed to provide global coverage with twenty-four to twenty-seven satellites. China currently uses three GEO satellites plus signals from ground stations to provide regional navigation services, but it wants to add two more GEO and thirty “non-GEO” satellites in order to have a global navigation satellite capability. Under current circumstances, the primary danger for satellites in MEO comes not from human action but from nature—that is, from the physical challenges of operating for extended periods of time in the harsh radiation environment around the Van Allen belts. Despite well publicized concerns about inexpensive jammers that can interfere with local reception of GPS signals, interfering with the satellites themselves is difficult: they are too high to reach using a modified missile for a ground-based KE ASAT attack; they do not use optical sensors that can be dazzled; and they have various forms of passive protection.204 If space-based missile defense interceptors or ASATs were deployed, navigation satellites would be more vulnerable to direct attack, but many satellites would have to be disabled or destroyed to significantly degrade the system’s capabilities.205

From an economic standpoint, a single worldwide satellite navigation system operated as a global public utility would make more sense than multiple constellations with potential interference and compatibility problems. This is unlikely as long as system operators are directed to seek national security advantages by controlling access to different space-based positioning, navigation, and timing services. Current U.S. policy aims not only to ensure that its own military has more precise GPS information than other users do but also to prevent adversaries and terrorist groups from using any space based positioning, navigation, or timing services, “particularly including services that are openly available.” 206 No other country has declared the aspiration to control who can or cannot use navigation satellite information from systems that belong to somebody else, and technical factors make such selective denial difficult. Current European refusal to allow foreign participation in Galileo’s decision-making body or to permit foreign access to its encrypted government-only Public Regulated Service is, however, a major reason why China, India, and Israel are all reconsidering their involvement and why China wants its own global system.

ENDNOTES

172. New propulsion technologies under development could reduce the amount of fuel needed to accomplish different maneuvers, but none of the propulsion technologies likely to be available in the foreseeable future could be used for rapid movements. See Wright, Grego, and Gronlund, The Physics of Space Security, 71–74.

173. Jaganath Sankaran, "Requirements and Feasibility for the Transition from a Ballistic Missile Capability to an Anti-Satellite (ASAT) Capability," CISSM working paper, Center for International and Security Studies at Maryland, December 2007, http://www.cissm.umd. edu/papers/files/sankaran_ASAT.pdf.

174. In the Indian case, the proliferation concern was not that India would put cryogenic engines into ballistic missiles (the fuel is too unstable to use in missiles that must be stored for extended periods of time and possibly launched on short notice). Rather, the concern was that Indian engineers would learn things about other aspects of advanced rocket design that they could then adapt for use in their ballistic missile program.

175. Frank Morring Jr., "Smallsats Grow Up," Aviation Week & Space Technology, December 8, 2003, 46. The estimated cost of small satellites comes from Sir Martin Sweeting, director of the Surrey Space Centre. Some people claim that small satellites can be built for only a few million dollars, but this capability is not widely demonstrated yet, and the small satellites built for DOD's TacSat program have cost about $40 million apiece. U.S. spy satellites are commonly said to cost about a billion dollars, but the director of national intelligence used the higher figure in Mike McConnell, "Overhauling Intelligence," Foreign Affairs, July/August 2007, 58. The actual costs to produce a satellite or launch vehicle can be quite different from what a commercial company might charge for that launch.

176. During the 1990s, commercial launches to GSO on average used 80–90 percent of the vehicle's carrying capacity, while launches to lower altitudes used less than half of the carrying capacity even when they had more than one satellite on board. Futron Corporation, Space Transportation Costs: Trends in Price per Pound to Orbit, 1990–2000 (Bethesda, MD: Futron Corporation, 2002), http://www.futron.com/pdf/resource_center/white_papers/ FutronLaunchCostWP.pdf.

177. The Space Report 2006 (30–31) has cost data for 18 out of 55 launch events in 2005. Twelve of the 18 involved the launch of a single communications satellite into GEO for $70 million per launch. The three other GEO launches also involved communications satellites and had launch costs ranging from $40 million to $140 million. Costs for the three LEO launches were much lower ($1.15 to $13 million), but these were much lighter satellites and two out of three launches were failures.

178. Andrea Maléter, "Strategies to Mitigate High Satellite Insurance Premiums," Satellite Finance 64, December 10, 2003, 46–47, http://www.futron.com/pdf/resource_center/ reports/SatFinanceAMaleter.pdf.

179. One study that used "man-years of labor per million grams to LEO" as a metric to analyze historical trends found that launch costs have remained essentially flat since the first decade of orbital space operations. See Dietrich Koelle, TRANSCOST: Statistical-Analytical Model for Cost Estimation and Economic Optimization of Space Transportation Systems (1991), quoted in London, "Reducing Launch Cost," 116.

180. Space Shuttle proponents initially claimed that they could reduce launch costs to LEO by a factor of ten or more by using completely reusable launch vehicles that would need little maintenance and could make weekly flights. As it turned out, the Air Force added design requirements that increased the Shuttle's base cost, it is only partially reusable, it needs expensive maintenance, and it flies at most eight missions per year; it is also not available for commercial use, and each flight costs NASA several hundred million dollars. EELV proponents set a more modest goal of reducing the government's recurring launch costs by 25 percent, but this assumed that customers in a rapidly expanding commercial launch market would pay for most of the fixed costs of the EELV. Demand for commercial launch services has been much lower than expected and foreign launch providers offer comparable capabilities at much lower prices. The government's share of the total EELV program cost is now estimated to be $32 billion, nearly double the original estimate of $17 billion. See London, "Reducing Launch Cost," 136; and GAO, Defense Space Activities.

181. Peter Taylor, "Why Are Launch Costs So High?" (September 2004), http://home.earthlink.net/~peter.a.taylor/launch.htm. See also John Jurist et al., "When Physics, Economics, and Reality Collide: The Challenges of Cheap Orbital Access" (paper presented at the "Space 2005" conference, American Institute of Aeronautics & Astronautics, Long Beach, CA).

182. London, "Reducing Launch Cost," 130–131.

183. After Orbital Science's Pegasus air-launch system suffered a few early failures in the 1990s, increased oversight and improved quality control drove the cost of a small-sat launch up from $6 million to $20–25 million, making it now one of the most expensive launch options. Some strategies to lower launch costs, such as design simplification, greater standardization, and more robust design margins, could also improve reliability. But it will be hard for capital-constrained companies that must start small in terms of the size of the satellites they launch or the number of launches they do per year to develop the track record needed to persuade customers that they should put at risk large, expensive satellites in order to save some small fraction of the satellite's value in launch costs.

184. AFSPC, Strategic Master Plan, 13. The Air Force also uses the term responsive space for the more limited objective of building and launching small satellites for short-term tactical objectives, such as persistent reconnaissance in a location that is not well covered by existing ISR satellites.

185. Jeff Foust, "Operationally Responsive Spacelift: A Solution Seeking a Problem?" The Space Review, October 13, 2003, http://www.thespacereview.com/article/52/1; and Dwayne Day, "How to Tell Your ORS from a Hole in the Ground," The Space Review, December 31, 2007, http://www.thespacereview.com/article/1027/1. The acronym ORS is now used more broadly to refer to "Operationally Responsive Space," which includes the TacSat demonstration program to build and launch smaller, more affordable satellites. DOD provided its Plan for Operationally Responsive Space to congressional defense committees in April 2007.

186. A congressionally mandated review of future national security space-launch requirements noted SPACECOM's interest in ORS but found "little hard documentation that equated to a verifiable need." The review concluded that "embarking on an extraordinary effort to develop a launch system more responsive than those that already exist would not be cost-effective until needs are clearly stated, operational concepts are defined, and, most importantly, a family of candidate payloads is within view." See National Security Space Launch Requirements Panel (NSSLRP), National Security Space Launch Report (Santa Monica, CA: RAND Corporation, 2006), xix, http://www.rand.org/pubs/monographs/2006/RAND_MG503.pdf.

187. Brian Berger, "Falcon 1 Failure Traced to a Busted Nut," Space.com, July 19, 2006, http://www.space.com/missionlaunches/060719_falcn1_update.html; and "SpaceX Declares Falcon 1 Rocket Operational Despite Less than Perfect Test," Space.com, March 28,2007, http://www.space.com/news/070328_spacex_falc1test_updt.html.

188. NSSLRP, National Security Space Launch Report, 35.

189. Musk is currently absorbing the extra development costs of design and procedure changes intended to increase reliability, but this venture has already proved far more expensive than he anticipated. Even before the March 2006 inaugural Falcon 1 launch failure, Musk had invested twice as much of his own money in SpaceX as he had anticipated. He described the rocket launch business as "a shortcut to making a large fortune into a small one" but declared his intention to keep trying to reduce the cost of launch by a factor of ten in hopes of revolutionizing how space is used. See Michael Fabey, "A Space Revolutionary," Defense News, June 13, 2005, 54. A review of efforts to reduce launch costs reached the over- all conclusion that even if some cost reductions were possible with a future increased rate of flight, "it still remains difficult today to project any costs less than $2,200/kg ($1,000/lb)." See Henry R. Herzfeld, Ray A. Williamson, and Nicolas Peter, Launch Vehicles: An Economic Perspective (Washington, DC: George Washington University Space Policy Institute, 2005), http://www.gwu.edu/~spi/publications/Study%20V-5.pdf.

190. Theresa Hitchens, Future Security in Space: Charting a Cooperative Course (Washington, DC: CDI, 2004), 39–52.

191. GAO, Critical Infrastructure Protection: Commercial Satellite Security Should Be More Fully Addressed, report prepared for the Senate Permanent Subcommittee on Investigations of the Committee on Governmental Affairs, GAO-02-781, August 2002, 14, http://www.gao.gov/new.items/d02781.pdf.

192. Several companies use GEO satellites to provide mobile phone service, but they lack coverage at the northern- and southernmost latitudes, require bulkier equipment, and produce more appreciable echoes and delays.

193. Iridium is the 77th element of the periodic table. Eventually, the design was changed to require only 66 satellites in 780 km polar orbits, but the Iridium name was retained, perhaps because the 66th element, dysprosium, has a root meaning of "bad approach." Joe Flower, "Iridium," Wired 1.05 (November 1993).

194. APS Study Group, Boost-Phase Intercept Systems for National Missile Defense: Scientific and Technical Issues (College Park, MD: APS, 2003), xxxvii–xxxviii.

195. When Iridium lost two satellites due to launch failures, the cost to Motorola (Iridium's parent company) was approximately $90 million. When Iridium stopped commercial service in August 2000, it had to figure out what to do with 88 satellites in orbit (66 operational, eight backup, and 14 defunct) whose total weight topped 53 U.S. tons. See MacCormack and Herman, "The Rise and Fall of Iridium," 10, 13.

196. APS Study Group, Boost-Phase Intercept Systems, xxxviii.

197. Some countries with short- or medium-range missiles also have the ability to develop homing interceptors, whereas others would have to use less sophisticated and potentially less effective types of ASAT attacks. Detonating a nuclear weapon in LEO would create an intense electromagnetic pulse that would destroy all unshielded satellites in the explosion's line of sight, as well as a persistent radiation environment that would slowly damage other unshielded satellites in LEO. Such an indiscriminate attack would be an act of desperation for any country, but might satisfy a terrorist's desire for shock and mass disruption.

198. Little is known about the capabilities of the stealth satellites in the Misty program. The objective is to prevent adversaries from calculating when any U.S. satellite is in position to observe their activities, but amateur astronomers have sometimes been able to observe and track the first two Misty satellites launched in 1990 and 1999. The program drew congressional attention in 2004 when it was learned that the projected cost for launching a third Misty satellite by the end of the decade had almost doubled from $5 billion to $9.5 billion. That effort was reportedly cancelled in 2007. See Jeffrey Richelson, "Satellite in the Shadows," Bulletin of the Atomic Scientists, May/June 2005, 26–33; and Mark Mazetti, "Spy Director Ends Program on Satellites," New York Times, June 22, 2007.

199. Marty Kauchak, "Eyes for a Sharper Image," Military Geospatial Technology 4, No. 5 (November 19, 2006), http://www.military-geospatial-technology.com/article.cfm?DocID=1787.

200. David E. Mosher and Steve Fetter, "The Limits of Space," CISSM working paper, Center for International and Security Studies at Maryland, forthcoming.

201. Dwayne Day, "In Defense of the Beleaguered Spy Satellite," The Space Review, June 14,2004, http://www.thespacereview.com/article/161/1.

202. The GAO's 2006 report on defense acquisitions described the space radar program as involving twenty-two satellites, costing about $23 billion, and being able to "find, identify, track, and monitor moving or stationary targets under all-weather conditions and on a near-continual basis across large swaths of the Earth's surface." See GAO, Defense Acquisitions: Assessments of Selected Weapon Programs (Washington, DC: GAO, 2006), 105, http:// www.gao.gov/new.items/d06391.pdf. The 2007 version of the same report described a much smaller eight-satellite system that could "provide persistent, all-weather, day and night surveillance and reconnaissance capabilities in denied areas" at a projected cost of $17.5 billion. See GAO, Defense Acquisitions, 127. No information was provided about the technical characteristics of these satellites, so the cost estimates in the GAO reports cannot be compared to the CBO estimates outlined in CBOhttp://www.cbo.gov/ftpdocs/76xx/doc7691/01-03-SpaceRadar.pdf., Alternatives for Military Space Radar, prepared for the Senate Subcommittee on Strategic Forces of the Committee on Armed Services, January 2007,

203. CBO, Alternatives for Military Space Radar. The range of cost estimates for the five, nine, and twenty-one satellite configurations reflect differences not only in the number of satellites but in the size of the radar array and in assumptions about advances in signal-processing algorithms and cost growth in this space acquisition program. In the mobile missile launcher scenario, the number of satellites needed to find the mobile launcher before it left the launching location and to track it continually until a strike aircraft could destroy it depends on the size of the satellites' radar array, their signal processing capabilities, and their maneuverability. The CBO did not estimate the cost of a satellite constellation that could track mobile targets, but it would likely be $100–$200 billion or more, comparable to recent cost estimates for the two most expensive defense acquisition programs, the Joint Strike Fighter and the Future Combat Systems.

204. See Wright, Grego, and Gronlund, The Physics of Space Security, 165–169.

205. Geoffrey Forden found that even if the six GPS satellites most relevant for service in Beijing stopped broadcasting, users in the region would still be able to see at least four satellites for all but roughly two hours per day. See "Appendix D: Sensitivity of GPS Coverage to Loss of One or More Satellites," in Ensuring America's Space Security (Washington, DC: Federation of American Scientists, 2004), http://www.fas.org/main/content.jsp?formAction=297&contentId=311.

206. The U.S. policy calls for improved capabilities to deny hostile access "without unduly disrupting" civil and commercial access to open signals outside the area of military operations, thus tacitly acknowledging that this type of counterspace operation would have major unintended consequences. The 2004 policy is detailed in OSTP "U.S. Space-Based Positioning, Navigation, and Timing Policy: Fact Sheet," December 15, 2004, 3, http://www.ostp.gov/html/FactSheetSPACE-BASEDPOSITIONINGNAVIGATIONTIMING.pdf"