The ‘black’ world of US spy satellites

Underpinning the US’ formidable intelligence-gathering capabilities are the varied and unique spy satellites operated by the National Reconnaissance Office (NRO). This article peers into the murky world of US space-based intelligence gathering, charting its progress from the Cold War to the present.

Back in the late-1970s, a US aerospace engineer drew my attention to a large and almost featureless building in Los Angeles. This was a rather unusual Lockheed facility, he told me. Company staff assigned to work there become ‘non-persons’ and were no longer listed in the company’s internal phone directory. He did not identify the programme being conducted there, but left it to me to deduce what it was. There seemed only one logical candidate for a facility of this size and degree of secrecy. I was confident that the mysterious building was the birthplace of the then-current generation of US reconnaissance satellites.

Between 1956 and 1960 the US had depended on overflights by the Lockheed U-2 reconnaissance aircraft as its sole source of imagery of areas of interest deep in what was then the Soviet Union and its Warsaw Pact allies. When the Soviet Union downed a U-2 on 1 May 1960, this brought to overflights programme to a halt, but on 18 August 1960, the US launched was what publicly announced as ‘Discoverer 14’, one of a series of US Air Force satellites intended to develop techniques for orbital manoeuvring and payload recovery. In practice it was the highly-classified KH-1 CORONA spacecraft 9009, and was to complete the first successful mission by a US photo-reconnaissance satellite. By the time it ejected its payload of film in a recoverable capsule on 19 August 1960, the spacecraft had photographed 64 Soviet airfields, detected 26 new SAM sites, and recorded more than 4,270,000 km² of Soviet territory – more coverage than had been obtained from all 24 U-2 flights that had overflown Soviet territory.

Six days after the recovery of CORONA 9009’s film capsule as it descended to Earth by parachute following its atmospheric re-entry, the US set up the National Reconnaissance Office (NRO), a new organisation tasked with running programmes such as CORONA. Its existence was to remain classified until 18 September 1992. Since 1977, NRO photo-reconnaissance satellites have maintained a continuous surveillance of the former Soviet Union. The only gap in coverage was for 15 days in 1981. In order to provide coverage during unexpected crises, the NRO kept a CORONA and its launch vehicle ready for launch, initially within seven days, but from mid-1965 onwards within one day.

As the reconnaissance satellite programme evolved, the resolution of the resulting imagery steadily improved thanks to newer types of camera. From the original 12 m, it fell to 30 m, then 1.5 m, while the KH-7 GAMBIT series managed between 0.9 and 0.6m. The long-running KH-8 (GAMBIT-3) programme conducted between July 1966 and April 1984 saw further improvement, and finally achieved a ground resolution of around 0.1 m.

The end of the film-based era

The traditional film-based camera had been the best sensor for aerial reconnaissance until the 1970s, but gave way to cameras that used an imaging array based on charge-coupled device (CCD) technology. The resulting digital data could be recorded on magnetic tape, or transmitted by downlink. While a film-based satellite had a useful lifetime measured in months, and became valueless once its film had run out, one with an optoelectronic sensor could remain effective for years until its on-board supply of the hydrazine propellant needed to adjust its orbit had been fully expended.

Several generations of the KH-11 optoelectronic reconnaissance have been fielded. The first five were the Block I configuration, and probably had the designation KENNEN. They were launched between 19 December 1976 and 17 November 1982. Four Block II satellites were then built, one of which was lost due to a launch failure. The designations KH-11B and CRYSTAL have been reported for this configuration.

Block III represented a major improvement, and has been reported to have the designation IMPROVED CRYSTAL. Heavier than the earlier variants, they are thought to have larger fuel capacity, and a typical operational lifetime of about 15 years. Satellites launched in October 2005, January 2011, and August 2013 were reported to have been the further-improved Block IV variant.

A new spacecraft designed to have a low radar and visual signature, and reported to have the designation MISTY, may have been a derivative of the KH-11. The first example was probably the classified payload of the US Space Shuttle Atlantis on mission STS-36 which began in February 1990. The second was launched from Vandenberg Air Force Base on 22 May 1999 using a Titan IVB, but according to a June 2007 press report, the MISTY programme was subsequently cancelled by the then incoming US Director of National Intelligence John Michael McConnell.

A glimpse of the capabilities of the KENNEN series was provided in 1984 when images taken of the first Soviet Kuznetsov class aircraft carrier (the Admiral Kuznetsov) while it was still under construction at the Nikolaiev 444 shipyard on the Black Sea were leaked to the UK-based Jane’s Defence Weekly magazine. The resolution of the images was around 0.75 m.

The current capability of US imaging satellites was revealed by President Trump on 30 August 2019, when he posted to Twitter an image of a damaged launch pad at Iran’s Semnan space facility that had been taken the previous day following the explosion of an Iranian satellite launch vehicle. Working from the angle of the image, experts were able to deduce that it had been taken from a distance of around 380 km by a KH-11 satellite whose unclassified designations are USA-224 and NROL-49. The ground resolution seems to 10 cm or even smaller, they deduced.

A secret COMSAT fleet

The NRO faced the problem of how to transmit the imagery to the ground. If a direct line-of-sight datalink was used, the satellite would only be able to transmit imagery while it was in view of a ground station. This would sharply reduce its output. Rather than accept the severe limitation that such an arrangement would pose, the NRO opted for a scheme in which the satellite would transmit its imagery to a communications satellite located in a higher orbit, and the data would then be relayed to the ground.

This scheme was implemented in the form of the Satellite Data System (SDS). First fielded as the SDS-1, which operated from 1976 to 1987, it was followed by the SDS-2 from 1989 to 1996, then the current SDS-3 which entered service in 1998. NROL-61 launched on 28 July 2016 was thought to be the first example of a new generation of SDS-type communication relay.

While some SDS satellites were launched into geosynchronous orbit, most were placed in a highly-inclined and highly-elliptical orbit, with a perigee of around 300 km and an apogee of around 39,000 km. The SDS constellation currently consists of two satellites in geostationary orbit and three in highly-elliptical orbits that create a good line of sight between SDS satellites and the low-flying reconnaissance satellites, particularly when these were imaging terrain at northern latitudes – at which the line of sight to geostationary orbit was poor.

As a security measure, communications between the imaging satellites and the SDS use a frequency of 60 GHz, which is subject to high attenuation if passing through the Earth’s atmosphere. As a result, ground-based ELINT stations are not able to intercept the signal.

Radar imaging – the all-weather sensor

While the existence of camera-equipped reconnaissance satellites is well known, the existence of reconnaissance satellites equipped with radar has attracted less public attention, and only a limited amount of information is available in the public domain.

The only satellite to be launched under the pioneering QUILL programme was orbited in 1964. Images of the ground obtained by its radar were stored on magnetic tape which was returned to Earth by the bucket system used by the CORONA reconnaissance satellites. It proved the basic concept, but its data proved of limited usefulness.

A radar-imaging satellite initially designated INDIGO and finally ONYX was developed during the 1980s, and the first was launched by the Space Shuttle in 1988. Further launches followed in 1991, 1997, 2000, and 2005. Radar resolution depended on operating mode, and is reported to have been around 1 m in the standard ‘pushbroom’ mode, in which the radar takes a continuous image. A synthetic-aperture mode in which the radar takes repeated images of the same small area is reported to have given a resolution of about 0.3 m.

SIGINT satellites – spying on RF emitters

The first US spacecraft to perform reconnaissance functions of what was then the Soviet Union was not equipped to take photographs but gather radio frequency (RF) information on Soviet military radars. The existence of GRAB 1 (Galactic Radiation and Background 1), the first US signal intelligence (SIGINT) satellite, was not declassified until 1998. Launched as a piggyback payload on a USN Transit IIA navigation satellite, it was operational from September 1960 to April 1961, and was followed on 29 June 1961 by GRAB 2. Both spacecraft successfully intercepted signals from Soviet air-defence radars, and retransmitted these to ground stations.

From 1961 onwards, the US used Agena-based satellite launch vehicles similar in general configuration to the CORONA series searched for and study the signals from all types of Soviet radar. The first SIGINT launches orbited scanning superheterodyne receivers and provided on-orbit digital processing of frequency, pulse width, and pulse interval data. By 1964 a wide-band magnetic tape recorder was being used aboard these spacecraft. These sensors provided valuable information on the radar coverage of Soviet and Warsaw Pact regions, and on the Soviet ability to track US satellites. As a result, the US became aware of how intensively its CORONA satellites were being monitored.

Nine of the KH-4A and KH-4B imaging missions also included the release of electronic intelligence (ELINT) sub-satellites into a higher orbit, as did 12 of the 19 KH-9 launches. These sub-satellites allowed the US to catalogue Soviet air defence radars, and to intercept voice and telemetry communications.

Early SIGINT spacecraft had operated in low-Earth orbit (LEO), but 1968 saw the launch of the first of a series of seven satellites codenamed CANYON, that would operate in geosynchronous orbit (GSO). These were followed by a series of four RHYOLITE satellites launched between 1970 and 1978 to monitor radar, communications and telemetry transmissions. MAGNUM seems to have been a derivative of RHYOLITE, and the first was orbited in 1985.

The ORION series of ELINT spacecraft developed as a replacement for the MAGNUM series are positioned in GSO. A total of eight have been launched – the first in 1995 and the most recent on 9 April 2024. NROL-32 was the fifth satellite in the series, and may have been the first to be classified as an ADVANCED ORION.

In September 2010, Bruce Carlson, then then-current director of the NRO described NROL-32 as “the largest satellite in the world.” In practice, ORION satellites are thought to be in the 5,000 kg class, but Carlson’s remark may have alluded to the spacecraft’s main antenna, which is reported to have a diameter of around 100 m when fully deployed. This is thought to make the spacecraft sensitive enough to detect telemetry transmissions, or even the output of cellular mobile phones. It also makes the ORION series the visually-brightest objects in geosynchronous orbit.

Reshaping the NRO constellations

Following the era of the early film-return imagery intelligence (IMINT) satellites, the NRO had relied on large spacecraft that were built in small numbers, and had long operational lives. This procurement pattern created problems. According to an NRO history document, ‘The National Reconnaissance Office at 50 Years: A Brief History’ published in 2011, “the first satellite of an NRO program was in effect, both the prototype and a production item. It is also common for the NRO to make changes – often major – from one satellite in a series to the next to add a capability or fix a deficiency. In effect, NRO programs never really went into production, at least in the way most military systems did; they were always in the development phase.”

The next generation of optical-imaging and radar-imaging spacecraft was originally intended to be created by the Future Imagery Architecture (FIA) programme. Begun in 1999 with Boeing as prime contractor, and Harris, Hughes Space and Communications Company, Kodak, and Raytheon as the main subcontractors, this covered the development of optical and radar imaging spacecraft. A parallel Integrated Overhead SIGINT Architecture (IOSA) was expected to provide an improved passive RF capability.

To meet the optical-reconnaissance requirement, Boeing had proposed a more innovative design than its competitor, Lockheed Martin. However the programme encountered technical problems which led to a major redesign of the spacecraft redesign, causing inevitable delays and cost overruns, so was cancelled in 2005. By 2009 the NRO had begun the Next Generation Electro-Optical (NGEO) programme. This was intended to create a lower-risk modular system that could be upgraded in increments over the lifetime of the programme.

The SIGINT component of NRO’s plan, the Integrated Overhead Signals Intelligence Architecture (IOSA), proceeded relatively smoothly. IOSA was mainly a process of consolidating payloads onto a smaller number of satellites and integrating the ground stations. IOSA satellites were either incrementally-improved versions of existing systems, or new systems created by contractors that had many years of experience building the earlier systems. The relay-satellite component of the plan was based on incremental change, so proceeded smoothly.

By 2009, the NRO had a new optronic programme designated ‘Next Generation Electro-Optical (NGEO)’. This was intended to be a lower-risk modular solution, capable of being modified in increments over its lifetime. It involved the procurement of two new satellites from Lockheed Martin, and the procurement of more data from commercial imaging satellite operators by the US National Geospatial-Intelligence Agency (NGA). Understood to be the Block V version of the KH-11 series, the two new satellites were respectively orbited by Delta IV Heavy boosters on 9 January 2019 and 26 April 2021.

An unidentified NRO satellite launched on 14 December 2006 failed soon after reaching orbit. According to unconfirmed reports, it was a technology-demonstration satellite for the planned FIA radar-imaging series. Shortly before its predicted re-entry, the satellite was intercepted above the Pacific Ocean on 21 February 2008 by an SM-3 missile launched by the US Navy’s Ticonderoga class missile cruiser USS Lake Erie (CG-70). The official reason for the engagement was to disperse 454 kg (1,000 lb) of hydrazine propellant carried by the spacecraft. Five subsequent launches thought to involve radar-imaging payloads designated TOPAZ were conducted between September 2010 and January 2018.

Growing use of commercial satellites

In 2021 the NRO announced a programme for what it called the Electro-Optical Commercial Layer (EOCL). It was already collecting around 50,000 commercial satellite images per week under a 2010 agreement with Maxar Technologies, and now planned to expand its acquisition programme to include other US suppliers of EO imagery. A May 2022 contract selected Maxar, BlackSky, and Planet Labs to provide imagery for the EOCL programme. A further expansion was announced on 5 December 2023. This brought Airbus US Space and Defense, Albedo, Hydrosat, Muon Space, and Turion Space into the programme. The NRO planned to evaluate the companies’ capabilities, then procure imagery.

Albedo plans to operate satellites in very low Earth orbit in order to gather ultra-high resolution visible imagery, and high-resolution thermal infrared imagery. It intends to be the first company to offer aerial-quality imagery from space, and plans to achieve a resolution of 10 cm in the visible-light band.

A similar plan to procure space-radar imagery from commercial providers had been announced on 7 October 2021. Five contracts were announced on 20 January 2022. These had been placed with Airbus US, Capella Space, ICEYE, PredaSAR, and Umbra.

Future NRO spacecraft will be of diverse size, stationed across diverse orbits, and the NRO plans to make continued use of commercial imaging spacecraft. “Within the next decade, NRO expects to quadruple the number of satellites we currently have on orbit. Different sizes, different orbits, both commercial and national”, deputy NRO director Major General Christopher Povak told the Mitchell Institute in October 2023. “These satellites will deliver over 10 times as many signals and images that we’re collecting today… We’ll create more persistent coverage over any area on the Earth, provide faster and [sic] revisit rates, and increase the accuracy and fidelity of our data.”

The NRO is also making growing use of other data from civilian-operated satellites. “We have done that over the last two years with electro optical imagery. We’ve done that with radar imagery, RF data, and also hyperspectral”, said Povak.

Ever-growing constellations

One potential method of countering future anti-satellite threats is to rely on large numbers of satellites able to form a constellation that will overstretch the attack capabilities of potential threat nations. Drawing on experience from its Starlink global communications network, SpaceX developed its Starshield programme to meet the needs of the United States Space Force, the US Space Development Agency, and the NRO. Starshield uses LEO satellites to handle optical and radio reconnaissance, tasks, as well as missile early warning. The first launches were conducted for the Space Development Agency and involved satellites equipped with advanced infrared sensors intended to detect and track ballistic and hypersonic missiles.

Addressing the House Armed Services Committee’s subcommittee on strategic forces in 1 May 2024, NRO principal deputy director Troy Meink stated that the then upcoming NROL-146 mission by a SpaceX Falcon 9 booster would be the first operational launch of the NRO’s new proliferated architecture made up of small satellites. “We have already launched a number of demonstrations over the last few years to verify cost and performance to make sure we’re really comfortable and we know what we’re doing,” he told the subcommittee.

The launch took place on 22 May 2024, but at the request of the NRO, SpaceX’s video imagery of the liftoff did not show the upper part of the Falcon 9 launch vehicle. A further five launches to expand the constellation are due to take place by the end of 2024.

The future shape and size of the constellations that today’s NRO plans to use would have amazed the engineers who built the first examples of the ‘Discoverer’ series of spy satellites. For them, it was still a struggle to get a fully-functioning payload into orbit, let alone return any part of it to Earth. When the first images were successfully returned, these had a resolution of 12 m. The best resolution available today is better than 10 cm, and close to the limits imposed by the Earth’s atmosphere.

However, at a time when even Google Earth has 50 cm resolution imagery of some areas, and some commercial satellite operators will release selected high-resolution images of locations of interest to the news media, the NRO’s visual view of the world is not as exclusive as it used to be. However, given that it can be integrated with radar and SIGINT data, and will inevitably cover targets not known outside of the classified world, the NRO constellations and their output are likely to remain ‘black’ for many years to come.

Doug Richardson