From Images to Insights: Data Integration in Satellite Surveillance

Space Surveillance as “National Technical Means”

Since the 1960s, the United States, the Soviet Union, and, later, also China along with other actors, have used satellites to spy on each other, replacing more vulnerable solutions such as the high-altitude U2 “Dragon Lady” reconnaissance aircraft, which was famously shot down over Soviet territory in 1960. On the positive side, information gathered by satellites, encompassed by the term “verification by national technical means (NTM)”¹ in arms control treaties, has increased transparency by helping to verify or refute claims regarding missile deployments and military bases. Historically, these capabilities remained closely held secrets; neither the raw imagery nor the sensor specifications were shared, in order to prevent adversaries from gauging the technology’s limitations.² Yet, as the nuclear superpowers appeared to be broadly compatible when it came to technological capabilities, it is reasonable to conclude that the information each gathered made a significant contribution to the stability of the nuclear balance.³ The first bilateral nuclear arms control agreements—SALT (Strategic Arms Limitation Talks, 1972) and the ABM Treaty (Anti-Ballistic Missile Treaty, 1972)—relied solely on national technical means for verification, with only the later INF Treaty (Intermediate-Range Nuclear Forces Treaty, 1987) and the START I (Strategic Arms Reduction Treaty, 1991) adding on-site inspections to NTM. At the same time, the treaties prohibited restrictions of NTM in the field of strategic weapons, including through concealment. However, monitoring another country’s strategic arsenal has not been the only use of satellites. Today, reconnaissance satellites are deployed for a much wider range of purposes, including detecting and tracking troop movements on the ground, in the air, and, within limits, at sea; assessing the destruction of infrastructure (“battle damage assessment”); and providing early warning of missile launches.

Among the key players, the U.S., China, and Russia lead in sheer numbers, each operating a fleet of more than 100 military satellites, including specialized reconnaissance satellites. Of the countries with smaller fleets, France, Israel, Germany, the UK, India, Italy, Japan, and South Korea are known to operate their own optical and radar-imaging satellites. Although such information is generally classified, Brazil is reportedly operating a radar satellite through a private company capable of penetrating the jungle canopy, while Indonesia and South Africa are believed to have similar programs, although these can presumably be described as nonoperational.

Earth observation from space relies on a range of remote sensing techniques, including electro-optical (visual), infrared (thermal), and synthetic aperture radar (SAR). Section 1 in the box on Technical Background provides detailed information on each sensing type and the current state of the art. Taken individually, each sensing modality is subject to distinct physical and technical limitations. However, these limitations can both be overcome through and provide a rationale for integrating multiple sensing domains into a unified analytical framework, yielding spectacular new results.

From Individual Images to Space Data Integration

As multiorbital constellations expand in scale and revisit frequency, space-based sensing is becoming increasingly central to the future of military air and multidomain operations.⁴ This transformation carries a paradox familiar to network-centric warfare theory:⁵ the integration of distributed sensing architectures enhances situational awareness and combat effectiveness, while simultaneously increasing dependence on a functioning and survivable satellite network.

The role of satellite technologies in cross-platform collaboration and detection⁶—particularly in enabling real-time monitoring and targeting, as well as improving anti-stealth capabilities—has increased significantly over the past few years. China’s recent investments in satellite constellations and autonomous sensing systems illustrate this broader shift. Systems such as the Ludi Tance No. 1 (LT-1) L-band synthetic-aperture radar satellite constitute enabling infrastructure: high-resolution optical imaging, combined with all-weather radar observation and increasing revisit rates, is contributing to unprecedented, persistent monitoring capabilities.

The analytical significance of modern satellite networks is vividly illustrated by a striking 2022 report: imagery from the Chinese Jilin-1 constellation appeared to capture a U.S. F-22 Raptor in flight. While tracking a stealth fighter from orbit is not yet a total breakthrough in stealth detection per se, it signals a massive shift in capability. This success is driven by a new “enabling infrastructure”—comprising systems such as the aforementioned Jilin-1 optical constellation and LT-1 radar satellites. By integrating high-resolution optical data with all-weather radar and rapid revisit rates, these shared architectures are turning sporadic glimpses into persistent, real-time monitoring. The decisive step is data fusion: by combining optical, radar, infrared, and radio frequency monitoring systems, this transforms detection from a platform-centric problem into a cross-platform correlation problem. Seemingly unrelated events can be integrated into a broader analytical picture, in which their combined pattern may be significant even though none of the individual events, considered in isolation, may be. Weak signals, environmental disturbances, thermal signatures, airfield activity patterns, and radio frequency (RF) anomalies may individually fall below actionable thresholds. When aggregated and analyzed across domains, however, they reduce ambiguity. Aircraft may minimize radar cross section in specific frequency bands and geometries, but they cannot eliminate all physical interactions with the electromagnetic and thermal environment. Engine exhaust produces infrared signatures; airframes perturb ambient radio frequency fields; and movement generates detectable pattern-of-life deviations. Much like assembling a jigsaw puzzle, putting the individual pieces—in this case, data from different satellites—together reveals the whole picture.

As orbital sensing layers become persistent and digitally integrated, these cross-domain perturbations become searchable. A complementary example of multidomain exploitation is provided by research on passive radio frequency monitoring. Chinese studies⁷ have explored how signals emitted by low-earth orbit (LEO) communication satellites, such as Starlink, could be monitored to detect aircraft or uav26 through subtle perturbations in the ambient electromagnetic environment. This principle—that weak, noncooperative signals can measurably interact with airborne platforms—is supported by foundational work on the effects of solar radio bursts on aviation frequency bands,⁸ as well as subsequent studies examining space weather impacts on drones⁹ and applied research into RF interference mitigation in UAV communication systems.¹⁰ These findings demonstrate that the RF spectrum is not a blank slate: it contains intrinsic weaknesses and ambient signals that cannot be eliminated. When combined with optical, radar, and other sensing modalities within a persistent, multiorbital informatics framework, these spectral observables create exploitable opportunities for detecting low-observable platforms.

The more extensive the sensing network, and the more sophisticated the analytic layer, the fewer environmental variables remain exploitable for undetected operation. While these capabilities enhance monitoring and treaty verification, they also raise questions about where legitimate verification ends and intrusive surveillance begins. Arms control arrangements, such as the Open Skies Treaty, have addressed this tension by establishing mutually agreed limits on sensing capabilities, recognizing that enduring verification depends not only on technical effectiveness but also on political acceptance of the scope of observation.

Enhanced Verification, Emerging Vulnerabilities: Rethinking Control in Data-Driven Conflict

From the perspective of arms control and verification, more accurate and more reliable information has always been regarded as positive. Sensing technologies have proliferated significantly, individually enabling more effective monitoring and verification. Data fusion has facilitated significant advances in verification and reconnaissance, particularly in monitoring movements of stealthy military assets. Based on this technology, states could, for example, verify that certain strategic assets such as stealth bombers have not been deployed near their border but remain at their distant home bases. But with the major bilateral nuclear disarmament treaties dead, satellite-based monitoring continues only informally through national means—without any treaty framework to mandate or structure it, thereby shifting the fine line between verification and spying more toward the latter than is (grudgingly) accepted.

States have consistently sought to keep certain secrets concealed, leaving their adversaries uncertain about the accuracy of their intelligence or whether they might still have an ace up their sleeve. With the advances in data-fusion technology, we are heading toward a state of “inevitable transparency”—with all its pros and cons. In addition, it might not only be opponents who gain a better understanding of security-relevant information; civilian actors, too, may be able to uncover sensitive knowledge and make it publicly accessible. This presents a challenge for remote sensing policy. Where comparable imagery can be obtained from international commercial providers, restricting domestic operators alone may do little to prevent access to sensitive information while imposing costs on legitimate civilian users. This raises the question: How will states react to this development?

Given that actors might balk at openly attacking both state-run and civilian space assets, resilience at the hardware layer can therefore generate vulnerabilities at the electromagnetic and informational layers. The focus of attacks would thus shift to data support functions, including ground stations or even AI computer centers responsible for the fusion of the data.

As a consequence, strategic competition is likely to shift away from kinetic anti-satellite engagement toward cyber and electronic means of interference.¹¹ Given that these activities are difficult to observe and attribute, they fall outside many of the assumptions underpinning traditional arms control verification. This suggests that future discussions may need to consider issues such as the resilience of verification systems, protection against interference with critical data flows, and measures to reduce the risk of false or escalatory warnings. However, states should be aware of the problems they might face in the not-so-distant future.

As verification increasingly relies on AI-assisted fusion of data from multiple space-based and terrestrial sensors, the integrity of the underlying data becomes progressively more important. One possible confidence-building measure would be a politically binding commitment to authenticate data entering agreed verification pipelines—for example, through digital signatures or comparable cryptographic mechanisms—to reduce the risk that manipulated or fabricated data could contribute to false assessments or unnecessary escalation. This would also involve civilian actors providing publicly available data. The extent to which states could or should regulate the public sale of high-resolution imagery is open to debate—traditionally, shutter control policies have had to balance state security requirements with the need to foster a viable commercial market for imagery products. The recent case of Planet Labs withholding commercial imagery of the Middle East sets a new precedent, while also presenting policymakers with a fresh challenge. As Kelevitch¹² argues, the fact that shutter control can work effectively enough to restrict access to commercially available imagery presents a concern. The international reach of shutter control to restrict civilian monitoring during conflict became evident when U.S. authorities submitted a diplomatic démarche to the EU on July 13, which subsequently resulted in a delay in the provision of Copernicus imagery under the EU Space Threat Response Architecture.

These developments raise questions about whether existing national frameworks remain adequate for an increasingly international commercial Earth observation market. In this context, Germany should reassess whether its National Data Security Policy for Space-Based Earth Remote Sensing Systems remains fit for purpose, particularly in balancing national security objectives with the scientific, commercial, and humanitarian benefits of commercial Earth observation. More broadly, shutter control policy should be revisited to ensure that measures are in place to preserve its legitimate security function while preventing its overapplication from inadvertently undermining the ability of the international community to independently monitor conflicts occurring on allied territory. Kelevitch recommends mandatory notification of allied governments whose territories fall within restricted imaging areas, alongside a protocol requiring states to specify the geographic scope and duration of an image hold, allowing affected states to identify alternative sources of imagery where necessary.

In sum, as in other domains of verification, political will here is also significantly lagging behind technological possibilities, particularly when combining the latest sensor technology with AI. While these techniques could prove valuable for future arms control agreements, adversaries might perceive them as intrusive surveillance, potentially increasing the risk of attacks on satellite infrastructure or even on satellites themselves. So, what are the political consequences of these latest developments?

When weighing the risks against the opportunities, the prospect of gaining significantly enhanced intelligence clearly outweighs the risks, particularly for emerging space powers such as Germany. States should therefore invest in AI-enabled intelligence. Rather than relying on collecting imagery alone, the strategic advantage increasingly depends on the integration and analysis of multisensor data. Policymakers should thus prioritize AI, data fusion, and analytical capacity. States should, however, bear in mind that non-state actors will also have increased AI capabilities. In light of this, it is important to maintain rigorous technology assessment. Governments should systematically evaluate claims of novel sensing technologies, allocating resources to research capable of distinguishing credible advances from strategic signaling, thereby avoiding technology races driven by the fear of being exposed.