Who Watches the Sky? What Radar and Drone Reports Can Actually Tell Us

Point Barrow long-range radar site, Alaska, 4 February 2023. US Air Force photo by Tech. Sgt. Curt Beach / DVIDS. The image does not depict the incidents discussed.

On a September night in 2025, reports of drones stopped air traffic at Copenhagen Airport. Nine months later, Danish police said they could neither establish nor rule out drone activity in the Copenhagen investigation. The disruption was real. The identity of what prompted it was not established. ^{[5] [10]}

That distinction sits at the heart of a wider problem. A radar return, a light in the sky and a report to police are different kinds of evidence. Each passes through choices about what to notice, what to investigate and when to act. Those choices can protect people before a threat is confirmed; they can also make an uncertain event look more definite than it is.

This investigation examines three separate episodes: NORAD’s response to a high-altitude Chinese balloon in 2023, reported drones in New Jersey in 2024, and the Copenhagen airport closure in 2025. It also examines US counter-drone powers in 2026. It does not claim the episodes were connected or identify a common operator.

In early 2023, a Chinese high-altitude balloon crossed North America and was shot down by the US. Canadian defence officials say NORAD detected and tracked it. The documented tracking matters: it would be wrong to say this balloon itself was filtered out. ^[2]

After that detection, NORAD revisited its radar settings. On 17 February, its Canadian deputy commander, Lieutenant-General Alain Pelletier, told Parliament that the command adjusted a filter, also called a gate, to detect objects moving more slowly and returning a smaller radar signal. He said the adjustment probably helped detect later objects, while increasing the risk of false contacts. ^[1]

A radar return is a measurement, not a photograph of a named aircraft. A setting that admits more weak or slow contacts can give operators more to examine; it can also produce more noise. In later US congressional testimony, NORAD commander General Gregory Guillot likewise described greater sensitivity and additional clutter. These are the commanders’ explanations of their systems, not independently published raw radar data. ^[3]

The adjustment gives a plausible reason why more objects might be detected without proving that more objects had suddenly appeared. It also does not prove what any particular return was, or that every previously missed object was invisible to every sensor.

Radar and a public aircraft-tracking display are not the same thing. The US Federal Aviation Administration says ADS-B Out broadcasts an aircraft’s position, speed and other information; its equipment requirement applies in specified airspace. A display relying on these transmissions cannot by itself show every object in the sky. ^[6]

That limitation cuts both ways. An aircraft missing from an app is not automatically secret, hostile or absent from military radar. Equally, an aircraft visible on an app does not explain a separate nearby sighting without a time-and-position match. Drone identification rules and radar coverage involve additional systems and cannot be inferred from ADS-B alone. A 2026 Government Accountability Office report describes cameras, acoustic sensors and ground radar as ways to detect aircraft that are not broadcasting. ^[16]

Primary radar detects reflected radio waves; secondary radar relies on replies from aircraft transponders. The FAA describes an airport system combining both, while its radar sustainment programme says ground-based radar can detect some aircraft that are not broadcasting ADS-B. Neither system guarantees detection of every object. ^{[11] [13]}

Small drones pose a further detection and classification problem. University of St Andrews researchers explain that small radar cross-sections, slow movement and low altitude can make them difficult targets; birds can resemble drones in size, speed and altitude. They study micro-Doppler signatures from movement such as propellers or wings to help distinguish targets. A promising technique is not proof that a particular system detected or identified an object in any of these cases. ^[18]

Surveillance data can cross institutional boundaries. The FAA says its Mode S system and co-located primary radars can provide correlated radar and beacon reports to air traffic facilities, the US Department of Defense and other users. That describes a capability, not a complete current map of who received which track in any incident. ^[12]

The US All-domain Anomaly Resolution Office reported in 2024 that lack of timely, actionable sensor data constrained its ability to resolve cases. It also described work on prototype processing for FAA and weather radar data to examine returns currently filtered out. This describes a research effort at that time, not a complete assessment of today’s sensors or the three incidents discussed here. ^[14]

By 17 December 2024, federal agencies said the FBI had received tips concerning more than 5,000 reported drone sightings in recent weeks, generating approximately 100 leads. The agencies said they had sent detection technology and trained observers to help investigate. ^[4]

Their joint assessment was that the sightings to date included lawful commercial, hobbyist and law-enforcement drones, as well as manned aircraft, helicopters and stars mistaken for drones. They said they had not identified anomalous activity or assessed a national-security or public-safety risk over the civilian airspace they described. ^[4]

The same statement separately acknowledged a limited number of visual drone sightings over military facilities, including restricted airspace. ^[4] In January 2025, the White House said many drones had been authorised for research and other reasons; a subsequently introduced House resolution records that statement. Neither account matches individual public reports to particular flights. ^[17]

That is an institutional assessment at a stated point in time. It is not a published case-by-case explanation of all 5,000 reports, nor does the count of leads mean 100 hostile drones. Many tips may refer to the same thing, and the statement does not provide a denominator of individually resolved sightings. The absence of an identified threat in the agencies’ assessment should not be converted into proof that every witness saw the same benign object.

Danish authorities halted air traffic at Copenhagen Airport after reports of drone observations on 22 September 2025. At the time, the reports generated concern about critical infrastructure. A closure is a precautionary decision: its purpose is to prevent a possible collision, not to certify the identity of the object first. ^[5]

In June 2026, Danish national police said the Copenhagen investigation had concluded. Their account says that in a small number of investigations, including Copenhagen and North Jutland, drone activity could neither be demonstrated nor ruled out. For most investigations elsewhere, police said drones could be ruled out. These are separate conclusions, not a finding that nothing was in the sky anywhere in Denmark. ^[5]

The earlier description of a capable operator and speculation about a foreign actor should therefore not be presented as an established explanation of the Copenhagen closure. Police did not establish a drone in that case; the available conclusion also does not prove there was no drone. A corroborated recording, sensor track or physical evidence could narrow the uncertainty, but the public record cited here does not provide it.

Munich offers a separate example of precautionary action: Bavarian and federal police reported multiple sightings around the airport on 2 October 2025, after which flight operations were suspended. Their initial statement is evidence of reports and disruption, not a finding that Copenhagen involved the same operator. ^[15]

Detection and legal authority are separate questions. In December 2025 Congress enacted the SAFER SKIES provisions in the fiscal 2026 defence law. The provisions expand specified counter-drone powers to qualifying state, local, tribal and territorial law-enforcement and correctional agencies, subject to conditions. They do not give every officer a general permission to interfere with any aircraft. ^[7]

An interim final rule issued by the Justice and Homeland Security departments on 6 July 2026 sets a framework for training and certification, authorised technologies, airspace coordination, reporting and privacy. The rule states an effective date of 1 July. It establishes separate levels for detection and warning, and for mitigation. Some passive detection that does not need the statutory powers is outside that training requirement; the rule does not treat every camera or radar as a regulated counter-drone operation. ^[8] The FBI has described a national training centre; that statement alone does not establish which local agencies have completed certification or deployed equipment. ^{[8] [9]}

The practical tension is clear: a potentially dangerous aircraft may need a quick response, but disabling or interfering with an airborne device can carry its own risks. Rules for approval, coordination and reporting are part of the response. Their existence is not evidence that they have been applied correctly in a particular incident.

These cases do not demonstrate a single hidden network. NORAD’s adjustment concerns a military radar process; the New Jersey statement concerns a collection of public tips and investigations; Copenhagen concerns reports and an airport safety decision. Treating them as one operation would go beyond the evidence.

There is also a stronger ordinary explanation than a coordinated mystery: busy airspace contains lawful aircraft and drones, perception at night is difficult, and better attention can produce more reports. That does not eliminate every real threat. It does mean that unexplained does not automatically mean extraordinary or hostile.

The public sources reviewed here do not establish the complete current sharing arrangements or coverage of every US civilian and military radar, identify every object reported in New Jersey, or identify what prompted each Copenhagen observation.

The sky is watched through overlapping systems, none of which turns every report into a complete identification. NORAD’s own testimony documents a trade-off: a more inclusive radar gate can reveal more contacts while adding false ones. The New Jersey agencies’ assessment and Danish police’s later investigation show different limits on what can be concluded from reports. ^{[1] [4] [5]}

The strongest supported answer is not that authorities ignore a known fleet, or that every mystery has been solved. It is that detection, identification, threat assessment and authority to intervene are distinct steps. A closure can be justified even when the object is never established; a large collection of reports can contain ordinary aircraft without proving the answer to every report. Better case-level evidence, not a more dramatic label, is what would change those conclusions.

Method: this article compares official parliamentary testimony, agency statements, police findings, FAA guidance, enacted US law and the implementing rule available by 4 November 2026. The agencies and commanders describe their own decisions and capabilities; their statements are evidence of those assessments, not independent access to all underlying sensor records. Danish police findings are read in Danish and checked against a contemporaneous Danish news report.