On the night of January 20, 2018, an amateur radio operator and electrical engineer named Scott Tilley sat in his backyard observatory in Roberts Creek, British Columbia, sweeping a software-defined radio across the S-band looking for a classified American spy satellite called Zuma. He never found Zuma. Instead, his receiver locked onto a steady carrier wave at 2275.905 MHz — a signal NASA had not heard in more than twelve years, broadcast by a spacecraft the agency had officially given up on in 2005. The transmitter was IMAGE, a magnetospheric observatory the size of a kitchen island, and it was still spinning, still drawing power from its solar panels, and still trying to phone home.
Tilley posted his findings to his blog the following day, describing how his hunt for the Zuma satellite had led to the discovery of the long-lost NASA spacecraft. Within seventy-two hours, NASA had reactivated the deep-space communications hardware needed to talk to a spacecraft nobody on the active operations roster had ever spoken to. The story that followed is one of the strangest resurrections in the history of orbital engineering — a piece of 1990s aerospace technology that refused to die quietly, found by a hobbyist looking for something else entirely.
A satellite built to watch the invisible
IMAGE, short for Imager for Magnetopause-to-Aurora Global Exploration, launched on a Delta II rocket from Vandenberg Air Force Base on March 25, 2000. Its mission was to photograph the Earth’s magnetosphere — the invisible bubble of charged particles wrapped around the planet by its magnetic field — using a suite of instruments tuned to ultraviolet, neutral-atom, and radio-plasma emissions. Before IMAGE, the magnetosphere had only ever been sampled point-by-point, the way you might try to map a thunderstorm by walking through it with a thermometer. IMAGE produced the first global pictures of it, including the first images of the plasmasphere from outside.
The spacecraft was built by Lockheed Martin Missiles and Space and placed in a highly elliptical polar orbit that swung from roughly 1,000 kilometres above the planet out to about 45,900 kilometres at apogee. From that vantage, it watched aurorae bloom over both poles simultaneously, mapped ring currents during geomagnetic storms, and gave researchers a moving picture of how solar wind dumps energy into the near-Earth environment. The original mission was scheduled for two years. NASA extended it. IMAGE kept working.
December 18, 2005: the silence
Then, without warning, it stopped. On December 18, 2005, IMAGE failed to check in during a routine pass. Engineers at Goddard Space Flight Center traced the most likely cause to a single-event upset in the spacecraft’s power controller — a stray cosmic ray, probably, that flipped a transistor and tripped the system into a state from which it could not recover on its own. The standard recovery procedure required the spacecraft to fall into eclipse, drain its battery completely, and reboot when sunlight returned. That eclipse season was projected for late 2007.
NASA waited. The eclipse came and went. Nothing. A failure review board closed the mission. The hardware was written off, the telemetry archives sealed, and the engineers who built the ground systems moved on to other projects. IMAGE, presumed inert, kept tracing its long ellipse around the Earth, drifting through more than twelve years of solar maxima and minima with no audience.
The amateur who looks for secrets
Scott Tilley is part of a small global community of radio amateurs who track classified satellites for sport. He runs antennas and software-defined receivers from his home on the British Columbia coast and publishes his findings openly. In January 2018, SpaceX launched the Zuma payload for the U.S. government, a mission so secret that even its sponsoring agency was never publicly identified. Tilley wanted to know whether Zuma had reached orbit — early reports suggested it had been lost — and he started sweeping likely frequencies for any carrier that fit the profile.
What he caught instead was a clean, narrow signal with a Doppler shift that matched an object in a highly elliptical orbit. He ran the orbital elements through a public tool called STRF, written by astronomer Cees Bassa, which fits orbits to radio signatures. The match came back as 2000-017A. IMAGE. The dormant satellite. Tilley double-checked, then triple-checked. The carrier was modulated. The spacecraft was not just reflecting sunlight — it was actively transmitting.
Waking the ground system
Within a day, Tilley had contacted Richard J. Burley, who had been IMAGE’s mission director at Goddard before the spacecraft went silent. Burley spread the word internally, and a small team began the work of figuring out whether any of the ground hardware needed to decode IMAGE telemetry still existed. Some of it did, in storage. Some had been cannibalised for other missions. The signal processing software ran on operating systems that had been retired years earlier and had to be reverse-engineered by engineers at the Johns Hopkins University Applied Physics Laboratory, who pointed their 18-metre satellite dish at the spacecraft’s predicted location and worked their way bit by bit to a baud rate of exactly 44.44 kilobits per second.
On January 30, 2018, NASA confirmed that the recovered signal was IMAGE. Goddard had coordinated five separate antennas — at Goddard itself, the APL facility in Maryland, NASA’s Wallops Flight Facility in Virginia, the White Sands Test Facility in New Mexico, and a ground station at UC Berkeley — to acquire the carrier and extract enough housekeeping data to identify the spacecraft conclusively. IMAGE was spin-stabilised, its solar panels were generating power, and its main transmitter was healthy. Somewhere in the intervening years, an eclipse season had apparently done exactly what the original recovery plan predicted — drained the battery, forced a reset, and brought the power controller back online. This is the same kind of patient, late-life ground work that still keeps the Voyager spacecraft alive with code written before most of its current operators were born.
Why nobody was listening
The painful part is that IMAGE may have been transmitting for years before Tilley caught it. NASA had stopped pointing antennas at its frequency in 2005, and the spacecraft’s signal is faint enough that no general-purpose tracking system would flag it automatically. A ground station has to be tuned to the right narrow band, pointed in the right direction, at the right moment in the satellite’s roughly fourteen-hour orbit. Without a reason to look, nobody looked.
This is a known hazard of how the agency catalogues failed missions. Once a spacecraft is declared lost, its frequencies are released back into the pool, its ground software is decommissioned, and the institutional memory of how to talk to it leaks out of the building one retirement at a time. The hardware in orbit, meanwhile, does not know it has been abandoned. It just keeps running its boot loop.
The reactivation that almost happened
Through the spring of 2018, NASA engineers worked to re-establish two-way contact. They confirmed the spacecraft’s spin, attitude, and thermal state from the carrier signal alone. They began rebuilding the command software that would let them send instructions back up. Patricia Reiff, a co-investigator on the original IMAGE mission at Rice University, told Spaceweather that the odds of recovery looked extremely good.
Then, on February 22, 2018, the signal began to break up, and by February 24 it had gone silent. It returned briefly in May and again sporadically through the summer, but the team was never able to lock on long enough to send commands the spacecraft would acknowledge. The most likely explanation was that whatever fault had caused the original 2005 failure had recurred, or that the spacecraft was cycling through brownout states as its aging battery struggled to hold charge through eclipses. The full scientific revival never happened.
What IMAGE left behind
The data IMAGE collected between 2000 and 2005 is still being mined. Its plasmaspheric imagery rewrote textbook descriptions of how the Earth’s magnetic environment behaves during storms, and its measurements of the auroral oval are still cited in papers about space-weather forecasting. The brief 2018 reactivation also produced something less tangible: a worked example of how cheap, modern radio equipment in private hands can find things government-scale infrastructure has stopped looking for.
Tilley has since identified other lost or unannounced satellites from his backyard, including the experimental UHF military communications satellite LES-5, which was launched in 1967 and was supposed to shut down in 1972. A loose network of amateur trackers around the world now shares Doppler curves and frequency catches the way birders share rare sightings. The same hobby that found IMAGE has caught Cold War-era surveillance birds still broadcasting, abandoned weather satellites, and the occasional secret military payload whose operators would prefer it stay unseen.
Old satellites, new economics
IMAGE’s accidental afterlife arrives at a strange moment for the satellite business. Companies are now building spacecraft designed to be serviced, refueled, and rebooted on purpose — DARPA’s robotic deep-space repair satellite is scheduled to launch in 2026, and operators like Vast are selling high-power satellite buses built around long operational lives. None of these designs assume that a single-event upset means the end of the spacecraft.
IMAGE belongs to an earlier philosophy: launch it, operate it, lose it, move on. Its survival was an accident of orbital mechanics — the spacecraft just happened to be in a configuration where eclipses could power-cycle it back into life. The next generation of satellites will be built for the kind of recovery that IMAGE managed on its own, by chance, while nobody was watching. The same kind of long-duration thinking now anchors missions like Voyager 1, which still answers commands sent across more than 22 hours of light-speed delay.
Still up there
The spacecraft is still in orbit. Its ellipse has shifted slightly over the years from solar pressure and the subtle tug of the Moon, but it traces roughly the same path it did when it was working. Every few months, an amateur tracker checks the old frequency to see whether the carrier has come back. Sometimes there is nothing. Sometimes there is a flicker — a few seconds of carrier, then silence — and the trackers post their waterfall plots online for anyone who wants to look.
Somewhere above the Pacific tonight, eighteen thousand kilometres up, a kitchen-island-sized spacecraft built during the Clinton administration is turning at half a revolution per minute in the dark, its gold foil weathered by twenty-six years of micrometeorites, its solar panels still catching enough sunlight to keep the lights on. It is doing what it was built to do. The only thing that has changed is whether anyone happens to be listening.
