For the last several years, I’ve been undertaking the most technically challenging photography series I’ve ever attempted: a project to photograph objects of unknown origin in orbit around the earth.

There are roughly 350 objects in orbit around the earth whose origins are unknown. These fall roughly into two categories: 1) Objects that the US Air Force tracks on radar and publishes orbital data for; 2) Objects that both amateur astronomers and foreign sources track and observe, but that the US military does not acknowledge, presumably because these unknown objects are classified.

UNKNOWN #89161 (Unclassified object near The Revenant of the Swan), 2023
Silver gelatin LE print
80 × 54 in.

The term “unid” is a term that amateur astronomers created to describe objects that they have observed in orbit, but whose identity they have failed to establish. In the first part of this text, I provide an overview of what we know about these objects, and review some attempts to identify them and their purpose. In the second part, I’ll describe some of the techniques I’ve used in my attempts to photograph them.


What are Unids?

The short answer is, nobody knows. The longer answer is that for some objects, somebody probably knows something about some of them, but they’re not saying. Or, that also might be wrong and actually nobody knows.

Some background: The US Space Force’s 18th Space Defense Squadron, located at Vandenberg Space Force Base1 on the California coast north of Santa Barbara, is tasked with operating the US’ Space Surveillance Network. This is a global network of powerful radar systems, classified telescopes, space-based surveillance platforms, and other sensor networks. The squadron’s job is to identify and keep tabs on tens of thousands of objects in orbit around the earth. Over the course of their work, they regularly track and observe nearly 350 objects whose origin and identity are unknown. The 18th SDS catalogs these as “well tracked analyst objects.”

The “well-tracked analyst objects” are described by the surveillance squadron as “on-orbit objects that are consistently tracked by the U.S. Space Surveillance Network that cannot be associated with a specific launch. These objects of unknown origin are not entered into the satellite catalog, but are maintained using satellite numbers between 80000 and 89999.” (In the military satellite catalog, satellites are cataloged sequentially, i.e. the rocket that launched Sputnik is catalog entry #1, Sputnik is entry #2, etc.)

Orbital Elements for Unknown Objects (Paglen Studio)

So what are these objects? The best answer is that, well, nobody knows. A more fine-grained answer involves some informed speculation. It is unlikely, however possible, that some of these objects are natural phenomena such as wayward asteroids. Undoubtedly, most of these “unknowns” are unidentified debris from satellite launches in places or times where the Air Force’s tracking capabilities are limited. But the story is almost certainly far more complicated.

The US’ National Reconnaissance Office (NRO) has a history of building satellites that attempt to disguise themselves as pieces of debris. This was the case for example with a spacecraft called “USA 53” (deployed from the Space Shuttle in 1990) that faked its own explosion, and again in 1999 when another “stealth” satellite deployed a balloon-like structure as a decoy. The Russian military has engaged in similar tactics, most recently with a spacecraft called Kosmos 2499, which behaved as if it were a debris object but which was almost certainly a satellite designed to attack other satellites. (Kosmos 2499 was mysteriously destroyed in early 2023, creating a small debris field.)

Visualization of Unknown Object Orbits (Paglen Studio)

Analyzing Unids
The only publicly available analysis of the “well-tracked analyst objects” that I’m aware of comes from a PhD dissertation written by space-security researcher James Pavur at the University of Oxford. Pavur took a novel approach to the analysis of these objects. He created a dataset of known satellites, and another dataset of known debris objects, and then trained a machine learning model on each. His idea was to build a classifier that could distinguish between a “generic satellite” and “generic debris object.” Pavur then used those models to analyze the orbit of Kosmos 2499, a satellite that pretends to be a debris object. His model correctly predicted that Kosmos 2499 was a satellite, not a piece of debris. Pavur then ran his model on the entirety of the “well-tracked analyst objects” data and discovered something remarkable: the model predicted with high confidence that a non-trivial number of unknown objects behaved, in fact, like spacecraft.

Excerpt from James Pavur’s dissertation “Securing new space: on satellite cyber-security” (2021)

Almost immediately after the publication of his dissertation, Pavur was tapped to work for the Department of Defense and is unable to speak about his current work. However, Pavur did provide me with a copy of the models he used in his analysis and I’m conducting a review of them to see if there’s more to learn about “analyst objects” from his work.

There are, however, a few limitations to Pavur’s approach. Firstly, Pavur’s classifier wasn’t designed to detect station-keeping maneuvers. Operational satellites in low-earth orbit are affected by small amounts of atmospheric drag in the upper atmosphere that slowly bring them back down to earth. To counter this, a satellite has to periodically “boost” itself back into its desired orbit using small thrusters located on the spacecraft. Satellites in higher orbits are affected by the gravitational influence of the moon, and from the uneven nature of Earth’s gravity field.2

The Plot Thickens…
In addition to objects in the 18th Space Defense Squadron’s publically available data, there are two additional sources of information about unknown objects. The first is a hybrid Russian civilian/military tracking program called “ISON” (International Scientific Observer Network), and the second is a database of classified objects maintained by a network of amateur satellite observers, unofficially known as the “See-Sat” group. Both of these groups have identified a handful of unknown objects in orbit whose existence is classified by the American military – in other words Top-Secret unknown objects.

Classified Unknown Objects in Earth Orbit (Paglen Studio)

So, to recap: There are many hundreds of unknown objects in orbit around the earth, many of which are tracked and acknowledged by the US military. Researchers who’ve analyzed these objects have concluded that a non-trivial number of them display characteristics more consistent with spacecraft than debris objects, although these results require further study. What’s more, there are more than a dozen other objects that are also “unknowns” but whose existence is classified and whose orbits are undisclosed.


Production documentation (Paglen Studio)

Photographing Unids
Photographing these objects is extremely difficult in every way, but can be done using good data, accurate modeling, and very specific optical equipment.

Step 1: Get the Data
The first thing one needs to photograph unids is a good source of data. I use two sources: two-line elements (a file format for describing satellite orbits) for “well-tracked analyst objects” are readily available by creating an account with “The Space Force,” on their portal for satellite information at space-track.org. This database provides a list of unclassified data. To retrieve data about classified unknown objects, the best source is a website maintained by satellite observer Mike McCants, who coallates observations from amateur satellite observers and publishes orbital elements based on those observations. Those elements need to be downloaded and filtered for both “unknown” objects and “ISON” objects.

Research documentation (Paglen Studio)

Step 2: Model the Orbits
Then I import that data into two different virtual planetarium software environments. (I use two in order to ensure that my predictions are accurate across multiple models and that I haven’t made a mistake). The first software I use is Stellarium (this is a superb piece of free astronomy software). To check my work, I load the same data into a second modeling program called Heavensat.

Using the modeling software, I can make predictions about when and where in the sky I might find a particular object.

It takes the better part of the day to model these orbits, and to select a series of targets for a given evening. Once I’ve selected the objects I want to image, I write a script for the evening in a software package I use to control the telescope, mount, and camera. The script tells the telescope to point to a particular point in the sky at a very precise moment, then instructs the camera to start making exposures before, during, and after the predicted pass of the unknown object. If I do everything correctly, I am able to capture the light-trail of the object as it passes through the telescope’s field-of-view.

Step 3: Equipment
The main difficulty in choosing an appropriate telescope for photographing unids is sourcing a telescope that can collect as much light as quickly as possible. Because unids tend to be both very faint and fast-moving, I use the “fastest” telescope that I can. In my case, that means a Rowe-Ackermann Schmidt Astrograph (RASA) astrograph.

The RASA design is designed above all for speed, but it sacrifices ease and multifunctionality to get there. The design is a variation on a Schmidt-Cassegrain Telescope that replaces the secondary mirror with a camera sensor. The advantage of this is that the telescope can collect far more light much faster than a telescope with a secondary mirror. There are, however, many disadvantages. First, the removal of the secondary mirror means that there is no possibility of including an eyepiece in the design – there is no way to look through the telescope and it can only be used with a specialized camera. Secondly, at f/2 the critical focus zone (known as “depth of field” in ‘normal’ photography) of the telescope is smaller than half the width of a human hair. This translates into extreme technical difficulties in positioning the camera sensor accurately, as slight imperfections in how the camera sensor is placed in its housing during the manufacturing process create optical anomalies that have to be manually compensated for. This process is not fun.

On any given night, I’m aiming for triple-redundancy: for each image I am using three separate telescopes to collect as much light as possible and to mitigate against any mechanical failures (which happen often).

When I’ve successfully photographed the light-trail of a unid, I task the telescopes with collecting additional data from that region in the sky to fill out the photograph. Each exposure ends up being about 10,000 seconds worth of data or about 3 hours, much of which is shot with an infrared filter to highlight the various stelliferous and gaseous regions in the sky that are invisible to our eyes.

Gustav Dore, The stormy blast of hell with restless fury drives the spirits on, c1890.

The night sky looks very different to infrared-sensitive equipment than it does to unaided eyes. Hydrogen, sulfur, and oxygen emissions reveal great cosmic clouds, stellar remnants, and galactic structures that recall Gustav Dore’s etchings of the Divine Comedy. Their names refer to ancient myths, stories, and ancestral star-gazers. Many of the stars in the sky have names so ancient that their origins of those names, and the stories they once referred to, have been long forgotten.

I have spent countless days and nights studying the unknown objects, plotting orbits, measuring light curves, and analyzing their movements over time to see how their behavior may have changed over the years. I’ve tried to learn anything and everything I can about their shape, size, and mass, the relative stability of their orbits, and the question of whether they receive energy from any non-natural sources. Some of the numbers are surprising.

But every analytical technique available supplies only tiny variations on a simple fact: the identity of these objects is “unknown.” Given this, I ask myself where my desire to “identify” them comes from. Where does my unconscious desire to place these objects into received categories come from? Why does my subconscious seek the comfort of pre-existing language and concepts in the face of these unknowns?

This post is part of a series of posts supporting works from the exhibition You’ve Just Been Fucked by PSYOPS at Pace Gallery, New York.

You can find more information about this project and other posts in the series here.


1 I’m sorry I really have a hard time saying or writing the words “Space Force.” (Back)

2 Gravitational anomalies are areas on Earth where the local gravity field is stronger or weaker than the global average. These variations can be caused by differences in the density of the Earth’s crust, the presence of large mountain ranges or ocean trenches, and variations in the distribution of the Earth’s mass. One of the most well-known gravity anomalies is the “Indian Ocean Geoid Low,” which is a region of low gravity field strength in the Indian Ocean. This anomaly is primarily due to the large mass of the Himalayas to the north and the Earth’s equatorial bulge. Gravitational anomalies affect satellites by subtly altering their inclination over time, and by causing them to drift longitudinally. (Back)