Rubin Observatory’s real-time sky alerts debut with 800,000 notifications in a night

High on a wind-swept peak in the Chilean Andes, a car-size camera inside a new observatory swung from one patch of sky to the next on a recent February night. Every 40 seconds, its shutter clicked, capturing a 3.2‑gigapixel image so detailed it would take hundreds of 4K television screens to display at full resolution.

By dawn on Feb. 25, those snapshots had been transformed into something astronomers have waited more than a decade to see: about 800,000 automated alerts, each one a digital flag that something in the universe had changed.

A milestone ahead of LSST

The National Science Foundation–Department of Energy Vera C. Rubin Observatory announced on Feb. 25 that, during the night of Feb. 24 local time in Chile, it had carried out the first large-scale public test of its real-time alert system. The exercise marks one of the last major milestones before the observatory begins its 10‑year Legacy Survey of Space and Time (LSST) later this year.

The test did more than show that a new telescope can take sharp pictures. It demonstrated that Rubin can detect and broadcast changes in the night sky within about two minutes of capturing an image, and that it can do so on a scale that will eventually reach up to roughly 7 million alerts every clear night.

“This is the launch of a real-time discovery machine for monitoring the night sky,” the observatory said in its statement announcing the test.

What an “alert” is—and what it can signal

Each alert corresponds to something that has appeared, brightened, faded or moved since the last time Rubin imaged that region. That could be an exploding star in a distant galaxy, a variable star within the Milky Way, a flare from matter falling into a supermassive black hole, or an asteroid shifting against the background of stars.

In practice, an alert is a compact data package, not a siren. It includes:

• a small cutout of the new image,
• a matching cutout from a deep “template” image taken on previous nights, and
• a third image showing the difference between the two.

Basic measurements of brightness, position and how that spot has behaved over time are added before the packet is sent out over the internet.

Rubin’s data systems compare every new exposure with those templates in a process known as difference imaging. When the software finds a significant change, it generates an alert and ships it to a network of subscribers around the world.

“The scale and speed of the alerts are unprecedented,” said Hsin‑Fang Chiang, a software developer at SLAC National Accelerator Laboratory who helps lead operations at Rubin’s U.S. data facility in California.

From a Chilean mountaintop to global subscribers

From the summit of Cerro Pachón, about 2,700 meters above sea level in Chile’s Coquimbo Region, raw images travel over high‑bandwidth fiber links to SLAC, where Rubin’s primary U.S. data center is located. There, automated pipelines analyze the data, compare them with the existing sky templates and create alerts, typically within about two minutes of the shutter closing on the telescope.

The Feb. 24 run was the first time Rubin sent a large volume of those alerts into the open, in near real time, at something approaching operational scale. The observatory and its partners have previously conducted internal tests and smaller public trials, but this night pushed the end‑to‑end system—from telescope to brokers—into territory not far from what LSST will see on a routine basis.

Why Rubin relies on “alert brokers”

The alerts are public by design. Rather than asking individual astronomers to cope with millions of nightly notifications, Rubin relies on an ecosystem of alert brokers—independent services that subscribe to the full stream and then filter, classify and redistribute it.

Among the brokers already active on Rubin data are ALeRCE, a Chilean‑led project focused on machine‑learning classification of transients, and Lasair, a United Kingdom‑based broker that previously handled alert streams from the Zwicky Transient Facility in California. The U.S. ANTARES system and other brokers are also preparing to ingest Rubin alerts.

These services examine each alert, cross‑match it with existing catalogs and apply algorithms to infer what kind of object or event it represents. They allow users to set up filters—for example, “supernova candidates brighter than a certain magnitude” or “fast‑moving objects that might be near‑Earth asteroids”—so that individual researchers receive only the subset relevant to their work.

ALeRCE has already begun processing Rubin alerts in what it describes as an iterative validation phase, refining how it handles the new stream as the observatory tunes its systems.

A survey built for scale

Rubin officials say this broker model is essential if the community is to make sense of the data volume that LSST will produce. When routine operations begin, the observatory expects to scan the accessible Southern sky roughly every three nights in six optical filters, building up a time‑lapse record of the universe over 10 years. The survey is forecast to catalog about 20 billion galaxies, 17 billion stars and millions of small bodies in the solar system.

The project’s four flagship science goals reflect that breadth. LSST is designed to:

1. probe the nature of dark matter and dark energy,
2. build a more complete inventory of asteroids and other solar system objects,
3. map the structure and formation history of the Milky Way, and
4. systematically explore the transient and variable sky.

The alert system sits at the heart of that last goal. Fast notifications are critical for catching events that change on human timescales—minutes to days—rather than over cosmic eons.

For supernovae, for example, astronomers often want to obtain spectra and follow‑up images as soon as possible after the explosion to understand how the star died and what elements it produced. For rare, fast transients such as kilonovae, which are associated with neutron‑star mergers and gravitational‑wave signals, a delay of a few hours can mean missing the peak of the light curve entirely.

In its first public night of alert production, Rubin reported detecting supernovae, variable stars, active galactic nuclei and solar system objects, among other phenomena.

Planetary defense and multi-messenger astronomy

The same system is expected to play a role in planetary defense. Over the course of the survey, Rubin is projected to increase the number of known small solar system bodies by a factor of several, including many near‑Earth asteroids that have not yet been found. Real‑time alerts allow follow‑up observatories to refine orbits quickly and assess whether any newly discovered object poses a potential hazard to Earth.

From its remote Chilean mountaintop, Rubin will not operate in isolation. Its alerts are expected to trigger follow‑up from a global fleet of ground‑based and space‑based observatories. They will be cross‑checked against gravitational‑wave detections from LIGO and Virgo, neutrino alerts from IceCube and gamma‑ray observations from missions such as NASA’s Fermi telescope and the upcoming Cherenkov Telescope Array.

A shift in how astronomy is done

Beyond specific discoveries, Rubin’s approach forces a broader shift in how astronomy is conducted. Instead of astronomers writing proposals for a few nights of observing time on a large telescope, the sky will generate its own observing cues in real time, and software will decide which ones merit follow‑up.

“The largest spot‑the‑difference effort ever has begun,” the observatory wrote in a social media post, echoing a metaphor staff have used to describe the comparison between new and old images.

Bob Blum, a senior leader at Rubin Observatory, told colleagues on an internal community forum that the start of public alerts represents “the beginning of science for the Observatory and its science community” and that Rubin and its partners are “changing how astronomy and astrophysics are done.”

The alerts are openly available, but access to the full scientific opportunity depends on more than just a network connection. Handling millions of events per night requires robust computing resources and sophisticated algorithms, and some researchers have raised questions about how to ensure rare or unexpected phenomena are not missed by machine‑learning systems trained on known types of events.

Rubin’s managers and the broker teams say they are building anomaly‑detection tools and flexible filters to address that concern, and they emphasize that the open nature of the alert stream allows institutions of all sizes, as well as citizen scientists, to participate.

A long-running megaproject nearing full operations

The Vera C. Rubin Observatory is the result of a long‑running U.S. scientific megaproject. Originally proposed as the Large Synoptic Survey Telescope, the facility is funded jointly by the National Science Foundation and the Department of Energy’s Office of Science, with construction formally beginning in 2014. In 2019, Congress passed and the president signed the Vera C. Rubin Observatory Designation Act, renaming the telescope for the late astronomer whose work on galaxy rotation curves provided some of the strongest early evidence for dark matter.

For now, the night of Feb. 24 stands as proof that the observatory’s intricate chain—a giant mirror in the Andes, a record‑breaking camera, a long fiber‑optic link, high‑performance computing and a network of brokers—can work together at something like the scale it was built for.

As LSST moves from testing into full operations, those 800,000 alerts are expected to become a nightly floor rather than a ceiling. Somewhere in the millions of future pings will be the first notice of an unusual asteroid, the earliest glimmer of a peculiar explosion or a subtle signal that challenges current models of the cosmos. The question now is how quickly scientists, and the software they rely on, can learn to hear those signals in the constant noise of a universe under continuous surveillance.