Seven years ago, an international collaboration of astronomers installed a state-of-the-art camera on a robotic telescope at the Palomar Observatory near San Diego. Today, the Zwicky Transient Facility holds the largest census of cosmic supernovae - flashes of light in the sky that tell us of stars dying in spectacular explosions.

“There are trillions of stars in the universe, and about every second, one of them explodes. ZTF detects hundreds of these explosions per night and a handful are then confirmed as supernovae. Systematically doing this for seven years has led to the most complete record of confirmed supernovae to date,” says Christoffer Fremling, a staff astronomer at Caltech who leads ZTF’s Bright Transient Survey (BTS) program that was dedicated to the supernova search.

The Bright Transient Survey is currently the primary discovery machine for cosmic flashes - called astronomical transients - in the world. ZTF shares a stream of transient detections with the wider astronomical community for further investigation via spectral analysis. This analysis uses instruments to split the light from the transient into its color composition to reveal its distance, type, and other physical properties. Most of the transients in the BTS sample are classified as one of two most common types - supernova Type Ia when a white dwarf steals material from another nearby star until it explodes or supernova Type II when massive stars die and collapse under their gravity.

Thanks to this treasure trove of data, astronomers are now better equipped to answer questions such as what transient demographics tell us about stellar evolution and death, how dark energy drives the expansion of the universe, and how exactly stars die.

“We have made incredible progress in our ability to browse the universe for transients. We have also built both the technology and the science community needed to enable significant strides in our understanding of the dynamic cosmos,” says Fremling.

About 70% of the supernovae in our BTS survey have been detected and classified by the ZTF team with a carefully designed relay between two telescopes at the Palomar Observatory near San Diego.

The first one is the 1.2m Samuel Oschin telescope. It scans the entire visible sky every two nights using an advanced 60Mpix wide-field camera mounted at its focus. To detect new astronomical events, astronomers subtract images of the same portion of the sky from subsequent scans.

In the next phase of the discovery process, members of the ZTF team study the subtracted images and trigger further observations for the most promising candidates with the neighboring 1.5m telescope that houses the ZTF spectrograph called the SEDM.

“We combine the brightness information from the ZTF camera with the data from the SEDM to correctly identify the origin and type of a transient, a process astronomers call transient classification,” says Yu-Jing Qin, a postdoc at Caltech who is running much of the daily operations of the BTS survey.

Since 2012, astronomers have kept track of the transient discoveries on a public platform called the Transient Name Server (TNS). The global community uses the platform to announce both detections and classifications of transient events. By sharing their discoveries, astronomers create a virtual global observatory where unclassified transients can be classified by other astronomy teams with spectroscopic facilities. As a result of this joint effort, today the TNS holds approximately 16,000 records of classified supernovae and other transients.

A critical bridge from detection to classification is the community of astronomers who examine image data from ZTF or other surveys and decide which transients merit further follow-up for classification. Within the ZTF partnership, the most avid among them is Jesper Sollerman, a professor at Stockholm University, Sweden, and a prolific supernova hunter.

“Because of the time difference between the US and Europe, I wake up to a set of freshly observed objects in the Bright Transient Survey every morning. In the last six years, I have invariably had my morning coffee while sieving through supernova candidates in the latest ZTF data,” says Sollerman.

Sollerman and astronomers like him explore the night sky from the comfort of their homes thousands of miles from the telescope. Things were quite different in 1933 when Fritz Zwicky began more systematic observations and discoveries of supernovae at Palomar. He spent long hours shivering in the cold dome during the observations, and many more examining hundreds of photographic plates that carried the images of the night sky at that time. Fritz found 120 supernovae in 52 years and many of his discoveries were made with the same telescope ZTF is mounted on today. No single person could match this record until 2009.

“Without having to freeze in the dark dome the way Fritz Zwicky did, I have discovered several thousands of these classified supernovae in the past six years”, says Sollerman.

“Most of the supernovae I find never even get classified. But I think my days as a primary supernova discoverer are counted, there is simply too much data to keep up with,” adds Sollerman.

Two years ago, Fremling teamed up with machine learning experts at Caltech to train computers to read the SEDM spectroscopic data, classify the supernovae, and report the results automatically to the Transient Name Server minutes after the actual observations. In 2023, PhD student Nabeel Rehemtulla at Northwestern University extended the use of machine learning to the full observation cycle. He developed the BTSbot system that is currently used in ZTF to discover, classify, and report supernovae without human involvement.

“Since BTSbot began operation it has found about half of the brightest ZTF supernovae before a human. For specific types of supernovae, we have automated the entire process and BTSbot has so far performed excellently in over a hundred cases. This is the future of supernova surveys especially when the Vera Rubin Observatory begins operations,” adds Rehemtulla.

The Vera Rubin Observatory is being built in the Chilean mountains and when complete it will be much more sensitive than ZTF making supernova discoveries in the millions.

“The machine learning and AI tools we have developed for ZTF will become essential when the Vera Rubin Observatory begins operations,” says Daniel Perley, an astronomer at Liverpool John Moores University in the UK who developed the search and discovery procedures for the BTS, and manages the public survey database. “We have already planned to work closely with Rubin to transfer our machine learning knowledge and technology,” adds Perley.

Thanks to an additional 1.6 million of funding from the National Science Foundation, ZTF will continue to scan the night sky and enrich the BTS survey in the coming two years.

“The period in 2025 and 2026 when ZTF and Vera Rubin can both operate in tandem is fantastic news for time-domain astronomers,” says Mansi Kasliwal, an astronomy professor at Caltech who will lead ZTF in the coming two years. “Combining data from both observatories, astronomers can directly address the physics of why supernovae explode and discover fast and young transients that are inaccessible to ZTF or Rubin alone. I am excited about the future,” adds Kasliwal.

Links

Bright Transient Survey

Transient Name Server

ZTF receives more support from the NSF

Science papers

The Zwicky Transient Facility Bright Transient Survey. III. BTSbot: Automated Identification and Follow-up of Bright Transients with Deep Learning

The Zwicky Transient Facility Bright Transient Survey. II. A Public Statistical Sample for Exploring Supernova Demographics

The Zwicky Transient Facility Bright Transient Survey. I. Spectroscopic Classification and the Redshift Completeness of Local Galaxy Catalogs

Contact

Science Contact
Christoffer Fremling
Staff astronomer (Caltech)
email: fremling [at] caltech [dot] edu

Media Contact
Ivona Kostadinova
ZTF Coordinator (Caltech)
email: ivonata [at] caltech [dot] edu