Research, methods and collaboration

Late 1980s–1999

Finding Supernovae in Batches

Follow the Supernova Cosmology Project’s search sequence from reference images to spectra, and work through a scheduling example for a short-lived target.

Original explanatory diagram: Finding Supernovae in Batches

The short answer

How can an observatory schedule time for transient events that have not yet appeared?

Read the evidence

Records from different observing stages

Discovery establishes a candidate. Follow-up establishes evidence for a comparison.

Original comparison based on the cited records.
StageWhat is recordedWhat comes next
ReferenceField before the new transientA matched image of the same field
DifferenceA candidate signal above the earlier sceneTests for artifacts and a spectrum
Follow-upClassification and brightness over timeCalibration and sample selection

Transient events and fixed calendars

A supernova survey needs more than images that happen to contain an explosion. It needs enough time after detection for spectroscopy and measurements as the source brightens or fades. Distant objects demand telescope time that is often allocated well in advance. An astronomer cannot name a particular new target months before it appears. The Supernova Cosmology Project managed this by arranging repeated visits to fields containing many galaxies. It scheduled a discovery epoch and secured nearby follow-up slots. The random event remained unpredictable in any single galaxy, but a field containing many plausible hosts could supply a batch.

Sources for this section: [3] · [2]

Subtract the scene before the event

A reference image records galaxies and other steady features before the discovery visit. Images of the same field are aligned and their apparent intensities made comparable. An added point of light can then stand out against the host after subtraction. Poorly matched images can also leave a residual. Seeing a candidate in the difference therefore begins a new decision: is it an astronomical transient and, if so, which kind? A spectrum and repeated brightness observations add information that the subtraction cannot supply. A batch is a group of candidate opportunities, not a final sample of certified Type Ia objects.

Sources for this section: [3] · [1]

An invented observing schedule

Suppose reference imaging is done on day 0, discovery imaging on day 17 and spectroscopy on day 19. A candidate that first becomes visible on day 13 can appear in the difference image and still have a chance of being measured on day 19. If the only scheduled spectrum falls on day 45, its usefulness depends on how much the transient has faded. Move the reference image to day 15 and the same source may already be present in it. These dates illustrate the scheduling logic. They are not the project’s observing log and they do not predict the lifetime of any real supernova.

Sources for this section: [3]

The first batch and the final sample

Berkeley Lab’s account recalls five discoveries from an early successful campaign in late 1993 and early 1994. Years later, the 1999 result combined 42 distant supernovae with a nearby comparison sample. The numbers represent separate stages. The early batch illustrated the ability to find targets during a planned run. The published analysis addressed cosmological parameters after classification and brightness calibration. In general, weather, source brightness and observing conditions can affect which discoveries become usable measurements. The resulting selection needs attention when interpreting the final sample.

Sources for this section: [2] · [4]

Why this technique changed the feasible question

By connecting the search to reserved follow-up time, the project could make repeated distance measurements instead of relying on occasional exceptional events. This improved the possibility of asking how the brightness–redshift relationship changes across the sample. The procedure itself did not decide the answer. It created the conditions for observations with enough common structure to be analysed together. When reading the resulting diagram, keep the sample-building steps nearby. The plotted point is the outcome of discovery, classification and a fitted light curve, and different failures along that chain can remove an object before it reaches the plot.

Sources for this section: [3] · [4]

Continue the chronology

Biography context

Explore United States through connected lives

Connected biographies

Researchers and collaborators

United StatesCulture

1959–

Saul Perlmutter

Experimental cosmologistExplore Saul Perlmutter’s career, the role of Carl Pennypacker and the Berkeley collaboration, and how scheduled searches made distant supernovae useful.Read biography
AustraliaCulture

1967–

Brian Schmidt

Astronomer and research leaderTrace Brian Schmidt from Arizona and Harvard to Mount Stromlo, covering his Type II distance method and leadership of the High-z supernova team.Read biography
United StatesCulture

1969–

Adam Riess

Observational cosmologistFollow Adam Riess from MIT and Harvard to the High-z Supernova Search Team, including his work on supernova colours, distance and cosmic expansion.Read biography

Quick answers

Common questions

Why take a reference image?
It records the steady field so a change can stand out when images are compared.
Does a difference image classify a supernova?
Its residual identifies a candidate. Classification requires other observations, especially a spectrum.

Primary and institutional sources

How this history was checked

Dates, legal milestones and historical claims are checked against the official and institutional records below. This article was reviewed on 13 November 2026.

  1. Nobel Prize OutreachSaul Perlmutter — biographical account
  2. Berkeley LabSaul Perlmutter Wins Nobel Prize in Physics
  3. Saul Perlmutter, Rev. Mod. Phys. 84, 1127Measuring the acceleration of the cosmic expansion using supernovae
  4. Supernova Cosmology Project, 1999Measurements of Omega and Lambda from 42 High-Redshift Supernovae