Science@Berkeley Lab nameplate Berkeley Lab logo
November 29, 2005
Supernovae: The Stellar Route to Understanding Dark Energy
Part 2: New Methods for Studying Type Ia Supernovae

Second in a series on the recent Supernova Workshop sponsored by SNAP, the SuperNova/Acceleration Probe collaboration, looking at the best ways to study dark energy with exploding stars

Distant supernovae are moving away from the observer at great speed; elemental lines that would be in the blue part of the spectrum if the object were not moving (the rest frame) are shifted well into the red or infrared. Although spectra are routinely corrected to see what they would look like in the rest frame, at very high redshifts and great distances some information is lost.

The more distant the object, the more the universe has expanded since the light we see was emitted. Features in the spectrum of a galaxy seen at rest (top) appear at longer wavelengths and lower frequencies (toward left) as redshift increases; redshift at bottom is z = 0.2. (After Sloan Digital Sky Survey)

An object at rest with respect to the observer has a redshift z = 0, meaning that the observed frequency and wavelength are identical to the emitted frequency and wavelength. At z = 1, the light observed has twice the wavelength and half the frequency of the light emitted. Like "nearby" and "distant," the terms "high and low redshift" are imprecise, but anything over z = 0.7 is high redshift and certainly distant, and anything over z = 1 is very far away indeed.

Besides establishing basic information like a supernova's type, measuring different colors at intervals may reveal differences in the shapes of their light curves. (A term often used as a kind of shorthand for the light curve is Δ m15, pronounced "delta em 15," introduced by Mark Phillips in 1993; it means the difference between the magnitude, or brightness, measured at maximum brightness and when measured again 15 days later.)

At the Supernova Workshop, Berkeley Lab's Gerson Goldhaber and Mark Strovink discussed some of the information that can be extracted from different color bands, using a color-magnitude diagram and the technique known as CMAGIC (for Color Magnitude Intercept Calibration).

One way Type Ia supernovae differ is that some of them, typically those said to have more "stretch" (a method for calibrating brightness by fitting the widths of light curves to a template), show an additional increase in brightness, a second peak or bump, in the B band (blue light). Said Goldhaber, " Type Ia's are not a one-parameter family — some supernovae with bumps have virtually the same stretch as others without bumps. Yet this and other information from CMAGIC analysis can be used to select the best standard candles for cosmology."

Color differences may be intrinsic, due to slight differences in the supernova's precursor or just how it exploded, although the question remains open. A factor that is certainly intrinsic, however, and gives insight into the nature of the explosion and its geometry is a supernova's polarization.

Berkeley Lab's Lifan Wang pioneered supernova polarimetry at the University of Texas and the European Southern Observatory. Although a supernova is a point source of light, when that light is polarized it strongly suggests that the source is asymmetric. Moreover, different parts of its spectrum may be polarized to different degrees.

Observers can use this information as another, perhaps more sensitive way to correct seemingly different Type Ia spectra and light curves. At the workshop Wang presented a technique he calls SQI, spectral quality index, intended to get rid of "manual errors" by removing the researcher's possible bias when comparing the spectra and light curves of a diverse population of Type Ia's.

Spectropolarimetry data can be used to model asymmetry. The light from an exploding white dwarf with a red giant companion may be polarized the same in every direction unless viewed directly in line with the "hole" the companion makes in the expanding ejecta (top, in this simulation). (Kasen and Nugent with LBNL Computational Research Division Visualization Group)

A more direct use of spectropolarimetry is as a measure of the shape of the supernova — whether the explosion has been observed through a surrounding accretion disk, for example — and whether it exploded smoothly, or off-center, or in clumps. Accretion disks could account for much of the variation in Type Ia observations: the spectrum of a supernova seen sideways, through an accretion disk, would differ from that of the same supernova seen end-on. Spectropolarimetry is critical to distinguishing between different points of view.

Cataclysmic modeling

Theoretical models realized on powerful supercomputers may be the best hope for a thorough understanding of Type Ia physics. Peter Nugent, a theorist in Berkeley Lab's Computational Research Division, said, "It would be great if we could explode a Type Ia ab initio" — from first principles — "but that's probably too much to ask just yet." Nevertheless, even simple computer models can help put limits on the observables. "To improve the empirical observations it's necessary to push the models."

A question that arose repeatedly during the workshop was how individual peculiarities in Type Ia spectra might be accounted for by physical processes. The notion of a two-stage thermonuclear explosion has long been popular — an explosion that begins with a subsonic deflagration but later makes the transition to a supersonic detonation. The way to test such ideas, said Fritz Röpke of the Max Planck Institute for Astrophysics, is to construct three-dimensional models as detailed as practical on available supercomputers, working as much as possible from first principles.

Röpke and his colleagues based their calculations on a canonical white dwarf star, made of carbon and oxygen and having a radius of about a thousand kilometers, as it reaches the Chandrasekhar mass. Starting from the moment ignition is triggered, the researchers apply mathematical concepts to understand what follows, including level sets and adaptive mesh refinement techniques originated by James Sethian and Phillip Colella of Berkeley Lab's Computational Research Division.

"Initially the flame is very narrow and there is no internal turbulence, but generic instabilities quickly create turbulent eddies that interact to unevenly accelerate the material inside the star," Röpke said. "Is there a later transition to detonation? So far there is no known physical mechanism for such a transition; we would have to introduce free parameters into our model. So we wonder if we can model what has been observed in actual Type Ia's without detonation."

A high-resolution, fully three-dimensional model has yet to be finished, but the best simulations so far — requiring half a million hours on a massively parallel IBM supercomputer — indicate that, at least for lower-energy Type Ia's, deflagration alone produces results that compare favorably with real supernovae. At low energies the distribution of nuclear species, evidence for clumpiness, and other differences noted by Stefano Benetti could be accounted for by this straightforward model.

Dan Kasen, now at Johns Hopkins University and the Space Telescope Science Institute, discussed models of Type Ia explosions developed with codes written in collaboration with Nugent and others when he was at Berkeley Lab. Like Röpke's, Kasen's models aim to match specific Type Ia observations, including data from spectropolarimetry, light-curve shape, and the position of elemental lines in the supernova's spectrum.

Computer simulations indicate that when a deflagration flame burns outward from the center of a white dwarf star, ascending blobs of burning material cause turbulence that deforms the flame and causes an asymmetric explosion. (Friedrich Röpke, Max Planck Institute for Astrophysics)

To account for features observed in the nearby Type Ia supernova SN 2001el, which was strongly polarized, with a distinctive calcium II absorption line differently polarized from the supernova as a whole, Kasen's code specified an off-center blob of calcium-rich material exploding outward and remaining distinct from the rest of the ejecta. In this case, the geometry of the supernova and the angle from which it was observed are crucial to understanding its physics.

Unfortunately, as discussions at the workshop made clear, spectropolarimetry is not easy to do; long observation times are required and adequate instruments are rare, challenges which must be met before this promising technique can be applied to its full potential.

The near and the far

Distant supernovae, 10 billion light years away or more, have become relatively easy to find from the ground: the Supernova Cosmology Project pioneered the technique of searching thousands of galaxies at a time at the right intervals to guarantee finding a dozen or more fresh exploding stars in each batch, most of them before they reached peak brightness.

Finding them is dependable, but distant supernovae are extremely faint, and their light is shifted so far into the red or infrared that some wavelengths of their spectra are difficult or impossible to recover. Ground-based observations are further hindered by the atmosphere's turbulence and absorption of infrared light.

Confidently removing systematic errors in distant observations thus depends largely on understanding the physics of Type Ia explosions. More urgent than good theoretical models, in the opinion of some participants — one called it "screamingly obvious" — is that "we need more nearby supernovae."

Nearby supernovae — those only a couple of billion light years away — can be measured with greater precision because they are brighter and more of their spectra is accessible. Unfortunately they are not easy to find; nearby galaxies are scattered, and no one patch of sky includes more than a few of them.

Three groups discussed their efforts to collect nearby supernovae at the workshop. Weidong Li of the University of California at Berkeley described a long-running search started in 1993 by UC Berkeley astronomy professor Alex Filippenko, which now uses a robotic telescope called KAIT (the Katzman Automatic Imaging Telescope) at the Lick Observatory to look at the same well-characterized galaxies repeatedly. Over the years KAIT has collected hundreds of supernovae, "of which 88 are really good Type Ia's," Li said. Analysis of candidates begins with quick image processing by what he called "an army of undergraduates."

Nearby-supernova searches presently underway include the Carnegie Supernova Project at the Las Campanas Observatory, the Nearby Supernova Factory using the University of Hawaii's 2.2 meter telescope on Mauna Kea, and an ongoing search with the Katzman Automatic Imaging Telescope at the Lick Observatory.

Greg Aldering, who leads the international Nearby Supernova Factory (SNfactory) based at Berkeley Lab, described a different approach, using images obtained from the Jet Propulsion Laboratory's Near Earth Asteroid Tracking program (NEAT). While searching the solar system for Earth-threatening asteroids, NEAT's automated telescopes incidentally image thousands of galaxies, revisiting the same regions roughly every six days during a typical 18-day observing period. When a supernova appears in the huge data stream from NEAT, SNfactory software running at DOE's National Energy Research Scientific Computing Center (NERSC) nabs it.

NEAT galaxies tend to be a little farther away than the ones observed by KAIT, said Aldering, "but still short of cosmology." The assumption is that most are in the "smooth Hubble flow": their motion away from the observer is caused only by the universe expanding and is less likely to be affected by a galaxy's idiosyncratic motions toward or away from the observer. Thus the spectra of supernova in these galaxies can be more confidently correlated with their maximum brightness.

Key to the effort is follow-up of newly discovered Type Ia supernovae with the sophisticated SNIFS spectrometer (the Supernova Integral Field Spectrograph), attached to the University of Hawaii's 2.2-meter telescope on Mauna Kea, an instrument assembled by the SNfactory's French team members. Of over 100 supernova candidates so far, the SNfactory has found 10 Type Ia's, although analysis is not complete; once the program's kinks have been ironed out, the researchers hope to find and study 300 to 600 Type Ia supernovae within a few years.

Mark Phillips, on the scientific staff of the Carnegie Observatories and associate director of Las Campanas Observatory in Chile, described early results of the Carnegie Supernova Project, which studies both Type I and II supernovae in the "local" universe. The goal is to obtain exceedingly well calibrated light curves and spectroscopy for some 200 supernovae of all types to pin down variations in spectra and establish means of correcting for reddening.

Like other searches, the Carnegie Project is searching for evidence of systematic differences due to evolutionary effects — for supernovae whose home galaxies are poor in heavier elements. The relative abundance of elements heavier than lithium is known as "metallicity"; all such elements are made in stars (and by supernovae themselves), and young galaxies presumably have fewer of them.

The Carnegie Supernova Project, which in its first season of searching for low-redshift (nearby) targets has collected 39 supernovae, has the advantage of what Phillips calls "a ridiculous number of nights to observe" — 300 nights a year on telescopes equipped with precision instruments at Las Campanas.

Collecting quality data is half the battle, however; analysis remains a challenge. The nearby-supernova searchers tossed around ideas about sharing Carnegie and KAIT data with the SNfactory, where SNIFS and NERSC provide advantages for follow-up spectroscopy and analysis.

The final installment in this series will report on the ongoing search for high-z and "higher-z" supernovae, and on the discussion of issues raised by the Supernova Workshop.

Additional information

Participants in the SNAP Supernova Workshop, September 2005:

Greg Aldering

Stefano Benetti

David Branch

Ben Dilday

Josh Frieman

Peter Garnavich

Gerson Goldhaber

Peter Hoeflich

Isobel Hook

David Jeffery

Daniel Kasen

Alex Kim

Kevin Krisciunas

Michael Levi

Weidong Li

Peter Nugent

Saul Perlmutter

Mark Phillips

Adam Riess

Fritz Röpke

Roger Romani

Masao Sako

Mark Strovink

Mark Sullivan

Nicholas Suntzeff

Lifan Wang