Title: Interloper bias in future large-scale structure surveys
Authors: A. R. Pullen, C. M. Hirata, O. Dore, A. Raccanelli
First Author’s Institution: Department of Physics, Carnegie Mellon University, Pittsburgh, PA
Status: To be submitted to PASJ
We look out into a universe that appears deceivingly two-dimensional. Our favorite constellations are often composed of stars that are separated by distances more immense than their proximity to each other suggests. This artificial two-dimensionality of the observed universe has forever been a bane of astronomy, for it takes a lot to squeeze information about the third dimension out of the universe. Deprojecting our 2D sky into a true 3D map by measuring distances to objects is an astronomical enterprise of its own, built up first from inch-long measuring sticks used exclusively for nearby objects, which are replaced by yardsticks as we move further out, to mile markers even further out, and so on. We can, for example, use predictably varying stars called classical Cepheids to determine distances up to about 30 Mpc, a little beyond the nearest galaxy cluster, Virgo. Type Ia supernovae, stellar explosions that achieve the same brightness each and every time they go off, no matter when or where they exploded, help us measure distances as much as 30 times further. Each measuring stick in the sequence is calibrated by the sequence of shorter measuring sticks that came before it, a sequence which astronomers have called the “distance ladder.” Thus errors and uncertainties in calibrating one measuring stick can propagate up the sequence, much like falling dominoes. We’ve directly measured the distances of only a small fraction of celestial objects; for a vast majority of the objects in the universe, we must turn to our sequence of sticks.
For objects far beyond the gravitational influence of our galactic neighborhood, the measuring stick of choice is the object’s redshift. This is unique to a universe that’s expanding uniformly and homogenously, which causes things further from you to appear to move away from you faster. Much like how the pitch of an emergency siren falls as it flys away from you, the wavelength of the light from an object moving away from you becomes longer and longer, causing it to look redder. The amount an object’s light is “redshifted” depends predictably on the object’s distance—a relation so robust that it has been codified into what’s known as Hubble’s law.
Hubble’s law has embolded cosmological cartographers to take up the herculean task of drawing a 3D map of our universe. The feat requires measuring redshifts of a huge sample of galaxies via large spectroscopic surveys. The first such survey, begun in the 1970s, contained a few thousand galaxies. The biggest survey completed to date, the Sloan Digital Sky Survey (SDSS), contains nearly a million galaxies. These maps revealed that the universe on its largest scales is fascinatingly varied and structured. There are walls of galaxies surrounding vast, empty voids; galaxies are often assembled together to form fractal-like filamentary strands; at nodes where the filaments intersect, one can find the densest and largest clusters of galaxies. The maps also contain clues to the physics and the cosmological parameters that govern the past and future evolution of our universe.
Thus even more ambitious surveys are in the works. Our quest for more galaxies requires us to search for ever fainter galaxies, for which reliable redshifts are difficult to measure. But it’s not impossible. One can look for an easy-to-find, strong spectral feature typical in galaxies and measure how much redder it’s become. It would have been a fairly straightforward task, except for one catch—there’s a handful of strong features that can easily be mistaken for each other. These interloping lines could cause a galaxy to be mistakenly given an incorrect redshift, and thus distance.
The authors of today’s paper thus asked, how much do galaxies with incorrect distances based on a single emission line affect our maps and the physics we infer from them? They looked at how upcoming spectroscopic redshift surveys undertaken with the Prime Focus Spectrograph (PFS) to be installed on the 8.2-meter Subaru Telescope and the Wide-Field InfraRed Survey Telescope (WFIRST) could be affected by interloping galaxies. In particular, the authors studied how the matter power spectrum, an important measure of the amount of mass found at varying cosmological size scales, derived from the two surveys would be affected. They found that if more than 0.2% of the galaxies were interlopers with incorrect distances, they can increase the total error by 10%. If more than 0.5% of the galaxies were interlopers, they can drastically skew the matter power spectrum at small scales. Such effects have consequences for many other cosmological studies, including those concerning dark energy and modified gravity.
Can the interlopers be weeded out somehow? The authors investigate two methods to identify interlopers. One could repeat the emission line analysis but for pairs of strong lines, since each of the strong lines pairs that PFS and WFIRST could measure have unique wavelength separations. Alternatively, one could independently measure the redshift of each galaxy based on the galaxy’s color, derived from a separate photometric survey. The authors tested these two interloper removal methods on a mock sample of galaxies and found that finding strong line pairs alone can help remove most of the interlopers in the PFS survey, while a combination of finding pairs and calculating photometric redshifts must be done together to remove interlopers in the WFIRST survey.
To see a video of the first author A. Pullen explaining this paper, follow this link.