Measuring Quasar Variability with Pan-STARRS1 and SDSS

TL;DR

This paper demonstrates that quasar variability measured through Pan-STARRS1 and SDSS data can effectively distinguish quasars from non-varying objects, improving selection purity and providing models for variability dependence on redshift, luminosity, and wavelength.

Contribution

The study introduces a method leveraging variability in large surveys to select quasars more efficiently, especially in redshift ranges where photometric methods are less effective.

Findings

Variability distinguishes quasars from non-varying objects with high efficiency.

Improves quasar selection purity from 4.1% to 48%.

Provides models for quasar variability dependence on redshift, luminosity, and wavelength.

Abstract

We measure quasar variability using the Panoramic Survey Telescope and Rapid Response System 1 Survey (Pan-STARRS1 or PS1) and the Sloan Digital Sky Survey (SDSS) and establish a method of selecting quasars via their variability in 10,000 square degree surveys. We use 100,000 spectroscopically confirmed quasars that have been well measured in both PS1 and SDSS and take advantage of the decadal time scales that separate SDSS measurements and PS1 measurements. A power law model fits the data well over the entire time range tested, 0.01 to 10 years. Variability in the current PS1-SDSS dataset can efficiently distinguish between quasars and non-varying objects. It improves the purity of a griz quasar color cut from 4.1% to 48% while maintaining 67% completeness. Variability will be very effective at finding quasars in datasets with no u band and in redshift ranges where exclusively photometric selection is not efficient. We show that quasars' rest-frame ensemble variability, measured as a root mean squared in delta magnitudes, is consistent with V(z, L, t) = A0 (1+z)^0.37 (L/L0)^-0.16 (t/1yr)^0.246 , where L0 = 10^46 ergs^-1 and A0 = 0.190, 0.162, 0.147 or 0.141 in the gP1 , rP1 , iP1 or zP1 filter, respectively. We also fit across all four filters and obtain median variability as a function of z, L and lambda as V(z, L, lambda, t) = 0.079(1 + z)^0.15 (L/L0 )^-0.2 (lambda/1000 nm)^-0.44 (t/1yr)^0.246 .