[Paper Review] Variability Studies with SDSS

Željko Ivezić, R. Lupton+11 more|ArXiv.org|Jan 20, 2003

Astronomy and Astrophysical Research4 citations

TL;DR

This paper demonstrates that the Sloan Digital Sky Survey (SDSS) enables high-precision, wide-area variability studies using multi-epoch photometry of 3 million point sources across 700 deg². With photometric accuracy of ~0.02 mag and well-understood errors, SDSS reveals that quasars dominate variable sources fainter than r ≈ 18 on timescales >3 months, while stars dominate at brighter magnitudes and shorter timescales (e.g., 3 hours).

ABSTRACT

The potential of the Sloan Digital Sky Survey for wide-field variability studies is illustrated using multi-epoch observations for 3,000,000 point sources observed in 700 deg2 of sky, with time spans ranging from 3 hours to 3 years. These repeated observations of the same sources demonstrate that SDSS delivers ~0.02 mag photometry with well behaved and understood errors. We show that quasars dominate optically faint (r > 18) point sources that are variable on time scales longer than a few months, while for shorter time scales, and at bright magnitudes, most variable sources are stars.

Motivation & Objective

• To assess the potential of SDSS for wide-field variability studies using repeated multi-epoch observations.
• To evaluate the photometric accuracy and error characteristics of SDSS data across five bands (u, g, r, i, z).
• To classify variable sources by magnitude and timescale, identifying dominant populations (quasars vs. stars).
• To demonstrate that SDSS photometric errors are well-behaved and nearly Gaussian, enabling robust statistical analysis of variability.
• To provide a foundation for identifying variable objects, including quasars and RR Lyrae stars, in large-scale surveys.

Proposed method

• Utilized 3 million point sources from 700 deg² of SDSS sky coverage with time baselines from 3 hours to 3 years.
• Applied a 3σ significance threshold and a minimum magnitude variation of 0.075 mag in both g and r bands to identify variable sources.
• Compared repeated observations (733 days apart and 3 hours apart) to assess photometric error distributions and accuracy.
• Used color-color and color-magnitude diagrams to classify variable sources based on their photometric colors (e.g., u-g, g-r, r-i).
• Calculated photometric error distributions using interquartile range and compared them to Gaussian models to validate error estimates.
• Employed SDSS photometric pipeline outputs to compute formal errors and normalized magnitude differences to test error model accuracy.

Experimental results

Research questions

• RQ1 What is the photometric accuracy of SDSS multi-epoch observations across five bands, and how well do formal errors match empirical distributions?
• RQ2 Which types of astrophysical objects dominate the variable source population at different magnitudes and timescales?
• RQ3 How do the color distributions of variable sources differ between short-timescale (3 hours) and long-timescale (2 years) observations?
• RQ4 To what extent do quasars dominate the variable source population at faint magnitudes (r ≥ 18) on timescales longer than a few months?
• RQ5 Can SDSS data reliably distinguish between variable stars (e.g., RR Lyrae) and quasars using photometric colors and variability amplitude?

Key findings

• SDSS achieves a photometric accuracy of ~0.02 mag with well-controlled error distributions, as confirmed by 3σ outlier fraction of 0.9% (vs. 0.3% expected for a perfect Gaussian).
• For time baselines of ~2 years, 77% of variable sources have colors typical of low-redshift quasars (u-g < 0.6), indicating quasars dominate the faint variable population (r ≥ 18).
• At 3-hour timescales, only 2% of variable sources have quasar-like colors, with RR Lyrae stars comprising 35% of the sample, highlighting their dominance on short timescales.
• The photometric error distribution for bright stars (r < 19) observed 3 hours apart closely follows a Gaussian with σ = 0.02 mag, confirming high data quality.
• The formal photometric errors computed by the SDSS pipeline are highly accurate, with normalized magnitude differences showing σ ≈ 1 across all bands.
• RR Lyrae stars are detectable as a distinct population due to their characteristic colors and variability, with a surface density of ~1 per deg², nearly independent of timescale or location.