[Paper Review] Setting UBVRI Photometric Zero-Points Using Sloan Digital Sky Survey ugriz Magnitudes

Taylor S. Chonis, C. M. Gaskell|Insecta mundi|Oct 31, 2007

Adaptive optics and wavefront sensing4 citations

TL;DR

This paper presents a practical method for setting UBVRI photometric zero points in CCD images using Sloan Digital Sky Survey (SDSS) ugriz magnitudes. It identifies systematic zero-point offsets in B, V, R, and I bands at g, r, i ≈ 14.5 and in U at u ≈ 16 due to PSF saturation, and recommends using only unsaturated stars with g, r, i in 14.5–19.5 and u in 16–19.5 for reliable calibration.

ABSTRACT

We discuss the use of Sloan Digital Sky Survey (SDSS) ugriz point-spread function (PSF) photometry for setting the zero points of UBVRI CCD images. From a comparison with the Landolt (1992) standards and our own photometry we find that there is a fairly abrupt change in B, V, R, & I zero points around g, r, i ~ 14.5, and in the U zero point at u ~ 16. These changes correspond to where there is significant interpolation due to saturation in the SDSS PSF fluxes. There also seems to be another, much smaller systematic effect for stars with g, r > 19.5. The latter effect is consistent with a small Malmquist bias. Because of the difficulties with PSF fluxes of brighter stars, we recommend that comparisons of ugriz and UBVRI photometry should only be made for unsaturated stars with g, r and i in the range 14.5 - 19.5, and u in the range 16 - 19.5. We give a prescription for setting the UBVRI zero points for CCD images, and general equations for transforming from ugriz to UBVRI.

Motivation & Objective

• To address the challenge of calibrating UBVRI CCD photometry when standard stars are unavailable.
• To identify systematic zero-point offsets in UBVRI magnitudes when derived from SDSS ugriz photometry.
• To provide a practical, empirically derived method for setting UBVRI zero points using SDSS data.
• To minimize errors from PSF saturation and Malmquist bias in the transformation process.

Proposed method

• The authors use SDSS Data Release 5 ugriz photometry of Landolt (1992) standard stars to derive transformation equations.
• They apply color-color cuts: 0.08 < r−i < 0.5 and 0.2 < g−r < 1.4, and remove outliers more than 2.5σ from linear fits.
• Transformation equations are derived for B, V, R, I, and U using linear fits in color space, with U derived independently due to its sensitivity.
• The method excludes stars with saturation warning flags and restricts analysis to stars with g, r, i in 14.5–19.5 and u in 16–19.5.
• A prescription is provided for setting zero points: match stars to SDSS, clean color-color plots, apply equations (1)–(5), and use mean magnitudes from fainter stars.
• The approach avoids higher-order color terms and focuses on practical calibration rather than astrophysical precision.

Experimental results

Research questions

• RQ1 What are the systematic offsets in UBVRI zero points when derived from SDSS ugriz PSF magnitudes for bright stars?
• RQ2 How does PSF saturation in SDSS affect the reliability of ugriz-to-UBVRI transformations?
• RQ3 What magnitude ranges ensure minimal error in UBVRI zero-point calibration using SDSS data?
• RQ4 Is there a Malmquist bias effect in the transformation at faint magnitudes, and how significant is it?
• RQ5 Can a simple linear transformation be used reliably across a restricted color and magnitude range for practical photometric calibration?

Key findings

• A sharp change in B, V, R, and I zero points occurs at g, r, i ≈ 14.5 due to PSF saturation in SDSS.
• The U band zero point shows a similar discontinuity at u ≈ 16, with SDSS ugriz magnitudes underpredicting U by up to 2 mag for bright stars.
• For stars with g, r ≥ 19.5, a small systematic offset is observed, consistent with a Malmquist bias effect.
• The Malmquist bias is estimated as a lower limit; its true magnitude could be up to twice the observed slope.
• The transformation equations (1)–(5) are valid only for stars with g, r, i in 14.5–19.5 and u in 16–19.5 to avoid saturation and bias effects.
• The U filter transformation is U = u − 0.854 ± 0.007, with no significant dependence on (u−g) color in the selected range.