Real-time Analysis and Selection Biases in the Supernova Legacy Survey

TL;DR

This paper discusses the real-time detection and analysis of supernovae in the SNLS, highlighting observational biases like Malmquist bias, and presents methods to correct these biases for accurate cosmological measurements.

Contribution

It introduces a detailed analysis of selection biases in SNLS and proposes correction methods to improve the reliability of supernova-based cosmology.

Findings

Biases become significant at z~0.6

Spectroscopic sample shifts towards brighter magnitudes at z>0.75

Bias correction methods reduce systematic uncertainties

Abstract

The Supernova Legacy Survey (SNLS) has produced a high-quality, homogeneous sample of Type Ia supernovae (SNe Ia) out to redshifts greater than z=1. In its first four years of full operation (to June 2007), the SNLS discovered more than 3000 transient candidates, 373 of which have been confirmed spectroscopically as SNe Ia. Use of these SNe Ia in precision cosmology critically depends on an analysis of the observational biases incurred in the SNLS survey due to the incomplete sampling of the underlying SN Ia population. This paper describes our real-time supernova detection and analysis procedures, and uses detailed Monte Carlo simulations to examine the effects of Malmquist bias and spectroscopic sampling. Such sampling effects are found to become apparent at z~0.6, with a significant shift in the average magnitude of the spectroscopically confirmed SN Ia sample towards brighter values for z>0.75. We describe our approach to correct for these selection biases in our three-year SNLS cosmological analysis (SNLS3), and present a breakdown of the systematic uncertainties involved.