Estimating Luminosity Function Constraints from High-Redshift Galaxy Surveys

TL;DR

This paper uses a Fisher matrix approach to forecast how different high-redshift galaxy survey strategies can constrain the galaxy luminosity function, emphasizing the importance of survey area, depth, and cosmic variance.

Contribution

It introduces a method to estimate luminosity function uncertainties considering sample variance and applies it to optimize high-redshift galaxy survey designs.

Findings

Wide area (~1 deg^2) surveys to H_AB~27 improve constraints.

Deeper UDF-like surveys to H_AB~30 are highly effective.

Splitting surveys into multiple fields offers limited gains due to cosmic variance.

Abstract

The installation of the Wide Field Camera 3 (WFC3) on the Hubble Space Telescope (HST) will revolutionize the study of high-redshift galaxy populations. Initial observations of the HST Ultra Deep Field (UDF) have yielded multiple z>~7 dropout candidates. Supplemented by the Great Observatory Origins Deep Survey (GOODS) Early Release Science (ERS) and further UDF pointings, these data will provide crucial information about the most distant known galaxies. However, achieving tight constraints on the z~7 galaxy luminosity function (LF) will require even more ambitious photometric surveys. Using a Fisher matrix approach to fully account for Poisson and cosmic sample variance, as well as covariances in the data, we estimate the uncertainties on LF parameters achieved by surveys of a given area and depth. Applying this method to WFC3 z~7 dropout galaxy samples, we forecast the LF parameter uncertainties for a variety of model surveys. We demonstrate that performing a wide area (~1 deg^2) survey to H_AB~27 depth or increasing the UDF depth to H_AB~30 provides excellent constraints on the high-z LF when combined with the existing UDF GO and GOODS ERS data. We also show that the shape of the matter power spectrum may limit the possible gain of splitting wide area (>~0.5 deg^2) high-redshift surveys into multiple fields to probe statistically independent regions; the increased root-mean-squared density fluctuations in smaller volumes mostly offset the improved variance gained from independent samples.