measure_cg_bias

This repository contains script to estimate bias on weak lensing (WL) measurements with the Large Synoptic Survey Telescope (LSST) from galaxy color gradients. For a more in-depth understanding, refer to this document describing measurement techniques and results in detail.

Along with python scripts to compute CG bias, the repository also contains the following directories:

1. notebooks has jupyter notebooks with images of the galaxies and PSF used in this analysis as well as the results.
2. data contains template SEDs, LSST filter response curves, and HST noise correlation functions used in the analysis.

What are galaxy color gradients?

Galaxies usually do not have a uniform spectral energy densities (intensity at each wavelength) across it’s spatial profile. Galaxy color gradients denote that the galaxy has different colors in different places, as can be seen by this image of the Pinwheel galaxy (M101).

Why is there a bias?

Weak lensing measurements involve estimating the correlations in the distortions of galaxy shapes due to gravitational lensing from matter along the line of sight. From this we can infer mass distribution between the galaxies and us. However, the observing telescope and the atmosphere can introduce distortions too. We clump these effects here as the point spread function (PSF). These PSF distortions are color dependent and thus their effect will be different across the galaxy due to color gradients. If this effect is incorrectly accounted for then the estimated shapes and thus, the WL shear measurements can be biased.

How do we estimate CG bias?

Several different systematics can bias shear estimates. Thus, to isolate bias from color gradients (CG) only we compare the shear measured from a galaxy with color gradients to an equivalent galaxy with no color gradients. Impact from all systematics other than CG will be the same for the two galaxies, thereby giving an estimate of bias from CG only.

This study is focused on estimating the multiplicative shear bias from CG, m_CG(z), produced when the PSF size depends on wavelength, at different galaxy redshifts, z.

What is the dataset?

CG bias is estimated for three types of galaxy images:

1. Reference parametric galaxy with bulge and disk: Since most galaxies important for WL measurements are well approximated by elliptical Sersic bulge + disk profiles, we simulate galaxies with no noise like this

The reference galaxy has extreme color gradients and is not representative of all galaxies LSST will see. We use this galaxy to place limits on the extent of CG bias.

1. Galaxies from CatSim (Catalog Simulator) with a range of color gradients. These galaxies are parametric bulge + disk models with galaxy size and color distribution that LSST is expected to see.
2. HST V/I band images from AEGIS survey are redrawn as would be seen by LSST. These are real galaxy images with noise.

The download-able AEGIS catalog contains postage stamp images of isolated galaxies and photometric measurements in V/I bands. Detailed document describing how the catalog was created can be found here

How big is the bias?

For noise-free parametric galaxy simulations, the value of half the maximum span of m_CG(z) in the redshift range [0, 1.2] is < 1.5 x 10^-3 for the reference galaxies with extreme color gradients and < 10^-4 CatSim galaxies. For input AEGIS galaxies with pixel noise, the estimated bias shows a strong dependence on SNR due to contributions from effects other than CG. However, for AEGIS galaxies with HST I-band SNR > 200 the magnitude of the mean estimated bias is <0.0009, while the value of half the maximum span of m_CG(z) is < 1.5 x 10^-4. Therefore, for both the noise-free parametric galaxies and for the AEGIS galaxies with SNR > 200, the half-maximum span is less than the LSST full-depth requirement of 0.003 on the total systematic uncertainty in the redshift-dependent shear calibration m_z(z).

The values of CG bias are summarized in the table below. Refer this document explaining the results in detail.