Detection of correlated galaxy ellipticities on CFHT data: first evidence for gravitational lensing by large-scale structures

TL;DR

This paper reports the first robust detection of cosmic shear in CFHT data, confirming weak lensing by large-scale structure at 0.5–3.5 arcmin scales with a 5.5-sigma significance and an amplitude around 2.2% at 1 arcmin. Using a heterogeneous 2-deg^2 CFHT dataset, the authors implement a KSB-based PSF-correction pipeline, carefully addressing shear estimation, star–galaxy anisotropy, and instrumental systematics. The measured shear variance agrees with nonlinear predictions and ray-tracing cosmological models, mildly favoring cluster-normalized and flat $\Lambda$CDM scenarios while acknowledging redshift distribution uncertainties and limited area. They outline steps toward more precise cosmological inferences with upcoming wide-field surveys and improved PSF-correction methods, aiming to map dark matter and constrain cosmology via multiple lensing statistics.

Abstract

We report the detection of a significant (5.5 sigma) excess of correlations between galaxy ellipticities at scales ranging from 0.5 to 3.5 arc-minutes. This detection of a gravitational lensing signal by large-scale structure was made using a composite high quality imaging survey of 6300 arcmin^2 obtained at the Canada France Hawaii Telescope (CFHT) with the UH8K and CFH12K panoramic CCD cameras. The amplitude of the excess correlation is 2.2\pm 0.2 % at 1 arcmin scale, in agreement with theoretical predictions of the lensing effect induced by large-scale structure.We provide a quantitative analysis of systematics which could contribute to the signal and show that the net effect is small and can be corrected for. We show that the measured ellipticity correlations behave as expected for a gravitational shear signal. The relatively small size of our survey precludes tight constraints on cosmological models. However the data are in favor of cluster normalized cosmological models, and marginally reject Cold Dark Matter models with (Omega=0.3, sigma_8<0.6) or (Omega=1, sigma_8=1). The detection of cosmic shear demonstrates the technical feasibility of using weak lensing surveys to measure dark matter clustering and the potential for cosmological parameter measurements, in particular with upcoming wide field CCD cameras.

Figures (19)

• Figure 1: Square-root of the variance of the measured shear as a function of the radius of the top-hat window (solid line). The maximum angular scale, 3.5 arc-minutes radius, is fixed by the maximum angular scale defined by individual CCDs (7'). Error bars are computed over $1000$ random realizations of the galaxy catalogue. The other lines are theoretical predictions of the same quantity for different cosmological models in the non-linear regime (using the fitting formula in PD96): the long-dashed line corresponds to $(\Omega=1, \Lambda=0, \sigma_8=0.6)$, the dashed line to $(\Omega=0.3, \Lambda=0, \sigma_8=0.6)$, and the dot-dashed line to $(\Omega=0.3, \Lambda=0.7, \sigma_8=0.6)$.
• Figure 2: For different smoothing sizes (indicated at the top of each panel), the value of the measured signal (given by the arrow) compared to the signal measured in the randomized catalogues (histograms). This figure shows how far the signal deviates from a pure random orientation of the galaxies. Note that the distribution of $\langle \gamma^2\rangle$ is not Gaussian.
• Figure 3: The thick solid line shows the signal as plotted on Figure \ref{['9700.f1.ps']}. It was obtained with a catalogue of galaxies where galaxies closer than $10$ pixels were rejected. The three other curves show the same signal measured with different rejection criteria: the diamond-dotted line is for no rejection at all, the triangle-dashed line for galaxies closer than $5$ pixels rejected, and the square-dashed line for galaxies closer than $20$ pixels rejected. This figure illustrates that the overlapping isophotes of close galaxies tends to overestimate the shear.
• Figure 4: Average galaxy ellipticity $\langle e_\alpha\rangle$ versus the average star ellipticity $\langle e^\star_\alpha\rangle$ for both components $\alpha=1,2$. The dashed lines are obtained from the fully corrected galaxy ellipticities, as given by Eq.(\ref{['imcateq']}). The solid lines are obtained from the galaxy ellipticities corrected from the seeing, but without the anisotropy correction term $P^{\rm sm}{\bf p}$ of Eq.(\ref{['imcateq']}). Each ellipticity bin contains about $N=16000$ galaxies, and the error bars are calculated assuming Gaussian errors $\propto N$. Except for a constant tiny bias along the $e_1$ direction, the corrected galaxies are uncorrelated with the stellar ellipticity, which demonstrates that the PSF correction method works well.
• Figure 5: Average galaxy ellipticity $\langle e_1\rangle$ (solid line) and $\langle e_2\rangle$ (dashed line) as a function of the object size $r_h$. It is shown that the systematic bias of $-1\%$ along the $e_1$ component is fairly galaxy independent.