A SPECTROSCOPICALLY CONFIRMED DOUBLE SOURCE PLANE LENS SYSTEM IN THE HYPER SUPRIME-CAM SUBARU STRATEGIC PROGRAM

Strong gravitational lensing is a powerful tool to map the matter distribution around galaxies on small scales. In particular, it is possible to infer the total mass enclosed within the Einstein radius, allowing us to separate the dark and baryonic mass distribution of massive galaxies and constrain the stellar initial mass function (IMF; e.g., Treu et al. 2010). However, these measurements are complicated by various model degeneracies.

Lenses with multiple sources at distinct redshifts offer a unique opportunity to break these degeneracies, leading to accurate measurements of the dark matter distribution and IMF (e.g., Sonnenfeld et al. 2012). In addition, such double source plane (DSP) lens systems can be a useful, complementary cosmological probe (e.g., Collett et al. 2012; Collett & Auger 2014; Linder 2016), provided the mass-sheet degeneracy is broken (e.g., Falco et al. 1985; Schneider 2014). However, DSP lenses are extremely rare. Only a handful are currently known (e.g., Gavazzi et al. 2008; Tu et al. 2009; Cooray et al. 2011; Brammer et al. 2012), and spectroscopic redshifts of the second sources have not been measured for these systems due to their faintness. Ongoing/upcoming wide-area imaging surveys are expected to increase the number of such rare but fruitful systems. Here, we report the discovery of a DSP lens system from the Hyper Suprime-Cam Subaru Strategic Program (hereafter the HSC survey).

The HSC survey is the largest observing program ever approved at the Subaru 8.2 m telescope (300 nights). With a three-layered survey strategy over ∼1400 deg², it aims to address several outstanding astrophysical questions such as the nature of dark energy, galaxy evolution, and cosmic reionization. The survey is $\sim 20 \%$ complete as of this writing. While the survey is still in an early phase, the exquisite data quality has led to the exciting discovery of this DSP lens. The system, HSC J142449−005322, was serendipitously discovered at a science school for undergraduate students hosted at the National Astronomical Observatory of Japan in 2015. While visually inspecting galaxy cluster candidates, we found a likely cluster member with a clear sign of gravitational lensing. A more careful inspection revealed the multi-source nature of the system. We dub it Eye of Horus, as the system resembles it.

We describe our photometric and spectroscopic data in Section 2. We discuss properties of the lens galaxy in Section 3 and describe our lens modeling procedures in Section 4. Our main results are presented in Section 5. We adopt ${{\rm{H}}}_{0}=70\ \mathrm{km}\ {{\rm{s}}}^{-1}\ {\mathrm{Mpc}}^{-1}$, ${{\rm{\Omega }}}_{M}=0.27$, and ${{\rm{\Omega }}}_{{\rm{\Lambda }}}=0.73$. Magnitudes are AB magnitudes.

In this section, we describe the photometric and spectroscopic properties of the system. The lens is located at (R.A., decl.) = (${14}^{{\rm{h}}}{24}^{{\rm{m}}}49\buildrel{\rm{s}}\over{.} 0$, $-00^\circ 53^{\prime} 21\buildrel{\prime\prime}\over{.} 65$) (J2000). A multicolor image of Eye of Horus constructed from HSC imaging data is shown in Figure 1. The two sources with distinct colors are clearly seen around the lens: a red inner source (S1) and a blue Einstein ring (S2). These arcs are approximately 3'' in radius. Several knots with similar colors are embedded in the arcs, which are useful features to constrain our lens models (Section 4).

Zoom In Zoom Out Reset image size

2.1. Photometric Data

Eye of Horus is located in the Wide layer of the HSC survey, which aims to cover ∼1400 deg² down to $r\sim 26$. This Letter is based on the S15A internal data release. The data are processed with hscPipe v3.8.5, a version of Large Synoptic Survey Telescope pipeline (Ivezic et al. 2008; Axelrod et al. 2010) calibrated against Pan-STARRS1 (Schlafly et al. 2012; Tonry et al. 2012; Magnier et al. 2013). A summary of the HSC data on this system is given in Table 1. The exposure times are short, but reach fairly deep thanks to the superb image quality. We find that the system has also been observed by KiDS (de Jong et al. 2013) and VISTA-VIKING (Edge et al. 2013). However, these shallower data do not reveal the double source nature of the system—only the outer arc is seen in the KiDS data, and VISTA-VIKING detect only the few brightest knots. We use only the photometric data from HSC in this Letter.

Table 1.  Photometric Data

Filter	Seeing ('')	Exposure Time (s)a	$5\sigma$ Depthb
g	0.81	450	26.7
r	0.62	450	26.4
i	0.61	1600	26.4
z	0.66	1600	25.5
y	0.66	2000	24.6

Notes.

^aSome of the y-band data are taken under poor conditions. ^b $5\sigma$ depth for point sources. The depths within 15 apertures, which may be more relevant for extended sources, are shallower by ∼0.2 mag.

Download table as:  ASCIITypeset image

2.2. Spectroscopic Data

The lens galaxy has a spectroscopic redshift of ${z}_{{\rm{L}}}=0.795$ from the Sloan Digital Sky Survey (Alam et al. 2015), and we discuss its properties in Section 3. In order to determine the redshifts of the background sources, we performed follow-up spectroscopic observations with the Folded-port Infrared Echellette (FIRE; Simcoe et al. 2013) on the 6.5 m Magellan I Baade telescope. FIRE covers the wavelength range 0.82–2.51 μm at a spectral resolution of R = 6000. The observations were taken on the nights of UT 2016 February 15–16. Conditions were clear, and the seeing varied from 06 to 10. A slit width of 1'' was used to image the primary features of S1 and S2 across six separate pointings with exposure times ranging from 10 to 40 minutes. The data are reduced using a combination of the FIREHOSE pipeline and custom IDL scripts.

We successfully determine the redshifts of most of the main features of both sources indicated in Figure 1, making this the first DSP lens with spectroscopic confirmation of both source redshifts. As shown in Figure 2, we measure the redshift of the outer source to be ${z}_{{\rm{S}}2}=1.988$ based on multiple detections of the [O ii], Hβ, [O iii], and Hα lines. Interestingly, S2 has two components that are slightly offset by ${\rm{\Delta }}z\approx 0.002$ in redshift space. However, they are likely physically associated given the small difference.

Zoom In Zoom Out Reset image size

The inner source is confirmed at ${z}_{{\rm{S}}1}=1.302$ with weaker detections of the [O iii] and Hα lines. In addition to our FIRE spectra, the spectrum from SDSS also confirms the redshift of S1—feature H is included within the SDSS fiber targeting the lens galaxy and we observe clear [O ii] emission from H once the continuum is subtracted. S1 also seems to have two components as indicated by the small redshift offset of ${\rm{\Delta }}z\approx 0.002$ between H and the southern segment of the arc (S1A-S) as well as a very marginal hint of the western segment (S1A-W) being at a redshift consistent with H, suggesting that S1A-W is the counterimage of H. Further follow-up observations will reveal the detailed dynamics of each source.

The lens galaxy is a passive galaxy with no clear signs of active star formation based on the SDSS spectrum. Preliminary results of the CAMIRA algorithm (Oguri 2014) applied to the HSC survey data suggest that there is a rich cluster of galaxies at ${z}_{\mathrm{phot}}\sim 0.81$ at the location of the lens. The richness of $N\sim 50$ implies a cluster mass of ${M}_{\mathrm{vir}}\sim 7\times {10}^{14}\,{M}_{\odot }$. Given the lens galaxy's large stellar mass (estimated below) and proximity to the cluster center, it is potentially the brightest cluster galaxy. A more detailed analysis of the lens environment based on multi-object spectroscopy from the Inamori Magellan Aerial Camera and Spectrograph (IMACS; Dressler et al. 2011) will be presented in a follow-up work (K. C. Wong et al., in preparation).

Despite the superb seeing, the lens and the surrounding arcs are still blended in the HSC imaging. In order to study the lens and perform detailed analyses of the background sources, we fit the lens with a double Sérsic model and subtract it from the images using Glee (Suyu & Halkola 2010; Suyu et al. 2012). We first fit the lens in the i-band, in which the lens is detected at a high signal-to-noise ratio of ∼500 within a $1\buildrel{\prime\prime}\over{.} 5$ aperture and the seeing is good ($0\buildrel{\prime\prime}\over{.} 61;$ see Table 1). The PSF is modeled in each exposure and is evaluated at the position of the lens by coadding the PSF models. We then fit the lens in the other bands with the structural parameters such as centroid, effective radius, ellipticity, and position angle fixed to those measured in the i-band. Only the flux normalization is allowed to vary. The PSF model in each band is evaluated in the same way as the i-band. The multicolor lens-subtracted image is shown in the left panel of Figure 3. An arc-like feature (H) on the left of the lens emerges. The images show no subtraction residuals at the location of the lens, demonstrating our accurate modeling of PSF and the lens galaxy light.

Zoom In Zoom Out Reset image size

Using the five-band photometry and the known redshift of the lens galaxy, we fit its spectral energy distribution (SED) using mizuki (Tanaka 2015). Due to the strong blending, we use the double Sérsic fit to the lens to estimate the flux in each band. The fit has a very extended envelope out to large radii, but we truncate the model at a radius of $3\buildrel{\prime\prime}\over{.} 3$ (∼25 kpc at the lens redshift), which is the radius of the S2 arc. This is a somewhat arbitrary choice, but it should include most of the lens light and also represent the enclosed stellar mass while keeping the flux from the (uncertain) envelope relatively small. The SED is then fit with a suite of model templates generated by Bruzual & Charlot (2003) models assuming a Chabrier (2003) IMF and Calzetti et al. (2000) dust attenuation curve. We apply Bayesian priors on the physical properties of galaxies to keep the model parameters within realistic ranges and to effectively let the templates evolve with redshift in an observationally motivated way (for more details, see Tanaka 2015). We find that the lens galaxy is very massive—it has a stellar mass of ${6.6}_{-0.1}^{+0.7}\times {10}^{11}\,{M}_{\odot }$, making it among the most massive galaxies around this redshift (Muzzin et al. 2013). The star formation rate from the fit is very small for its stellar mass (SFR $\,=\,{0.9}_{-0.1}^{+1.3}\,{M}_{\odot }\,{\mathrm{yr}}^{-1}$), as is the dust attenuation (${\tau }_{V}={0.2}_{-0.1}^{+0.2}$). Together with the early-type morphology, the lens may be the brightest cluster galaxy. Note that the given uncertainties are just statistical uncertainties and do not account for systematic effects.

We model the system based on the data in hand using two independent codes: Glee (Suyu & Halkola 2010; Suyu et al. 2012) and Glafic (Oguri 2010). The system is complex, and our goal is to reproduce the main features of the image configuration of both sources to obtain a basic understanding of the system. More detailed lens models will be presented in a future paper (K. C. Wong et al., in preparation).

4.1. Pixelated Source Modeling

4.2. Parameterized Source Modeling

The reconstructed image configurations are shown in the middle and right panels of Figure 3. We find that both Glee and Glafic successfully reproduce the main features of the system. Both methods prefer a shallow mass profile ($\gamma \sim 1.5$, where $\rho \propto {r}^{-\gamma }$) for L, suggesting that there is a large dark matter fraction, possibly due to the cluster environment. Both models also require an additional mass component between images A and C to split this feature into two images. We have assumed that this mass component, which is undetected in the HSC imaging and FIRE spectroscopy, is at the same redshift as the lens galaxy, although it could potentially be a line-of-sight structure as well. The image splitting is unlikely to be due to S1 as the positions are not coincident. The velocity dispersion of L1, assuming an SIS profile, is ∼60–100 km s⁻¹, although the uncertainties are large and there are likely degeneracies with the centroid position relative to features A and C. Deeper and higher-resolution imaging is needed to potentially detect L1.

Both models find that S1 has a velocity dispersion of ∼230 km s⁻¹ (assuming an SIE profile) with moderate to high ellipticity. The Glee model maps feature H to S1A-W, while S1A-S maps to a segment of H just above the lens galaxy, resulting in a source that either consists of two components or is highly elongated (e.g., an edge-on disk). The slight velocity offset of H and S1A-W with S1A-S indicated by the FIRE data supports this interpretation. Unfortunately, the upper segment of H was not targeted in our FIRE observations. There are likely degeneracies among the physical properties of L and S1. More detailed mass models using improved data will explore these degeneracies and more accurately determine the model parameters.

By extrapolating the results from CFHTLS (i.e., one confirmed DSP lens from Tu et al. 2009), we expect to discover ∼10 DSP lenses in the HSC survey, of which Eye of Horus is the first. Our initial results highlight the potential wealth of information provided by DSP lenses, as well as the challenges of DSP lens modeling and the importance of high-quality data and multiple lens model analyses to understand all of the degeneracies and systematic effects. The spectroscopic confirmation of both sources and the presence of the potential substructure causing the image splitting of features A and C make Eye of Horus a valuable system for studies of cosmology and galaxy structure.

We thank Aleksi Halkola for his contributions to this project. This work is based on data collected at Subaru Telescope, which is operated by the National Astronomical Observatory of Japan, and is supported by JSPS KAKENHI grant numbers JP15K17617, JP26800093, and JP15H05892. S.H.S. gratefully acknowledges support from the Max Planck Society through the Max Planck Research Group. We thank the referee for a helpful report. The Hyper Suprime-Cam (HSC) collaboration includes the astronomical communities of Japan and Taiwan, and Princeton University. The HSC instrumentation and software were developed by the National Astronomical Observatory of Japan (NAOJ), the Kavli Institute for the Physics and Mathematics of the universe (Kavli IPMU), the University of Tokyo, the High Energy Accelerator Research Organization (KEK), the Academia Sinica Institute for Astronomy and Astrophysics in Taiwan (ASIAA), and Princeton University. Funding was contributed by the FIRST program from Japanese Cabinet Office, the Ministry of Education, Culture, Sports, Science and Technology (MEXT), the Japan Society for the Promotion of Science (JSPS), Japan Science and Technology Agency (JST), the Toray Science Foundation, NAOJ, Kavli IPMU, KEK, ASIAA, and Princeton University. This paper makes use of software developed for the Large Synoptic Survey Telescope. We thank the LSST Project for making their code freely available.¹³ The Pan-STARRS1 (PS1) Surveys have been made possible through contributions of the Institute for Astronomy, the University of Hawaii, the Pan-STARRS Project Office, the Max-Planck Society and its participating institutes, the Max Planck Institute for Astronomy and the Max Planck Institute for Extraterrestrial Physics, The Johns Hopkins University, Durham University, the University of Edinburgh, Queen's University Belfast, the Harvard-Smithsonian Center for Astrophysics, the Las Cumbres Observatory Global Telescope Network Incorporated, the National Central University of Taiwan, the Space Telescope Science Institute, the National Aeronautics and Space Administration under grant No. NNX08AR22G issued through the Planetary Science Division of the NASA Science Mission Directorate, the National Science Foundation under grant No. AST-1238877, the University of Maryland, and Eotvos Lorand University (ELTE).