MNRAS 473, 5641–5657 (2018) doi:10.1093/mnras/stx2463
Advance Access publication 2017 September 27
First results from GeMS/GSAOI for project SUNBIRD: Supernovae
UNmasked By Infra-Red Detection
E. C. Kool,1,2‹ S. Ryder,2 E. Kankare,3 S. Mattila,4 T. Reynolds,4
R. M. McDermid,1,2 M. A. Pe´rez-Torres,5 R. Herrero-Illana,6 M. Schirmer,7
A. Efstathiou,8 F. E. Bauer,9,10,11 J. Kotilainen,4,12 P. Va¨isa¨nen,13 C. Baldwin,1
C. Romero-Can˜izales10,14 and A. Alberdi5
1Department of Physics and Astronomy, Macquarie University, NSW 2109, Australia
2Australian Astronomical Observatory, 105 Delhi Rd, North Ryde, NSW 2113, Australia
3Astrophysics Research Centre, School of Mathematics and Physics, Queen’s University Belfast, Belfast BT7 1NN, UK
4Tuorla observatory, Department of Physics and Astronomy, University of Turku, Va¨isa¨la¨ntie 20, FI-21500 Piikkio¨, Finland
5Instituto de Astrofı´sica de Andalucı´a (CSIC), Glorieta de la Astronomı´a s/n, E-18080 Granada, Spain
6European Southern Observatory (ESO), Alonso de Co´rdova 3107, Vitacura, Casilla 19001, Santiago de Chile, Chile
7Gemini Observatory, Casilla 603, La Serena, Chile
8School of Sciences, European University Cyprus, Diogenes Street, Engomi, 1516 Nicosia, Cyprus
9Instituto de Astrofı´sica and Centro de Astroingenierı´a, Pontificia Universidad Cato´lica de Chile, Casilla 306, Santiago 22, Chile
10Millennium Institute of Astrophysics (MAS), Nuncio Monsen˜or So´tero Sanz 100, Providencia, Santiago, Chile
11Space Science Institute, 4750 Walnut Street, Suite 205, Boulder, Colorado 80301, USA
12Finnish Centre for Astronomy with ESO (FINCA), University of Turku, Va¨isa¨la¨ntie 20, FI-21500 Piikkio¨, Finland
13South African Astronomical Observatory, P.O. Box 9, Observatory, 7935 Cape Town, South Africa
14Nu´cleo de Astronomı´a de la Facultad de Ingenierı´a y Ciencias, Universidad Diego Portales, Av. Eje´rcito 441, Santiago, Chile
Accepted 2017 September 21. Received 2017 September 20; in original form 2017 May 26
ABSTRACT
Core collapse supernova (CCSN) rates suffer from large uncertainties as many CCSNe explod-
ing in regions of bright background emission and significant dust extinction remain unobserved.
Such a shortfall is particularly prominent in luminous infrared galaxies (LIRGs), which have
high star formation (and thus CCSN) rates and host bright and crowded nuclear regions, where
large extinctions and reduced search detection efficiency likely lead to a significant fraction
of CCSNe remaining undiscovered. We present the first results of project SUNBIRD (Super-
novae UNmasked By Infra-Red Detection), where we aim to uncover CCSNe that otherwise
would remain hidden in the complex nuclear regions of LIRGs, and in this way improve
the constraints on the fraction that is missed by optical seeing-limited surveys. We observe
in the near-infrared 2.15µm Ks-band, which is less affected by dust extinction compared to
the optical, using the multiconjugate adaptive optics imager GeMS/GSAOI on Gemini South,
allowing us to achieve a spatial resolution that lets us probe close in to the nuclear regions.
During our pilot program and subsequent first full year we have discovered three CCSNe and
one candidate with projected nuclear offsets as small as 200 pc. When compared to the total
sample of LIRG CCSNe discovered in the near-IR and optical, we show that our method is
singularly effective in uncovering CCSNe in nuclear regions and we conclude that the majority
of CCSNe exploding in LIRGs are not detected as a result of dust obscuration and poor spatial
resolution.
Key words: instrumentation: adaptive optics – surveys – supernovae: general – galaxies: star-
burst – infrared: galaxies.
E-mail: erik.kool@mq.edu.au
C© 2017 The Authors
Published by Oxford University Press on behalf of the Royal Astronomical Society
Downloaded from https://academic.oup.com/mnras/article-abstract/473/4/5641/4259573
by Turun Yliopiston Kirjasto user
on 29 March 2018
5642 E. C. Kool et al.
1 IN T RO D U C T I O N
Luminous and ultraluminous infrared galaxies (LIRGs and
ULIRGs; LIR > 1011 L and LIR > 1012 L, respectively) ex-
hibit high star formation (SF) rates, and their corresponding high
infrared (IR) luminosity is due to ultraviolet (UV) light from hot
young stars getting absorbed by dust and re-emitted in the IR. In the
local Universe the relative SF contribution of (U)LIRGs is small,
but by a redshift of ∼1 the SF in these dusty galaxies dominates
over that in normal galaxies (Magnelli et al. 2011).
Based on their LIR, starburst dominated (U)LIRGs have an ele-
vated expected core-collapse supernova (CCSN) rate of the order of
one per year (Mattila & Meikle 2001), which is ∼100 times larger
than the Milky Way CCSN rate (Adams et al. 2013). Due to the short
lifetime of their progenitors, CCSNe act as a relatively direct tracer
of the current rate of massive SF. However, even with the onset
of wide field optical SN searches this past decade (such as CRTS,
Drake et al. 2009; ASAS-SN, Shappee et al. 2014; Pan-STARRS1,
Chambers et al. 2016; SkyMapper, Scalzo et al. 2017) surprisingly
few CCSNe have been found in LIRGs.
This shortfall of CCSNe is not unique to LIRGs: Horiuchi et al.
(2011) claimed that up to half of all CCSNe beyond the local
volume, as predicted by the well-defined SF rate history, are not
observed: the so-called Supernova Rate Problem. In the very lo-
cal Universe, however, such a discrepancy is not seen (Botticella
et al. 2012; Xiao & Eldridge 2015), and Cappellaro et al. (2015)
argue that systematic errors in the SN and SF rates remain too large
to invoke a supernova rate problem in the first place. One of the
main uncertainties in the observed CCSN rate is the fraction that
is missed in obscured and clumpy galaxies, such as LIRGs. Based
on monitoring of the LIRG Arp 299, Mattila et al. (2012) placed
empirical limits on the fraction of CCSNe missed by optical surveys
as a function of redshift concluding that the optical missed-fraction
rises from 20 per cent locally to 40 per cent at redshift of ∼1. These
corrections were successfully applied to bridge the gap between
observed and predicted CCSN rates in several studies (e.g. Dahlen
et al. 2012; Melinder et al. 2012) or used as a high extinction sce-
nario for CCSN rates at high redshift (z = 2; Strolger et al. 2015),
but are based on a small SN sample.
In order to improve the constraints on the missed fraction of CC-
SNe, a larger statistical sample of CCSNe in obscured and clumpy
galaxies is required. LIRGs fit these properties well and also display
a clear deficit in CCSN discoveries. They have therefore been tar-
gets for several supernova surveys in the near-IR, where extinction
is vastly reduced compared to optical wavelengths (AK ∼ 0.1 × AV),
to uncover their CCSN population. Mannucci et al. (2003) observed
46 local LIRGs in natural seeing conditions in K′ band and detected
three CCSNe, an order of magnitude smaller than the rate estimated
from the galaxies LFIR. Cresci et al. (2007) observed 17 LIRGs
with HST/NICMOS in the F160W filter and did not find any con-
firmed SNe. Miluzio et al. (2013) observed 30 LIRGs across three
semesters in service time with HAWK-I on the Very Large Tele-
scope (VLT) in natural seeing conditions in K band and detected
five CCSNe. They claimed good agreement with the expected rate,
but assumed a large fraction (∼60–75 per cent) remained hidden in
the nuclear regions (<2 kpc) due to reduced search efficiency and
extinction.
The results of these studies demonstrate that there remains con-
siderable uncertainty in the CCSN rate from LIRGs. With a dozen
CCSN discoveries in total, the number of detections have lagged
behind the expected CCSN rate, typically attributed to limitations in
temporal coverage, lack of contrast against the extremely luminous
background and/or inadequate spatial resolution in order to resolve
the crowded and complex nuclear regions in LIRGs. Recent studies
using near-IR ground-based adaptive optics (AO) imaging to pro-
vide the necessary high spatial resolution have had more success
uncovering SNe in these conditions, with an additional five near-IR
CCSN discoveries (Mattila et al. 2007; Kankare et al. 2008, 2012).
Promisingly, several of these CCSNe have been within a few hun-
dred pc from the hosts’ nuclei, with extinctions up to 16 magnitudes
in V band.
Building on these early results, we commenced in 2015 the Su-
pernova UNmasked By Infra-Red Detection (SUNBIRD) project: a
systematic search for CCSNe in a sample of LIRGs within 120 Mpc
using laser guide star AO (LGSAO) imaging with the Gemini South
Adaptive Optics Imager (GSAOI, McGregor et al. 2004; Carrasco
et al. 2012) with the Gemini Multi-Conjugate Adaptive Optics Sys-
tem (GeMS, Rigaut et al. 2014; Neichel et al. 2014a) on the Gemini
South telescope. The SUNBIRD project aims to characterize the
population of CCSNe in the dusty and crowded star forming re-
gions of LIRGs and in this way improve the constraints on the
fraction of CCSNe missed due to dust obscuration and/or nuclear
vicinity. In this first paper we introduce this ongoing survey and
report on the first results of project SUNBIRD: three LIRG CCSN
discoveries and one CCSN candidate.
During preparation of this manuscript, project SUNBIRD has
extended coverage to the Northern hemisphere through use of the
Keck telescope. A description of the Keck campaign and a detailed
CCSN rate analysis of the complete sample of LIRGs in SUNBIRD
will appear in a forthcoming paper.
Section 2 of this paper contains a description of the survey, in-
cluding the galaxy sample and observing strategy. In Section 3
we describe the data reduction and analysis. We present the SUN-
BIRD supernova detections made so far in Section 4, followed by
Sections 5 and 6 where we investigate the supernova types and
explore the impact of these new discoveries on the total number
of CCSNe discovered in LIRGs, respectively. Finally in Section 7
we draw our conclusions. Throughout this paper we assume H0 =
70 km s−1 Mpc−1,  = 0.7 and M = 0.3.
2 PRO J E C T S U N B I R D
2.1 Galaxy sample
The sample of LIRGs observed in SUNBIRD was selected from the
IRAS Revised Bright Galaxy Sample (RBGS; Sanders et al. 2003).
The main constraints to the sample selection originated from the
instrument we used, GeMS/GSAOI on the Gemini South telescope,
which requires guide stars of sufficient brightness and vicinity for
the AO correction (see Section 2.2). Additionally we limited the
sample to galaxies that are closer than 120 Mpc (z = 0.027) in order
to be able to resolve the central regions as close to the nucleus
as possible. At this distance a typical AO corrected full width at
half-maximum (FWHM) of 0.1 arcsec corresponds to ∼60 pc. We
included targets with IR luminosities L(8–1000µm) in RBGS of
log(LIR) > 10.9, see Table 1. Finally, we omit LIRGs where a
significant active galactic nucleus (AGN) contamination to the IR
luminosity could be expected, and thus exclude targets with ‘warm’
IRAS colours, requiring f25/f60 < 0.2 (e.g. Farrah et al. 2005). The
only exception is IRAS 08355-4944 with f25/f60 = 0.24 which is
included in the sample based on SED fitting results, where it was
shown to have a high SF rate of ∼85 M yr−1 (Dopita et al. 2011),
typical for a SF dominated LIRG.
MNRAS 473, 5641–5657 (2018)Downloaded from https://academic.oup.com/mnras/article-abstract/473/4/5641/4259573
by Turun Yliopiston Kirjasto user
on 29 March 2018
First results from GeMS/GSAOI for SUNBIRD 5643
Table 1. SUNBIRD GeMS/GSAOI LIRG sample. Distances are from the NASA/IPAC Extragalactic Database (NED1), Virgo/GA
corrected. CCSN rates are based on the empirical relation with LIR from Mattila & Meikle (2001), unless otherwise indicated: CCSN
rate based on SED fits from Herrero-Illana et al. (2017), denoted by a, or this work, denoted by b. Values of log LIR are from Sanders
et al. (2003), adjusted for updated distances. Final column shows number of epochs obtained with GeMS/GSAOI.
LIRG RA Dec. Distance log LIR rCCSN Epochs
(J2000) (J2000) (Mpc) (L) (yr−1) #
NGC 1204 03 04 40.5 −12 20 26 64 10.96 0.25 1
ESO 491-G020 07 09 47.0 −27 34 10 43 10.97 0.25 1
MCG +02-20-003 07 35 42.5 +11 42 36 72 11.13 0.37 1
IRAS 08355-4944 08 37 02.3 −49 54 32 115 11.62 1.12 2
NGC 3110 10 04 02.7 −06 28 35 79 11.36 0.14b 4
ESO 264-G036 10 43 07.0 −46 12 43 99 11.34 0.59 4
ESO 264-G057 10 59 02.4 −43 26 33 82 11.15 0.38 1
NGC 3508 11 03 00.1 −16 17 23 61 10.97 0.25 2
ESO 440-IG058 12 06 53.0 −31 57 08 111 11.45 0.51a 4
ESO 267-G030 12 14 12.6 −47 13 37 96 11.26 0.49 5
NGC 4575 12 37 52.1 −40 32 20 63 11.03 0.29 2
IRAS 17138-1017 17 16 36.3 −10 20 40 83 11.49 0.75a 5
IRAS 18293-3413 18 32 40.2 −34 11 26 85 11.74 1.97a 4
Note. 1The NASA/IPAC Extragalactic Database (NED) is operated by the Jet Propulsion Laboratory, California Institute of Technology,
under contract with the National Aeronautics and Space Administration.
2.2 Observing strategy
The near-IR observations were obtained with GeMS/GSAOI on the
Gemini South telescope. GSAOI is a near-IR AO imaging camera
fed by GeMS and records images in a 85 arcsec × 85 arcsec field-of-
view (FOV) with a pixel scale of 0.0197 arcsec pixel−1, delivering
close to diffraction limited images between 0.9 and 2.4 µm. An
optimal uniform AO correction across the FOV of GeMS requires
three natural guide stars (NGS) in addition to the five-point sodium
laser guide star (LGS) pattern. The minimum requirement for AO
correction at the time of the observations was at least one NGS of
sufficient brightness (mR < 15.5 mag) available within the 1 arcmin
patrol field of the wave front sensor probes and one on-detector
guide window star (mH < 13.5 mag) within the 40 arcsec FOV of
any of the four GSAOI detectors at all dither positions.
The SN search was conducted in Ks band, as this is where, com-
pared to J and H, AO performs best and extinction due to dust is
lowest. Each target was imaged with a nine step dither pattern for
120 s at each position with a step size large enough (>5 arcsec) to
cover the gaps between the detectors. The targets were typically
centred on one of GSAOI’s four arrays, with orientation depending
on the galaxy and the locations of the NGS. Employing the efficient
cadence strategy from Mattila & Meikle (2001) for near-IR CCSN
searches, we aimed to observe each galaxy twice each semester.
In practice we achieved this cadence for half of the sample while
the remainder of the sample galaxies was observed less frequently,
due to seasonal weather variations, sodium layer return, and inter-
ruptions due to aircraft and satellite avoidance (see Table 1). If a
night did not allow for coverage of all observable targets, priority
was given to galaxies with a high expected SN rate and those for
which at least one GeMS/GSAOI epoch was already available. Ex-
pected SN rates were based on the empirical relation from (Mattila
& Meikle 2001)
rSN = 2.7 × 10−12 × LIR/L yr−1 (1)
The targets in our sample with just one epoch were checked
for SNe against archival high-resolution VLT/NACO (Nasmyth
Adaptive Optics System Near-Infrared Imager and Spectrograph,
0.055 arcsec pixel−1; Lenzen et al. 2003; Rousset et al. 2003) AO
images, obtained by members of the SUNBIRD collaboration as
part of a predecessor program (Randriamanakoto et al. 2013) and
available for the whole sample.
Our total sample of LIRGs covered with GeMS/GSAOI so far
consists of 13 galaxies up to a distance of 115 Mpc. Even though
for some LIRGs there were only one or two NGS available and
AO correction was not optimal, across our full data set a typical
point-spread function (PSF) of ∼0.07 arcsec – 0.12 arcsec FWHM
was achieved.
2.3 Multiwavelength follow up
Following a potential SN detection in Ks, the source was first
checked for proper motion between exposures to exclude a passing
minor planet.1 Follow up with GeMS/GSAOI in H and J was done
as soon as possible, which due to observing constraints, typically
occurred in the next GeMS/GSAOI observing window. As these ob-
serving windows were two to three months apart, rapid follow up of
the SN candidate in the near-IR/optical was done with other instru-
ments: in JHK with NACO on the VLT, or contemporaneously with
the Nordic Optical Telescope (NOT, Djupvik & Andersen 2010)
in r′ and i′ band with ALFOSC2 (Andalucia Faint Object Spectro-
graph and Camera, 0.19 arcsec pixel−1) and in JHK with NOTCam
(Nordic Optical Telescope near-infrared Camera and spectrograph,
0.234 arcsec pixel−1).
In addition to near-IR and optical imaging, two SN candidates
were observed at radio wavelengths with the Karl G. Jansky Very
Large Array (JVLA). A detection would provide important infor-
mation about the nature of the SN and rule out a Type Ia SN,
as even the most nearby Type Ia have yet to be detected at radio
wavelengths (e.g. Hancock, Gaensler & Murphy 2011; Perez-Torres
et al. 2015; Chomiuk et al. 2016). The intrinsic rate of Type Ia SNe
in LIRGS is estimated to be ∼5 per cent of that of CCSNe (Mattila
1 http://www.minorplanetcenter.net/cgi-bin/checkmp.cgi
2 The data presented here were obtained in part with ALFOSC, which is
provided by the Instituto de Astrofisica de Andalucia (IAA–CSIC) under a
joint agreement with the University of Copenhagen and NOTSA.
MNRAS 473, 5641–5657 (2018)Downloaded from https://academic.oup.com/mnras/article-abstract/473/4/5641/4259573
by Turun Yliopiston Kirjasto user
on 29 March 2018
5644 E. C. Kool et al.
et al. 2007). As a final step, when possible near-IR spectroscopic
coverage was obtained, under natural seeing conditions using the
cross-dispersed mode of the Gemini Near Infra-Red Spectrograph
(GNIRS) on Gemini North (Elias et al. 2006a,b). The follow up
observations are described in more detail in Section 4.
3 DATA R E D U C T I O N A N D A NA LY S I S
M E T H O D S
3.1 GeMS/GSAOI data reduction
The GeMS/GSAOI data were reduced using THELI3 (Erben
et al. 2005; Schirmer 2013), generally following the proce-
dures described in Schirmer (2013) and Schirmer et al. (2015).
GeMS/GSAOI exposures suffer from a distortion pattern with a
static component introduced by GeMS, and variable distortion com-
ponents depending on NGS configurations, position angle and ele-
vation (Neichel et al. 2014b; Schirmer et al. 2015). THELI uses Scamp
(Bertin 2006) for astrometric calibration and distortion correction
of individual exposures prior to the final co-addition, based on refer-
ence catalogues of point sources measured in distortion-free images
of the same field, such as from HAWK-I on the VLT or VIRCAM
on the VISTA telescope. In this way an optimal data quality across
the FOV is obtained. Optimisations or adaptations of the proce-
dures mentioned above are described in Appendix A. The solutions
are specific to this project, but applicable to any GeMS/GSAOI
observations with a low number source density.
3.2 NOT and NACO data reduction
The near-IR NOTCam instrument data were reduced with a slightly
modified version of the external NOTCAM package4 (v. 2.5) within
IRAF.5 The reduction steps included flat-field correction, distortion
correction, sky subtraction and stacking of the individual expo-
sures for increased signal-to-noise ratio. The optical r′- and i′-band
ALFOSC images were reduced using the QUBA pipeline (Valenti
et al. 2011), including bias subtraction and flat-field correction.
The near-IR data taken with VLT/NACO were reduced using
the NACO pipeline, which is based on the ESO COMMON PIPELINE
LIBRARY(CPL)6 The jittered on-source images were flat-field cor-
rected, then median-combined to create a sky frame. Bad pixels
were removed and the sky was subtracted from the individual im-
ages. The images were then stacked using a 2D cross-correlation
routine.
3.3 GNIRS near-IR spectroscopy data reduction
We used the cross-dispersed spectroscopy mode, providing a com-
plete spectrum within 0.8–2.5µm at an instrumental resolution of
R ∼ 1700. The data were reduced using version 2.6 of the XDGNIRS
pipeline.7 Briefly, the spectra were cleaned of pattern noise caused
3 https://www.astro.uni-bonn.de/theli/
4 http://www.not.iac.es/instruments/notcam/guide/observe.html
5 IRAF is distributed by the National Optical Astronomy Observatory, which
is operated by the Association of Universities for Research in Astronomy
(AURA) under cooperative agreement with the National Science Foundation
(Tody 1993).
6 http://www.eso.org/sci/software/cpl/
7 http://drforum.gemini.edu/topic/gnirs-xd-reduction-script/
by the detector controller, using the PYTHON code from the Gem-
ini website.8 Radiation events from the radioactive lens coatings
on the GNIRS short camera were identified and interpolated over
using IRAF’s fixpix task. Files were divided by a combined master
flat-field created from a combination of quartz-halogen and IR flats
taken after each science observation. Subtraction of a sky frame
removed night sky emission lines as well as other static artefacts
in the detector, such as stable hot pixels. The orders were rectified
using daytime pinhole flats, wavelength calibrated using an argon
arc frame and 1D spectra were extracted from each order using IRAF
task apall. Telluric correction and flux calibration were done using
an A-type star which was observed immediately before or after each
science object. Each order of the science target was multiplied by
a blackbody spectrum of the telluric star’s effective temperature,
scaled to the K-band flux of the standard star for an approximate
2MASS flux calibration, and the orders were joined together using
the IRAF task odcombine.
3.4 Image subtraction
Subtraction of different epochs was done using a slightly modified
(to accept manual stamp selection) version of image subtraction
package ISIS 2.2 (Alard & Lupton 1998; Alard 2000), where the
software matches the PSF as well as flux and background levels of
a previously aligned pair of images by deriving an optimal convo-
lution kernel based on a selection of small windows (or ‘stamps’)
around objects with high signal-to-noise. See Figs 1–3 for the result-
ing subtractions. In the case of IRAS 18293-3413 the subtraction
process resulted in considerable residuals at locations of high signal-
to-noise, such as the nucleus and other bright compact sources. This
was a result of having to prioritise an optimal subtraction for the
central regions, as the SN is located very close to the nucleus. AO
is optimized for point sources, but for a bright nuclear region with
some intrinsic morphology the PSF is likely to vary between epochs.
As such, the extraction of a SN signal in a location with such a steep
background gradient is non-trivial, because any deviations from a
perfect PSF match will produce large residuals. Despite our best
efforts it was not possible to obtain a uniform subtraction across the
full image. The smoothest subtraction of the nucleus was obtained
by selecting a large number (>10) of small stamps near and in the
galaxy to map the PSF around the SN location as well as possible,
but this resulted in the residual patterns visible in the subtraction at
the locations of bright compact objects in the field.
3.5 Photometry
The photometry of the objects was measured using the SNOOPY9
package in IRAF, where a PSF is fitted to the SN residual in the sub-
tracted image. The PSF was derived from 10 isolated field stars in
the FOV of the one image that was not convolved during subtraction
(i.e. the image with poorest image quality). SNOOPY removes a simple
background estimate from the region surrounding the SNe (exclud-
ing the innermost region around the object) during PSF fitting, but as
the SN detections are generally located in nuclear regions with large
and variable background signal, the PSF fitting is performed on the
8 http://www.gemini.edu/sciops/instruments/gnirs/data-format-and
-reduction/cleanir-removing-electronic-pattern-0
9 SNOOPY, originally presented in Patat (1996), has been implemented in IRAF
by E. Cappellaro. The package is based on daophot, but optimised for SN
magnitude measurements.
MNRAS 473, 5641–5657 (2018)Downloaded from https://academic.oup.com/mnras/article-abstract/473/4/5641/4259573
by Turun Yliopiston Kirjasto user
on 29 March 2018
First results from GeMS/GSAOI for SUNBIRD 5645
Figure 1. SN 2013if at RA = 18h32m41.s10 and Decl. = −34◦11′27.′′24 in IRAS 18293-3413 with GeMS/GSAOI. From left to right with linear scaling:
reference image (2015 June), discovery image (2013 April) and the image subtraction.
SN residual in the subtracted image, where the complex background
has already been removed. The photometry was calibrated against
five 2MASS stars in the FOV, when available. Systematic errors in
the local background subtraction were estimated by simulating and
PSF fitting nine artificial stars in a three by three grid pattern around
the SN position in the subtracted image. These typically dominated
the photometric errors.
The FOV of NGC 3110 did not offer any suitable catalogue
reference stars and only one clear isolated field star for PSF fitting
purposes. This field star acted as the PSF model and photometric
reference for SN 2015ca after the field star’s JHKs magnitudes
were determined from NOT images calibrated against five 2MASS
sources.
In case of a non-detection, a 5σ upper limit was determined by
simulating stars of decreasing brightness at the SN location using
the task mkobjects in IRAF package artdata, prior to subtraction. The
signal-to-noise was based on the aperture flux of the residual of the
simulated star compared to 80 empty positions in the field around
the SN location using IRAF task phot.
3.6 Light-curve template fitting
To determine the type of the discovered SNe, the observed light-
curve data points were fitted using (reduced) χ2 minimization to
three prototypical template light curves representing Type IIP, IIn
and stripped envelope Type IIb and Ib/c SNe, similarly to Kankare
et al. (2008, 2014). The three prototypical CCSNe templates were
chosen on the basis of being well characterized and having a well-
sampled light curve in the near-IR covering the different stages of
SN evolution during at least the first 150 d since explosion. It should
be noted that there is non-negligible diversity in SNe even within a
given type, and thus the sources detected here could differ from the
templates. The templates we used (see Figs 6, 7 and 8) are given
below.
(i) Type IIP: a type IIP template fit was carried out based on the
photometric evolution of SN 1999em. UBVRI light curves of SN
1999em (Leonard et al. 2002) were transferred into ugri using the
conversions of Jester et al. (2005). JHK light curves were obtained
from Krisciunas et al. (2009). The distance modulus, total V-band
line-of-sight extinction and explosion date of the used template SNe
were adopted for the analysis from the literature. For SN 1999em,
μ = 30.34 mag, te = 2451 475.0 in JD and AV = 0.34 mag were
reported by Krisciunas et al. (2009).
(ii) Type IIn: a type IIn template fit was carried out using the pub-
lished light curves of SN 1998S (Fassia et al. 2000; Liu et al. 2000;
Mattila & Meikle 2001). Similar to SN 1999em, the Johnson–
Cousins light curves were converted into the SDSS system using
the transformations of Jester et al. (2005). For SN 1998S, μ =
31.15 mag, te = 2450 872.5 in JD and AV = 0.68 mag were adopted
from Fassia et al. (2000).
(iii) Type IIb/Ib/Ic: well-sampled multiband light curves of
Type IIb SN 2011dh (Ergon et al. 2014, 2015) were used as a
general template for stripped-envelope SNe, since the photometric
evolution of Type IIb SNe appears to be fairly similar to that of
Type Ib/c SNe (e.g. Arcavi et al. 2012). Ergon et al. (2014) finds for
SN 2011dh values of μ = 29.46 mag, te = 2455 713.0 in JD and AV
= 0.22 mag.
All the bands are fitted simultaneously with three free parameters:
the line-of-sight extinction AV, time t0 between explosion date and
discovery and a fixed constant C applied to all bands, representing
the intrinsic magnitude difference between SNe. Upper limits are
used to constrain the template fits where necessary. Galactic line-
of-sight extinction values are adopted from NED based on the dust
maps of Schlafly & Finkbeiner (2011) and are fixed in the fit. For
host galaxy and Galactic extinction, the Cardelli, Clayton & Mathis
(1989) extinction law was used.
4 O BSERVATI ONS
4.1 SN 2013if in IRAS 18293-3413
SN 2013if (Kankare et al. 2017) in IRAS 18293-3413 was dis-
covered with GeMS/GSAOI on 2013 April 21, see Fig. 1. Sub-
traction of a Ks-band image taken with NACO on the VLT on
2004 September 13 (Mattila et al. 2007) showed a positive residual
0.2 arcsec North and 0.4 arcsec West (200 pc projected distance)
from the nucleus. WCS matching in THELI with a catalogue of 180
sources extracted from a VISTA image (VISTA Hemisphere Survey
or VHS10) yielded RA = 18h32m41.s10 and Decl. = −34◦11′27.′′24,
with 0.03 arcsec and 0.03 arcsec uncertainty in RA and Decl., re-
spectively. SN positions in this paper were determined using cen-
troiding in IRAF in the subtracted images to avoid the effects of
strong background. Follow-up observations were made with NACO
10 http://www.vista-vhs.org/
MNRAS 473, 5641–5657 (2018)Downloaded from https://academic.oup.com/mnras/article-abstract/473/4/5641/4259573
by Turun Yliopiston Kirjasto user
on 29 March 2018
5646 E. C. Kool et al.
Figure 2. SN 2015ca and AT 2015cf in NGC 3110 with GeMS/GSAOI, at RA = 10h04m01.s57 and Decl. = −06◦28′25.′′48 and RA = 10h04m01.s53 and
Decl. = −06◦28′25.′′84, respectively. Top row, with linear scaling, shows the reference image (2016 February) and discovery image (2015 March). Bottom row
shows the full image subtraction and zoomed in around SN 2015ca, which shows AT 2015cf visible to the south-west.
on 2013 May 8 in Ks and H, and with GeMS/GSAOI in Ks on 2013
May 24 and in Ks, H and J on 2013 June 11. SN 2013if was detected
in all the follow up images with the exception of the final J-band
image, see Table 2. Supernova free comparison GeMS/GSAOI im-
ages required for optimal image subtraction were obtained in Ks, H
and J on 2015 June 2.
4.2 SN 2015ca in NGC 3110
The discovery image of SN 2015ca (Kool et al. 2016) in NGC
3110 was observed on 2015 March 11; see Fig. 2. As a reference
image a NACO Ks-band image from 2010 December 28 was used
(Randriamanakoto et al. 2013). The subtracted image revealed a
point source along the northern spiral arm of the galaxy 4.3 arcsec
North and 8.0 arcsec West from the nucleus, corresponding to a
projected distance of ∼3.5 kpc. WCS matching in THELI with a cat-
alogue of >300 sources extracted from a HAWK-I image (Miluzio
et al. 2013) yielded RA = 10h04m01.s57 and Decl. = −06◦28′25.′′48,
with 0.03 arcsec and 0.04 arcsec uncertainty in RA and Decl., re-
spectively. Follow up observations with the NOT were carried out
on 2015 March 27 in r′ and i′ and on April 5 in Ks, H and J, and
with GeMS/GSAOI in Ks, H and J on 2015 May 29 and 31 and in
Ks, H and J on 2016 February 19. SN 2015ca was detected in all
near-IR bands in 2015 April and May, but was not detected in any
optical bands; see Table 2. It had faded below our detection limits
in 2016 February.
4.3 SN 2015cb in IRAS 17138-1017
SN 2015cb (Kool et al. 2017a) in IRAS 17138-1017 was discov-
ered with GeMS/GSAOI on 2015 March 6; see Fig. 3. Subtraction
of a GeMS/GSAOI image from 2013 March 22 revealed a residual
MNRAS 473, 5641–5657 (2018)Downloaded from https://academic.oup.com/mnras/article-abstract/473/4/5641/4259573
by Turun Yliopiston Kirjasto user
on 29 March 2018
First results from GeMS/GSAOI for SUNBIRD 5647
Table 2. Supernova near-IR and optical photometry.
SN 2013if (IRAS 18293-3413)
2012 September 14.1 NACO Ks >19.2
2013 April 21.3 GeMS/GSAOI Ks 18.53 ± 0.09
2013 May 8.4 NACO Ks 18.61 ± 0.12
2013 May 8.4 NACO H 18.93 ± 0.21
2013 May 24.3 GeMS/GSAOI Ks 19.22 ± 0.11
2013 June 11.1 GeMS/GSAOI Ks 19.12 ± 0.11
2013 June 11.1 GeMS/GSAOI H 19.03 ± 0.16
2013 June 11.1 GeMS/GSAOI J >18.7
SN 2015ca (NGC 3110)
2015 March 11.1 GeMS/GSAOI Ks 18.73 ± 0.11
2015 March 27.8 NOT r′ >20.0
2015 March 27.9 NOT i′ >21.0
2015 April 5.9 NOT Ks 17.91 ± 0.22
2015 April 6.0 NOT H 18.63 ± 0.19
2015 April 6.0 NOT J 19.39 ± 0.21
2015 May 29.9 GeMS/GSAOI Ks 20.22 ± 0.12
2015 May 31.9 GeMS/GSAOI H 20.67 ± 0.09
2015 May 31.9 GeMS/GSAOI J 21.25 ± 0.08
2016 February 19.2 GeMS/GSAOI Ks >21.5
2016 February 19.3 GeMS/GSAOI H >22.0
2016 February 19.4 GeMS/GSAOI J >22.8
SN 2015cb (IRAS 17138-1017)
2015 March 6.3 GeMS/GSAOI Ks 16.56 ± 0.09
2015 March 17.2 NOT r′ >22.5
2015 March 17.3 NOT i′ 21.10 ± 0.12
2015 April 6.2 NOT Ks 17.12 ± 0.22
2015 April 6.3 NOT H 17.06 ± 0.10
2015 April 6.3 NOT J 18.34 ± 0.18
2015 May 2.3 NOT J 18.71 ± 0.57
2015 June 1.2 GeMS/GSAOI Ks 18.97 ± 0.10
AT 2015cf (NGC 3110)
2015 March 11.1 GeMS/GSAOI Ks 20.97 ± 0.13
2015 May 29.9 GeMS/GSAOI Ks 21.4 ± 0.2
2015 May 31.9 GeMS/GSAOI H >22.7
2015 May 31.9 GeMS/GSAOI J >22.8
point source 1.4 arcsec North and 0.6 arcsec East (600 pc projected
distance) from the nucleus. WCS matching in THELI with a cata-
logue of 76 sources extracted from a VISTA image yielded RA
= 17h16m35.s84 and Decl. = −10◦20′37.′′48, with 0.04 arcsec and
0.04 arcsec uncertainty in RA and Decl., respectively. Follow up
observations with the NOT were carried out on 2015 March 17 in
optical (in i′ and r′ band, FWHM ∼1.1 arcsec), on April 6 in Ks,
H and J (FWHM ∼1.0 arcsec) and with GeMS/GSAOI in Ks on
2015 June 1. SN 2015cb was detected in all follow up observations
except in r′; see Table 2.
4.4 AT 2015cf in NGC 3110
The discovery image of SN 2015ca in NGC 3110 from 2015 March
11 showed a second transient source at RA = 10h04m01.s53 and
Decl. = −06◦28′25.′′84; see Fig. 2. AT 2015cf (Kool et al. 2017b)
is just 0.6 arcsec S and 0.4 arcsec W of SN 2015ca, as shown in the
zoomed panel in Fig. 2. The PSF of this source matches well with
that of SN 2015ca and field stars in the image. The magnitude for
this source at this epoch, bootstrapped off of SN 2015ca, is 20.97
± 0.13. It is not visible in any of the NOT epochs, which is not
surprising as with a mere 0.7 arcsec separation from SN 2015ca it
was likely blended with it in the subtraction. The GeMS/GSAOI Ks
image from 2015 May 29 does show a residual at the same position
with a magnitude of 21.4 ± 0.2 confirming that it is in fact a real
transient. The H and J observations from the same epoch did not
show the source, likely as a result of poorer image quality and/or
due to significant extinction. In 2016 February it was not visible in
any band.
4.5 Radio observations
4.5.1 VLA observations
We observed IRAS 17138-1017 and NGC 3110 with the VLA on
2015 April 8–9 under program 15A-471 (PI: Pe´rez-Torres), while
the VLA was in B-configuration. We observed IRAS 17138-1017
at K band (centred at 22 GHz) with a total bandwidth of 8 GHz, and
Figure 3. SN 2015cb at RA = 17h16m35.s84 and Decl. = −10◦20′37.′′48 in IRAS 17138-1017 with GeMS/GSAOI. From left to right, with linear scaling:
reference image (2013 March), discovery image (2015 March) and image subtraction.
MNRAS 473, 5641–5657 (2018)Downloaded from https://academic.oup.com/mnras/article-abstract/473/4/5641/4259573
by Turun Yliopiston Kirjasto user
on 29 March 2018
5648 E. C. Kool et al.
Figure 4. K-band (22 GHz) VLA contours overlaid on GeMS/GSAOI de-
tection image of SN 2015cb, with the SN indicated by tick marks.
Figure 5. X-band (10 GHz) VLA contours overlaid on GeMS/GSAOI de-
tection image of SN 2015ca and AT 2015cf. SN 2015ca is indicated by tick
marks, AT 2015cf is separated by just 0.7 arcsec.
NGC 3110 at X band (10 GHz), with 4 GHz of bandwidth, in both
cases using full polarization.
We used the bright quasar 3C286 for flux and bandpass calibra-
tion, and J0943-0819 and J1733-1304 to calibrate the phases of
IRAS 17138-1017 and NGC 3110, respectively. We performed a
standard data reduction using the Common Astronomy Software
Applications package (CASA; McMullin et al. 2007). We imaged
our data sets using multifrequency synthesis (MFS) with natu-
ral weighting, yielding a synthesized beam size and an rms of
0.56 arcsec × 0.30 arcsec and 9µJy beam−1 for IRAS 17138-1017
and 0.96 arcsec × 0.66 arcsec and 15µJy beam−1 for NGC 3110,
respectively.
Neither observation showed a local maximum coincident with
the SN position. In the case of SN 2015cb in IRAS 17138-1017,
any point source was most likely blended with the significant back-
ground signal in the central regions of the LIRG; see Fig. 4. SN
2015ca and AT 2015cf in NGC 3110 also coincide with local radio
emission potentially masking possible source detections; see Fig. 5.
For further details on the radio limits, see Section 5.
4.5.2 eEVN observation
We also observed the region around SN 2015ca and AT 2015cf with
the electronic European VLBI Network (eEVN) on 2016 May 10.
We used the eEVN at an observing frequency of 5.0 GHz, using
an array of nine telescopes for about 2 hr, including overheads for
calibration purposes and slew time. We observed the transients
phase-referenced to the nearby source J0959-0828, using a typical
duty cycle of four minutes. We used the strong source 3C273B as
fringe finder and bandpass calibrator. All the data were correlated
at the EVN MkIV data processor of the Joint Institute for VLBI in
Europe (JIVE, the Netherlands), using an averaging time of 2 s.
We used Astronomical Image Processing System (AIPS) for cal-
ibration, data inspection and flagging of our eEVN data using stan-
dard procedures. We then imaged a FOV of 1 arcsec × 1 arcsec
centred at RA = 10:04:01.572 and DEC = −06:28:25.480, and
applied standard imaging procedures using AIPS, without aver-
aging the data either in time or frequency, to prevent time- and
band- width smearing of the images. We did not detect any signal
above 101 µJy/b (=5σ ) in the field surrounding SN 2015ca. Several
sources at the 3σ level were detected, but repeated imaging with
different cleaning schemes showed that these were spurious detec-
tions, implying that there was no evidence of radio emission from
SN 2015ca. AT 2015cf was not covered in the 1 arcsec × 1 arcsec
FOV.
4.6 Near-IR spectroscopic observations
We obtained near-IR spectroscopy of SN 2015ca and SN 2015cb
through Director’s Discretionary Time with GNIRS on the Gemini
North telescope on 2015 May 23. These observations were obtained
in natural seeing conditions ∼70 d after discovery. We opted for
near-IR spectroscopy as the SNe were located in crowded regions
and expected to be significantly dust extincted. Image quality in
the near-IR is typically better and less affected by dust extinction
than the optical. The SNe were too faint to acquire directly, so a
blind offset from a bright star was required. There was no clear
indication of a point source in the slit, and the spectra did not show
any emission lines associated with CCSNe. As such the obtained
spectra could not be used to constrain the SN types of SN 2015ca
or SN 2015cb.
5 A NA LY SIS
5.1 SN 2013if
The best template fit for SN 2013if with a χ˜2 of 4.7 is a Type IIP
caught in the tail phase (see Fig. 6 and Table 3), although the other
templates fit the data with comparable values of χ˜2. A Type IIP
caught in the tail phase is deemed most likely as, in addition to the
fitting results, it is the only fit where no magnitude shift C is required
and the SN discovery is not before or at the maximum of the light
curve. This means it does not require a sub-luminous supernova as
opposed to the Type IIn fit with similar χ˜2, and it does not require
discovery during the very short-lived phase around maximum light.
The fit of the tail phase Type IIP is along a linearly extrapolated
tail with a very slow colour evolution, so the fitting parameter time t0
between explosion and detection is poorly constrained by the data.
However a NACO image from 2012 September 14, 219 d prior to
discovery, does not show a detection with an upper limit of 19.2
in Ks, allowing for the constraints on t0. Regardless of SN type,
it is noteworthy that all four template fits are best fitted with an
extinction AV of 0. This is surprising as the SN is very close to the
MNRAS 473, 5641–5657 (2018)Downloaded from https://academic.oup.com/mnras/article-abstract/473/4/5641/4259573
by Turun Yliopiston Kirjasto user
on 29 March 2018
First results from GeMS/GSAOI for SUNBIRD 5649
Figure 6. Template light-curve fits for SN 2013if in IRAS 18293-3413 are shown: Type IIP fitted in the plateau and tail phase, Type IIn and Type IIb/Ib/Ic.
nucleus where significant extinction would be expected. Negligible
extinction suggests we are observing the SN in the foreground of
the host’s nuclear regions.
5.2 SN 2015ca
SN 2015ca is best fitted by the Type IIP template, with χ˜2 = 3.8, but
the stripped envelope template fits the data too with χ˜2 = 5.0; see
Fig. 7 and Table 3. In both cases the fit requires moderate extinction
(AV = 3.4+1.0−1.7 and AV = 2.8+0.3−0.4, respectively) and a magnitude
shift of ∼1.5. However neither template fits well the brightest epoch
from 2015 April 6 with the NOT. The Type IIn template is poorly
fitted by the data and seems inconsistent with the JHK upper limits
of the final epoch. We also fitted a Type Ia light curve to the data
based on the light curve of SN 2011fe (Matheson et al. 2012; Zhang
et al. 2016) because SN 2015ca exploded in an isolated location
seemingly far from recent SF. With a χ˜2 = 17.9 the Type Ia fit was
clearly inferior to the core collapse scenarios. The Type IIn, stripped
envelope and Type Ia fits have been forced to take into account the i′-
band limit from the NOT, i.e. the i′-band curve is basically matching
with the i-band limit. The Type IIP fit was consistent with the limit
based on the detections.
The follow up at radio wavelengths with the VLA and eEVN did
not show any detections of SN 2015ca. For the two likely scenar-
ios the radio luminosity for Type IIP SNe typically peaks between
∼20–70 d post-explosion and are much fainter than stripped enve-
lope SNe, which peak between 10–150 d post-explosion (Romero-
Can˜izales et al. 2014). In the case of the IIP fit, the VLA non-
detection would have been 86+14−5 days post-explosion, meaning the
relatively faint radio signature of the SN would already have been on
the decline. For the stripped envelope fit, the observation occurred
44+6−2 days post-explosion, which coincides more closely with ra-
dio peak luminosities for the type. For instance for SN 2011dh,
the prototypical SN used for the stripped envelope template, the
observation would have been at peak (∼40 d) with a luminosity
forty times brighter than SN 1999em, the SN used for the Type IIP
template (∼8 × 1026 erg s−1 Hz−1 versus ∼2 × 1025 erg s−1 Hz−1,
respectively). However, as the SN is coincident with strong host
galaxy signal, it is not possible to exclude certain SN types defini-
tively based on the VLA observation.
The eEVN non-detection of SN 2015ca at 5 GHz with a 3σ upper
limit of 60 µJy beam−1 (<4 × 1026 erg s−1 Hz−1) occurred 60 d
post-discovery. This would have been 118+14−5 days for the Type IIP
fit and 76+6−2 days post-explosion for the stripped envelope fit. This
means a Type IIP SN would have been well past peak luminosity
and at any point along the light curve below the detection limit. On
the other hand a stripped envelope SN would have been much closer
in time to peak with a luminosity well above the upper limit. This is
shown in Fig. 5 of Romero-Can˜izales et al. (2014) for the majority
of stripped envelope SNe with the exception of more extensively
MNRAS 473, 5641–5657 (2018)Downloaded from https://academic.oup.com/mnras/article-abstract/473/4/5641/4259573
by Turun Yliopiston Kirjasto user
on 29 March 2018
5650 E. C. Kool et al.
Table 3. Results from the fitting light-curve templates to the three SNe,
with line-of-sight extinction AV, time t0 between explosion date and discov-
ery, and a fixed constant C representing the intrinsic magnitude difference
between SNe. The final column shows the resulting χ˜2 for each fit.
Template AV t0 C χ˜2
SN 2013if (IRAS 18293-3413)
IIP, plateau 0.0+2.8−0.0 67
+1
−14 1.8
+0.1
−0.4 7.1
IIP, tail 0.0+2.5−0.0 136
+83
−13 0.0
+0.5
−1.7 4.7
IIn 0.0+2.7−0.0 19
+12
−3 3.1
+0.1
−0.5 5.0
IIb/Ib/Ic 0.0+2.9−0.0 18
+3
−2 1.5
+0.1
−0.5 7.4
SN 2015ca (NGC 3110)
IIP 3.4+1.0−1.7 58
+14
−5 1.5
+0.4
−0.3 3.8
IIn 2.7+0.2−0.7 25
+17
−9 3.0
+0.2
−0.2 10.3
IIb/Ib/Ic 2.8+0.3−0.4 16
+5
−2 1.3
+0.2
−0.2 5.0
Ia 6.9+0.1−0.7 5
+1
−1 0.4
+0.3
−0.2 17.9
SN 2015cb (IRAS 17138-1017)
IIP, plateau 4.6+0.3−0.1 64
+11
−7 −0.7+0.1−0.3 11.7
IIP, tail 3.6+0.4−0.1 134
+15
−4 −2.2+0.2−0.3 29.0
IIn 5.2+0.3−0.1 24
+5
−4 0.5
+0.2
−0.2 32.4
IIb/Ib/Ic 4.7+0.1−0.6 19
+10
−1 −1.1+0.1−0.2 11.2
AT 2015cf (NGC 3110)
IIP >7.0 >139 <1.6 −
IIb/Ib/Ic >0.0 >105 <2.0 −
stripped Type Ic SNe, like 2002ap or 2007gr. As such the eEVN
upper limit rules out a large fraction of stripped envelope SNe
implying a Type IIP is more likely for SN 2015ca.
5.3 SN 2015cb
The SN 2015cb data is best fitted by the templates of a Type IIP
caught in the plateau phase or a stripped envelope SN, with χ˜2 =
11.7 and χ˜2 = 11.2, respectively; see Fig. 8 and Table 3. In both
cases a slightly more luminous than average SN has been observed
(magnitude shifts C = −0.7+0.1−0.3 and C = −1.1+0.1−0.2, respectively)
with a moderate extinction of ∼4.5 magnitudes in V. To take into
account the r′-band limit, all template fits were required for the
limit to at least match the r′-band template curve. Similar to SN
2015ca, the lack of a radio detection with the VLA a month after
discovery favours a Type IIP scenario, but the presence of significant
contamination around the SN location prevents obtaining a strong
upper limit to conclusively exclude a stripped envelope SN.
5.4 AT 2015cf
If AT 2015cf were a young SN, this would yield an unlikely
combination of extremely high line-of-sight extinction and/or in-
trinsically very sub-luminous SN. Furthermore, the Ks-band de-
cline rate of AT 2015cf from 2015 March 11 until 2015 May 29
(0.5 ± 0.4 mag/100 d) is within errors roughly consistent with the
theoretical 56Co decay rate with complete γ -ray trapping. However,
typically H-poor SNe decline more rapidly in the tail phase. None
the less, to estimate parameter limits, we consider both the Type IIP
and IIb/Ib/Ic template options, with the former providing a some-
what better fit due to the aforementioned decline rate, see Table 3
and Fig. 9. The NOT observations in JHKs from 2015 April 6 cover
the site of AT 2015cf, but due to the poorer data quality (FWHM ∼
1.0 arcsec) and the small separation from SN 2015ca of 0.7 arcsec,
the presence of the residual of SN 2015ca in the subtracted im-
age prevents us from obtaining any meaningful upper limits for AT
2015cf. The NOT optical upper limits from 2015 March 27 are the
same as obtained for SN 2015ca, but the AT 2015cf data are fitted to
a much fainter stage in the templates which means the optical upper
limits do not provide useful constraints to the template fits. As such
we have opted not to include the NOT upper limits in Table 3 and
Fig. 9. The eEVN non-detection of SN 2015ca does not cover the
position of AT 2015cf and cannot be used to constrain the SN type.
We conclude that the observations of AT 2015cf are most consistent
with an old, possibly H-rich, CCSN.
5.5 CCSN rate of NGC 3110
Table 4 shows for NGC 3110 the SF rate, starburst age, CCSN
rate and the origin of its bolometric luminosity. This is based on
modelling the multiwavelength SED of NGC 3110, using data points
available from the literature ranging from optical to submillimetre
photometry (U et al. 2012), by combining libraries of starburst, AGN
torus and disc component models. For more details see Herrero-
Illana et al. (2017) and references therein, with the difference that a
spheroidal/cirrus component was fitted instead of a disc (Efstathiou
et al, in preparation).
We found that the SED of the host galaxy NGC 3110 was best
fitted by a disc model with a minor starburst component. Table 4
shows the best-fitting parameters and Fig. 10 the best-fitting model.
This means the dominant contributor to the galaxy’s total luminosity
is not the starburst luminosity, but rather the luminosity of the
disc component, which is in strong contrast with the other two
SN hosts IRAS 18293-3413 and IRAS 17138-1017 (Herrero-Illana
et al. 2017). This results in a CCSN rate for NGC 3110, which is
significantly lower (0.14+0.01−0.01 SNyr−1) than the expectation from the
LIRG’s IR luminosity, which would be ∼0.6 SNyr−1 when applying
equation (1). The reason for this is that as a disc-dominated object,
most of the IR luminosity in this galaxy is due to reprocessed
radiation from old stars which are not massive enough to explode
as SNe.
Finding two concurrent CCSNe in a galaxy with a CCSN rate
of 0.14yr−1 is peculiar, but not impossible. If we assume a Poisson
distribution and that SNe are detectable for 3 m, as has been the
case for the discoveries in this paper, the probability of discovering
two CCSNe in any of the four epochs of NGC 3110 is 4 per cent
and 0.2 per cent for an expected yearly SN rate of 0.6 and 0.14,
respectively. If we assume the same rates for our whole sample
(36 epochs in total), this increases to 30.8 per cent and 1.8 per cent,
respectively.
The CCSN nature of SN 2015ca is well established. The CCSN
nature of AT 2015cf remains poorly constrained, but other plausible
scenarios with sufficiently high rates are not known to the authors.
One possible explanation for this discrepancy is that the starburst
age is significantly underestimated. A higher starburst age would
give a higher CCSN rate. One important difference between the
model fit for NGC 3110 and the other LIRGs studied by Herrero-
Illana et al. (2017) is that in the case of NGC 3110 we lack Spitzer
IRS data for the whole galaxy (because of its large angular size)
which would constrain our model further.
5.6 Detection efficiency
To evaluate how effective our detection method was at recovering
SNe, we calculated the detection efficiency in the data by simulating
MNRAS 473, 5641–5657 (2018)Downloaded from https://academic.oup.com/mnras/article-abstract/473/4/5641/4259573
by Turun Yliopiston Kirjasto user
on 29 March 2018
First results from GeMS/GSAOI for SUNBIRD 5651
Figure 7. Template light-curve fits for SN 2015ca in NGC 3110 are shown: Type IIP fitted in the plateau phase, Type IIn, Type IIb/Ib/Ic and Type Ia.
artificial sources in an epoch typical for our sample. We chose to
simulate SNe in the GeMS/GSAOI epoch from 2013 April 21 of
IRAS 18293-3413, as it is a fairly typical LIRG at a typical distance
within our sample and host to SUNBIRD discovery SN 2013if. SNe
were simulated in the central region of the galaxy, defined as con-
taining 80 per cent of the galaxy light. We constructed a model PSF
from three bright field stars, placed it in a random position in this
region and then performed a subtraction as described in Section 3.4.
To recover the source, aperture photometry was taken at the sim-
ulation location and a 3 × 3 grid of apertures around it, separated
by twice the FWHM. The source was considered recovered when
the source flux exceeded three times the standard deviation of the
counts recorded in the surrounding aperture grid. The results were
split into three regions: the central 100 pc (<0.25 arcsec), where the
residuals from subtraction are very large, a wider nuclear region of
300 pc omitting the central 100 pc (0.25 arcsec–0.75 arcsec), and the
remaining area which extended to 600–800 pc (0.5 arcsec–2 arcsec).
This was repeated for a range of magnitudes and an S-curve
(e.g. Dahlen, Strolger & Riess 2008; Kankare et al. 2012) was
fitted to the data. We derive a preliminary 50 per cent detection
efficiency at magnitudes of 16.7, 19.3 and 20.7 for the three regions
referred to above, respectively. As expected, our detection efficiency
is significantly reduced in the central 100 pc of the galaxy. SN 2013if
was detected in this galaxy at magnitude 18.53 at 200 pc from the
nucleus, halfway the second region, which agrees well with the
derived 50 per cent detection efficiency for the region. For LIRGs
such as this, there is significant variation in the chances of detection,
heavily dependent on the local galaxy structure and proximity to
the nucleus. Additionally, differences in the data quality between
epochs affects detection efficiency. This is reflected by the non-
detection of SN 2013if in the J band (where AO-performance in
near-IR typically is worst) from 2013 June 11, with an upper limit
magnitude of 18.7. We will present an expanded approach to the
evaluation of this across multiple near-IR AO surveys, including
SUNBIRD, in a future publication (Reynolds et al, in preparation).
6 D I SCUSSI ON
With the detections of SN 2013if, SN 2015ca, SN 2015cb and AT
2015cf, currently there is a total of 60 CCSN optical and near-IR
discoveries in LIRGs reported in the literature since 1968: 39 op-
tical discoveries (see Table 5) and 21 in the near-IR (see Table 6).
Although the discoveries presented in this paper are a minor con-
tribution to the total sample, they are a significant addition to the
five previous CCSN discoveries in crowded and obscured regions
using near-IR AO observations. We next consider the usefulness of
a starburst CCSN survey in the near-IR with the use of LGSAO as
compared to the alternatives. Starting with the most well covered
temporal baseline, in total 39 reported CCSNe have been discov-
ered in LIRGs in the optical, with 29 since 2000; see Table 5
for a complete list. This table is a result of cross referencing all
galaxies in the IRAS RBGS with LIR > 1011 L (corrected for
MNRAS 473, 5641–5657 (2018)Downloaded from https://academic.oup.com/mnras/article-abstract/473/4/5641/4259573
by Turun Yliopiston Kirjasto user
on 29 March 2018
5652 E. C. Kool et al.
Figure 8. Template light-curve fits for SN 2015cb in IRAS 17138-1017 are shown: Type IIP fitted in the plateau and tail phase, Type IIn and Type IIb/Ib/Ic.
Figure 9. Template light-curve fits for AT 2015cf in NGC 3110 are shown: Type IIP and Type IIb/Ib/Ic.
H0 = 70 km s−1) with the most up to date SN catalogues available:
Open Supernova Catalog (Guillochon et al. 2017); Asiago Super-
nova Catalog (Barbon et al. 1999); Transient Name Server;11 and
ASAS-SN (Shappee et al. 2014). When equation (1) is applied to the
11 https://wis-tns.weizmann.ac.il/
same sample of galaxies, the collective expected LIRG CCSN rate
would be ∼250 yr−1. This would amount to ∼4000 CCSNe since
the start of this century, which is two orders of magnitude larger than
the actual observed CCSN rate for LIRGs. We note that equation
(1) is based on SF-dominated LIRGs, with negligible AGN contri-
bution to LIR. Furthermore a proper comparison can only be done
if one would also know the search characteristics of each survey,
MNRAS 473, 5641–5657 (2018)Downloaded from https://academic.oup.com/mnras/article-abstract/473/4/5641/4259573
by Turun Yliopiston Kirjasto user
on 29 March 2018
First results from GeMS/GSAOI for SUNBIRD 5653
Table 4. NGC 3110 SED model fitting parameters and derived physical
quantities.
Total luminosity (1011 L) 2.59+0.03−0.08
Starburst luminosity (1011 L) 0.42+0.08−0.02
Disc luminosity (1011 L) 2.17+0.03−0.13
AGN luminosity (1011 L) <0.01
SF rate, averaged
over the past 50 Myr (M yr−1) 12.7+0.7−1.1
Starburst age (Myr) 13.8+0.9−2.4
Core-collapse supernova rate (SNyr−1) 0.14+0.01−0.01
Figure 10. Best SED fitting model of NGC 3110 with the data points indi-
cated by the dots, starburst contribution by the red line and disc component
by the green line. No AGN contribution was required to fit the data. The
total fit is shown by the black line.
such as cadence and magnitude limits, which is beyond the scope
of this paper. However it is clear that almost all CCSNe exploding
in LIRGs are not being observed optically. This is not surprising
as current all-sky wide field SN searches are biased towards iso-
lated and bright SNe, and CCSNe hosted by LIRGs are at relatively
large distances (most LIRGs are > 50 Mpc) superimposed on bright
background emission and likely affected by dust extinction. For ex-
ample ASAS-SN, an all sky optical supernova search running since
2013 June has discovered ∼500 SNe so far none of which were
located in LIRGs (Holoien et al. 2017a,b,c).
Discoveries in the near-IR add another 21 CCSNe to the sample,
where we include SNe that were typed based on photometric data.
Throughout this discussion we treat AT 2015cf as a real CCSN, but
the overall conclusions do not change if this were not the case. The
discoveries summarized in Table 6 are all a result of SN searches
targeting starburst galaxies and include this work. The table is split
into CCSN discoveries in natural seeing conditions (non-AO) and
discoveries assisted by AO. As this is primarily a division in spa-
tial resolution, the space-based SPIRITS CCSN discoveries with
Spitzer/IRAC (1.5 arcsec diffraction limit) have been included in
the non-AO sample. The table shows a clear distinction: all but
one of the near-IR CCSNe discovered in natural seeing conditions
occurred outside the hosts’ inner kpc. In contrast, the AO-assisted
discoveries have primarily occurred within the inner kpc, with six
out of nine discoveries with nuclear offsets <1 kpc.
When the optical CCSN discoveries are added, the benefit of AO
to CCSN searches in starbursts becomes even more evident, see
Fig. 11. This figure shows the (projected) nuclear offsets of all 60
CCSN discoveries in LIRGs and it is clear that the near-IR programs
using AO have been much more effective in finding nuclear CCSNe
than both non-AO near-IR and optical programs, even when we
disregard discoveries with offsets > 4 kpc. This cutoff is applied to
account for selection effects due to differences in FOV. The FOV
of AO-imagers is limited and 4 kpc corresponds to 11 arcsec at a
typical distance of 75 Mpc of a LIRG hosting a discovered SN.
This is the limit of the CCSN search radius of ALTAIR/NIRI on
Gemini North, the smallest FOV (22 arcsec) of the AO instruments
with which CCSNe have been discovered in LIRGs. As such, this
should be treated as a lower limit, since for example the FOV of
GeMS/GSAOI is 85 arcsec.
The plotted numbers in Fig. 11 have not been normalized to ac-
count for coverage, so directly comparing the number of discoveries
between optical, near-IR and near-IR AO in each radial bin is not
appropriate. Normalization would require knowledge of how often
LIRGs have been observed in the optical and the near-IR to this day,
which is beyond the scope of this paper. However, we can make as-
sumptions and get an order of magnitude estimate of the historical
coverages of the three different distributions. For simplicity’s sake
we will express coverage in terms of the number of epochs.
In the near-IR we assume AO-assisted observations of LIRGs
are dominated by targeted SN searches, and we know the coverage
of these programs. Across Mattila et al. (2007) with VLT/NACO,
Kankare et al. (2008, 2012) with ALTAIR/NIRI, and this work with
GeMS/GSAOI there have been ∼180 epochs in near-IR with AO.
If we add the 17 NICMOS epochs from Cresci et al. (2007), this
amounts to a total of ∼200 epochs of high spatial resolution (AO)
imaging of LIRGs.
Individual near-IR non-AO SN programs such as Mannucci et al.
(2003) (234 epochs) and Miluzio et al. (2013) (210 epochs) al-
ready exceed the AO sum total. Other programs have targeted
starburst galaxies, including LIRGs, such as Grossan et al. (1999)
(∼500 epochs), Mattila, Meikle & Greimel (2004) (120 epochs)
and the ongoing SPIRITS survey (Jencson et al. 2017). Just based
on near-IR observations from targeted SN surveys, the coverage of
non-AO near-IR observations of LIRGs is a factor of five higher
than in the near-IR with AO. Based on this we assume the total
coverage in near-IR non-AO is at least an order of magnitude larger
than in AO.
By virtue of optical surveys having a larger FOV and greater
availability, the coverage of LIRGs in the optical extends over a
longer time baseline with higher cadence than in the near-IR. For
example the Lick Observatory Supernova Survey (LOSS, Leaman
et al. 2011) observed 14 882 galaxies over the course of 11 years,
with an average of 150 observations per galaxy. As this sample
included ∼100 LIRGs, this amounts to 15 000 epochs of LIRGs in
the optical for LOSS alone. If we also account for optical wide-field
surveys such as ASAS-SN (Shappee et al. 2014), PanSTARRS1
(Chambers et al. 2016) and CRTS (Drake et al. 2009) that observe
large swaths of the sky on a regular basis, it is safe to say the
coverage of LIRG observations in the optical is again at least an
order of magnitude higher than in the near-IR.
Based on these relative coverages and the numbers shown in
Fig. 11, we can conclude the following: First, despite a massive
difference in coverage, since 2000 there have been almost as many
CCSN discoveries in LIRGs in the near-IR (21) as there have been
MNRAS 473, 5641–5657 (2018)Downloaded from https://academic.oup.com/mnras/article-abstract/473/4/5641/4259573
by Turun Yliopiston Kirjasto user
on 29 March 2018
5654 E. C. Kool et al.
Table 5. CCSNe hosted by LIRGs discovered in optical, from cross referencing the IRAS Revised Bright Galaxy Sample with the Open Supernova Catalog,
Asiago Supernova Catalog, the Transient Name Server and the ASAS-SN supernova sample.
Name LIRG host RA Dec. Type Projected Discovered by
(J2000) (J2000) distance (kpc)
SN1968A NGC 1275 03:19:49 +41:30.2 I 12 Lovas
SN1983V NGC 1365 03:33:31.63 −36:08:55.0 Ic 7 Evans et al.
SN1993G NGC 3690 11:28:33.43 +58:33:31.0 II 3.7 Treffers et al.
SN1996D NGC 1614 04:34:00.30 −08:34:44.0 Ic 2.1 Drissen et al.
SN1998T NGC 3690 11:28:33.14 +58:33:44.0 Ib 1.5 BAOSS
SN1999bx NGC 6745 19:01:41.44 +40:44:52.3 II 5 LOSS
SN1999D NGC 3690 11:28:28.38 +58:33:39.0 II 6 BAOSS
SN1999ec IC 2163/NGC 2207 06:16:16.16 −21:22:09.8 Ib 14 LOSS
SN1999ex IC 5179 22:16:07.27 −36:50:53.7 Ib 6 Martin et al.
SN1999gl NGC 317B 00:57:40.07 +43:47:35.6 II 1.9 Boles
SN2000bg NGC 6240 16:52:58.18 +02:23:51.5 IIn 9 LOSS
SN2000cr NGC 5395 13:58:38.37 +37:26:12.4 Ic 10 Migliardi, Dimai
SN2001du NGC 1365 03:33:29.11 −36:08:32.5 II 7 Evans
SN2001eq PGC 70417 23:04:56.78 +19:33:04.8 Ic 6 LOTOSS
SN2001is NGC 1961 05:42:09.07 +69:21:54.8 Ib 14 BAOSS
SN2003H IC 2163/NGC 2207 06:16:25.68 −21:22:23.8 Ib Pec 6 LOTOSS
SN2003hg NGC 7771 23:51:24.13 +20:06:38.3 IIP 3.2 LOTOSS
SN2004bf UGC 8739 13:49:15.45 +35:15:12.5 Ic 8 LOSS
SN2004ed NGC 6786 19:10:53.62 +73:24:27.6 II 5 Armstrong
SN2004gh MCG-04-25-06 10:24:31.60 −23:33:18.4 II 2.0 LOSS
SN2005H NGC 838 02:09:38.52 −10:08:43.6 II 0.4 LOSS
SN2007ch NGC 6000 15:49:47.82 −29:23:13.7 II 3.6 Monard
SN2008fq NGC 6907 20:25:06.19 −24:48:27.6 II 1.4 LOSS
SN2009ap ESO 138-G27 17:26:43.23 −59:55:57.9 Ic 1.2 Pignata et al.
SN2010as NGC 6000 15:49:49.23 −29:23:09.7 Ib/c 0.6 Maza et al.
SN2010bt NGC 7130 21:48:20.22 −34:57:16.5 II 6 Monard
SN2010gg ESO 602-G25 22:31:25.47 −19:01:54.8 II 6 Pignata et al.
SN2010gk NGC 5433 14:02:35.94 +32:30:30.7 Ic 2.0 LOSS
SN2010jp IC 2163/NGC 2207 06:16:30.63 −21:24:36.3 II 130 Maza et al.
SN2010O NGC 3690 11:28:33.86 +58:33:51.6 Ib 1.6 Newton, Puckett
SN2012by UGC 8335 13:15:28.90 +62:07:47.8 II 10 Rich
SN2013ai IC 2163/NGC 2207 06:16:18.35 −21:22:32.9 II 9 Conseil
SN2013cc NGC 1961 05:41:58.76 +69:21:40.9 II 19 Itagaki
SN2013dc NGC 6240 16:52:58.97 +02:24:25.2 IIP 11 Block
SN2014dj NGC 317B 00:57:40.18 +43:47:35.1 Ic 1.5 Rich
SN2014eh NGC 6907 20:25:03.86 −24:49:13.3 Ic 14 LOSS
SN2015ae NGC 7753 23:47:06.15 +29:29:07.4 II 6 Itagaki
SN2015U NGC 2388 07:28:53.87 +33:49:10.6 Ibn 1.8 LOSS
PS15aaa IC 564 09:46:20.73 +03:04:22.1 II 3.2 Pan-STARRS
in the optical (29), which implies SN discovery in LIRGs is sub-
stantially more efficient in the near-IR. Secondly, despite a signifi-
cant difference in coverage, in the near-IR there have been almost
as many CCSN discoveries from AO programs (9) as from non-
AO programs (13), with the AO discoveries weighted towards the
nucleus regions. The simplest explanation for these conclusions
is the combination of reduced dust extinction in the near-IR, and
the improved spatial resolution for AO-assisted imaging at these
wavelengths, both of which provide enhanced sensitivity to CCSNe
detection in these dusty and compact star forming objects.
The CCSN radial distribution as seen in near-IR AO studies
agrees well with work on the spatial distribution of CCSNe and
CCSN remnants in the nuclear regions of three starburst galax-
ies and a sample of spiral galaxies using high-angular resolution
(0.1 arcsec) radio VLBI observations studied by Herrero-Illana,
Pe´rez-Torres & Alberdi (2012). Here it was found that the SN ra-
dial distribution in the LIRGs Arp 220 and Arp 299 was centrally
peaked with a very steep SN surface number density profile when
compared to the SN radial distribution in regular spiral galaxies from
Hakobyan et al. (2009). The SNe and SN remnants in Herrero-Illana
et al. (2012) have radial distances almost exclusively smaller than
the most centrally located SN in Table 6, which is not surprising
given the superior spatial resolution, but limited FOV, of VLBI.
7 C O N C L U S I O N S
We have introduced project SUNBIRD, a systematic search for
CCSNe in nearby LIRGs using AO imaging at near-IR wavelengths,
where we aim to characterize the population of CCSNe hidden in
obscured and crowded regions of star forming galaxies. We have
presented the first results of this project: so far we have covered
a sample of 13 LIRGs with GeMS/GSAOI on the Gemini South
telescope and discovered three photometrically confirmed CCSNe
and one CCSN candidate. Two of the three CCSNe occurred close
to the nucleus of their host galaxies. SN 2013if had a projected
distance from the nucleus as small as 200 pc, which makes it the
second most nuclear CCSN discovery in a LIRG to date in the
optical and near-IR after SN 2010cu (Kankare et al. 2012).
MNRAS 473, 5641–5657 (2018)Downloaded from https://academic.oup.com/mnras/article-abstract/473/4/5641/4259573
by Turun Yliopiston Kirjasto user
on 29 March 2018
First results from GeMS/GSAOI for SUNBIRD 5655
Table 6. CCSNe discovered in near-IR as found in the literature, divided into discoveries in natural seeing conditions and with the assistance of AO. Space
based CCSN discoveries are distributed based on the spatial resolution of the discovery image. This list includes CCSNe without spectral confirmation, and
are therefore officially potential SNe (PSN) or astronomical transients (AT). Nuclear offsets of the CCSN discoveries in this work are given with 0.1 arcsec
precision to reflect the uncertainty of the centre of the hosts nucleus.
Name LIRG host RA Dec. Type Reference Extinction Projected
(J2000) (J2000) AV (mag) distance (kpc)
Non-AO
SN1992bu NGC 3690 11:28:31.5 +58:33:38 Ib/c? van Buren et al. (1994) – 1.3
SN1999gw UGC 4881 09:15:54.7 +44:19:55 II Maiolino et al. (2002) – 2.9
SN2001db NGC 3256 10:27:50.4 −43:54:21 II Maiolino et al. (2002) 5.5 1.5
SN2005U NGC 3690 11:28:33.22 +58:33:42.5 IIb Mattila et al. (2005a) – 1.4
SN2005V NGC 2146 06:18:38.28 +78:21:28.8 Ib/c Mattila et al. (2005b) – 0.5
SN2010hp MCG-02-01-52 00:18:50.01 −10:21:40.6 IIP Miluzio et al. (2013) 0.5 2.1
SN2010P NGC 3690 11:28:31.38 +58:33:49.3 Ib/IIb Kankare et al. (2014) 7 1.2
PSN2010 IC 4687 18:13:40.213 −57:43:28.00 IIP? Miluzio et al. (2013) 0–8 6.5
PSN2011 IC 1623A 01:07:46.229 −17:30:29.48 Ic? Miluzio et al. (2013) 0.5 3.2
SN2011ee NGC 7674 23:27:57.34 +08:46:38.1 Ic Miluzio et al. (2013) 0 8.9
SPIRITS 14buu IC 2163/NGC 2207 06:16:27.2 −21:22:29.2 IIP? Jencson et al. (2017) 1.5 2.0
SPIRITS 15c IC 2163/NGC 2207 06:16:28.49 −21:22:42.2 IIb Jencson et al. (2017) 2.2 2.1
AO
SN2004ip IRAS 18293-3413 18:32:41.15 −34:11:27.5 II Mattila et al. (2007) 5–40 0.5
SN2004iq IRAS 17138-1017 17:16:35.90 −10:20:37.9 II Kankare et al. (2008) 0-4 0.66
SN2008cs IRAS 17138-1017 17:16:35.86 −10:20:43.0 II Kankare et al. (2008) 17–19 1.5
SN2010cu IC 883 13:20:35.36 +34:08:22.2 II Kankare et al. (2012) 0–1 0.18
SN2011hi IC 883 13:20:35.38 +34:08:22.23 II Kankare et al. (2012) 5–7 0.38
SN2013if IRAS 18293-3413 18:32:41.10 −34:11:27.24 IIP This work 0-3 0.2
SN2015ca NGC 3110 10:04:01.57 −06:28:25.48 IIP This work 3 3.5
SN2015cb IRAS 17138-1017 17:16:35.84 −10:20:37.48 II This work 4.5 0.6
AT2015cf NGC 3110 10:04:01.53 −06:28:25.84 II? This work 2–5 3.5
Figure 11. Nuclear offset distribution for all CCSNe discovered in LIRGs.
In blue are shown the discoveries in the near-IR with AO instruments, in
green the near-IR CCSN discoveries in natural seeing, and in red CCSN dis-
coveries in the optical. The dotted line at 4 kpc marks the limit of the smallest
FOV of the instrument used for the AO CCSN discoveries, ALTAIR/NIRI
on Gemini North, at a typical distance for LIRGs with SN discoveries of
75 Mpc.
To investigate the impact of these new discoveries and the ef-
fectiveness of our near-IR AO CCSN search, we gathered from the
literature all CCSN discoveries in LIRGs in the optical and the
near-IR. This sample consists of 60 events, out of which nine were
discovered using near-IR AO, including this project. We show a
clear distinction in nuclear offset distribution between these nine
events and the other optical and near-IR discoveries. Almost all
CCSN discoveries from AO programs occurred in the inner kpc
of their hosts, but only a fraction of the non-AO near-IR and op-
tical CCSNe had offsets smaller than 1 kpc. This tells us that our
approach is singularly effective in uncovering CCSNe in nuclear
regions, and crucial in characterizing this population of CSSNe that
will remain invisible through other means.
AC K N OW L E D G E M E N T S
We would like to thank the anonymous referee for constructive
comments. This publication is based on observations obtained at
the Gemini Observatory, which is operated by the Association of
Universities for Research in Astronomy, Inc., under a coopera-
tive agreement with the NSF on behalf of the Gemini partnership:
the National Science Foundation (United States); the National Re-
search Council (Canada); CONICYT (Chile); Ministerio de Cien-
cia, Tecnologı´a e Innovacio´n Productiva (Argentina); and Ministe´rio
da Cieˆncia, Tecnologia e Inovac¸a˜o (Brazil). The relevant program
codes are: GS–2012B–SV–407 (PI: S. Ryder); GS–2013A–Q–9 (PI:
S. Ryder/F. Bauer); GS–2015A–C–2, (PI: S. Sweet/R. Sharp); GS–
2015A–Q–6 (PI: S. Ryder); GS–2015A–Q–7 (PI: S. Ryder); GN–
2015A–DD–4 (PI: S. Ryder) and GS–2016A–C–1 (PI: E. Kool).
This publication is based on observations made with the Nordic
Optical Telescope, operated by the Nordic Optical Telescope Scien-
tific Association at the Observatorio del Roque de los Muchachos,
La Palma, Spain, of the Instituto de Astrofisica de Canarias.
This publication is based on observations made with ESO Tele-
scopes at the La Silla Paranal Observatory under programme ID’s
073.D–0406 (PI: P. Va¨isa¨nen), 086.B–0901 (PI: A. Escala) and
089.D–0847 (PI: S. Mattila) and on data obtained from the ESO
MNRAS 473, 5641–5657 (2018)Downloaded from https://academic.oup.com/mnras/article-abstract/473/4/5641/4259573
by Turun Yliopiston Kirjasto user
on 29 March 2018
5656 E. C. Kool et al.
Science Archive Facility under request numbers eckool194907, –
196798 and –226876.
The National Radio Astronomy Observatory is a facility of the
National Science Foundation operated under cooperative agreement
by Associated Universities, Inc.
The European VLBI Network is a joint facility of independent
European, African, Asian and North American radio astronomy
institutes. Scientific results from data presented in this publication
are derived from the following EVN project code(s): RSP13 (P.I.
Pe´rez-Torres).
RMcD is the recipient of an Australian Research Council Future
Fellowship (project number FT150100333).
ECK is grateful for financial support provided by the Interna-
tional Macquarie University Research Excellence Scholarship and
the Australian Astronomical Observatory (AAO) through the AAO
PhD Scholarship Scheme.
MAPT, RHI and AA acknowledge support by the Span-
ish MINECO through grants AYA2012–38491–C02–02 and
AYA2015–63939–C2–1–P, co-funded with FEDER funds.
FEB acknowledges support from CONICYT–Chile Basal–
CATA PFB–06/2007 and FONDECYT Regular 1141218. CRC ac-
knowledges support from CONICYT through FONDECYT grant
3150238. FEB and CRC acknowledge support from the Ministry of
Economy, Development, and Tourism’s Millennium Science Initia-
tive through grant IC120009, awarded to The Millennium Institute
of Astrophysics, MAS.
R E F E R E N C E S
Adams S. M., Kochanek C. S., Beacom J. F., Vagins M. R., Stanek K. Z.,
2013, ApJ, 778, 164
Alard C., 2000, A&AS, 144, 363
Alard C., Lupton R. H., 1998, ApJ, 503, 325
Arcavi I. et al., 2012, ApJ, 756, L30
Barbon R., Buondı´ V., Cappellaro E., Turatto M., 1999, A&AS, 139, 531
Bertin E., 2006, in Gabriel C., Arviset C., Ponz D., Enrique S., eds, ASP
Conf. Ser. Vol. 351, Astronomical Data Analysis Software and Systems
XV. Astron. Soc. Pac., San Francisco, p. 112
Bertin E., Arnouts S., 1996, A&AS, 117, 393
Bertin E., Mellier Y., Radovich M., Missonnier G., Didelon P., Morin B.,
2002, in Bohlender D. A., Durand D., Handley T. H., eds, ASP Conf.
Ser. Vol. 281, Astronomical Data Analysis Software and Systems XI.
Astron. Soc. Pac., San Francisco, p. 228
Botticella M. T., Smartt S. J., Kennicutt R. C., Cappellaro E., Sereno M.,
Lee J. C., 2012, A&A, 537, A132
Cappellaro E. et al., 2015, A&A, 584, A62
Cardelli J. A., Clayton G. C., Mathis J. S., 1989, ApJ, 345, 245
Carrasco E. R. et al., 2012, in Ellerbroek B. L., Marchetti E., Ve´ran J.-P,
eds, Proc. SPIE Conf. Ser. Vol. 8447, Adaptive Optics Systems III. SPIE,
Bellingham, p. 84470N
Chambers K. C. et al., 2016, preprint (arXiv:1612.05560)
Chomiuk L. et al., 2016, ApJ, 821, 119
Cresci G., Mannucci F., Della Valle M., Maiolino R., 2007, A&A, 462, 927
Dahlen T., Strolger L.-G., Riess A. G., 2008, ApJ, 681, 462
Dahlen T., Strolger L.-G., Riess A. G., Mattila S., Kankare E., Mobasher
B., 2012, ApJ, 757, 70
Djupvik A. A., Andersen J., 2010, Ap&SS, 14, 211
Dopita M. A. et al., 2011, Ap&SS, 333, 225
Drake A. J. et al., 2009, ApJ, 696, 870
Elias J. H., Rodgers B., Joyce R. R., Lazo M., Doppmann G., Winge C.,
Rodrı´guez-Ardila A., 2006a, in McLean I. S., Iye M., eds, Proc. SPIE
Conf. Ser. Vol. 6269, Ground-based and Airborne Instrumentation for
Astronomy. SPIE, Bellingham, p. 626914
Elias J. H., Joyce R. R., Liang M., Muller G. P., Hileman E. A., George
J. R., 2006b, in McLean I. S., Iye M., eds, Proc. SPIE Conf. Ser.
Vol. 6269, Ground-based and Airborne Instrumentation for Astronomy.
SPIE, Bellingham, p. 62694C
Erben T. et al., 2005, Astron. Nachr., 326, 432
Ergon M. et al., 2014, A&A, 562, A17
Ergon M. et al., 2015, A&A, 580, A142
Farrah D., Surace J. A., Veilleux S., Sanders D. B., Vacca W. D., 2005, ApJ,
626, 70
Fassia A. et al., 2000, MNRAS, 318, 1093
Grossan B., Spillar E., Tripp R., Pirzkal N., Sutin B. M., Johnson P., Barnaby
D., 1999, AJ, 118, 705
Guillochon J., Parrent J., Kelley L. Z., Margutti R., 2017, ApJ, 835, 64
Hakobyan A. A., Mamon G. A., Petrosian A. R., Kunth D., Turatto M.,
2009, A&A, 508, 1259
Hancock P. J., Gaensler B. M., Murphy T., 2011, ApJ, 735, L35
Herrero-Illana R., Pe´rez-Torres M. ´A., Alberdi A., 2012, A&A, 540, L5
Herrero-Illana R. et al., 2017, MNRAS, 471, 1634
Holoien T. W.-S. et al., 2017a, MNRAS, 471, 4966
Holoien T. W.-S. et al., 2017b, MNRAS, 464, 2672
Holoien T. W.-S. et al., 2017c, MNRAS, 467, 1098
Horiuchi S., Beacom J. F., Kochanek C. S., Prieto J. L., Stanek K. Z.,
Thompson T. A., 2011, ApJ, 738, 154
Jencson J. E. et al., 2017, ApJ, 837, 167
Jester S. et al., 2005, AJ, 130, 873
Kankare E. et al., 2008, ApJ, 689, L97
Kankare E. et al., 2012, ApJ, 744, L19
Kankare E. et al., 2014, MNRAS, 440, 1052
Kankare E., Kool E., Mattila S., Ryder S., 2017, Transient Name Serv.
Discovery Rep., 10
Kool E., Ryder S., Mattila S., Kankare E., 2016, Transient Name Serv.
Discovery Rep., 1073
Kool E., Ryder S., Mattila S., Kankare E., 2017a, Transient Name Serv.
Discovery Rep., 9
Kool E., Ryder S., Mattila S., Kankare E., 2017b, Transient Name Serv.
Discovery Rep., 555
Krisciunas K. et al., 2009, AJ, 137, 34
Leaman J., Li W., Chornock R., Filippenko A. V., 2011, MNRAS, 412,
1419
Lenzen R. et al., 2003, in Iye M., Moorwood A. F. M., eds, Proc. SPIE Conf.
Ser. Vol. 4841, Instrument Design and Performance for Optical/Infrared
Ground-based Telescopes. SPIE, Bellingham, p. 944
Leonard D. C. et al., 2002, PASP, 114, 35
Liu Q.-Z., Hu J.-Y., Hang H.-R., Qiu Y.-L., Zhu Z.-X., Qiao Q.-Y., 2000,
A&AS, 144, 219
Magnelli B., Elbaz D., Chary R. R., Dickinson M., Le Borgne D., Frayer D.
T., Willmer C. N. A., 2011, A&A, 528, A35
Maiolino R., Vanzi L., Mannucci F., Cresci G., Ghinassi F., Della Valle M.,
2002, A&A, 389, 84
Mannucci F. et al., 2003, A&A, 401, 519
Matheson T. et al., 2012, ApJ, 754, 19
Mattila S., Meikle W. P. S., 2001, MNRAS, 324, 325
Mattila S., Meikle W. P. S., Greimel R., 2004, New Astron. Rev., 48,
595
Mattila S. et al., 2005a, in Green D. W. E., ed., IAU Circ., 8473
Mattila S. et al., 2005b, in Green D. W. E., ed., IAU Circ., 8474
Mattila S. et al., 2007, ApJ, 659, L9
Mattila S. et al., 2012, ApJ, 756, 111
McGregor P. et al., 2004, in Moorwood A. F. M., Iye M., eds, Proc. SPIE
Conf. Ser. Vol. 5492, Ground-based Instrumentation for Astronomy.
SPIE, Bellingham, p. 1033
McMullin J. P., Waters B., Schiebel D., Young W., Golap K., 2007, in Shaw
R. A., Hill F., Bell D. J., eds, ASP Conf. Ser. Vol. 376, Astronomical Data
Analysis Software and Systems XVI. Astron. Soc. Pac., San Francisco,
p. 127
Melinder J. et al., 2012, A&A, 545, A96
Miluzio M. et al., 2013, A&A, 554, A127
Neichel B. et al., 2014a, MNRAS, 440, 1002
Neichel B., Lu J. R., Rigaut F., Ammons S. M., Carrasco E. R., Lassalle E.,
2014b, MNRAS, 445, 500
MNRAS 473, 5641–5657 (2018)Downloaded from https://academic.oup.com/mnras/article-abstract/473/4/5641/4259573
by Turun Yliopiston Kirjasto user
on 29 March 2018
First results from GeMS/GSAOI for SUNBIRD 5657
Patat F., 1996, Ph.D. thesis, Univ. Padova
Perez-Torres M. et al., 2015, Proc. Sci., Core-collapse and Type Ia super-
novae with the SKA. SISSA, Trieste, PoS#60
Randriamanakoto Z., Escala A., Va¨isa¨nen P., Kankare E., Kotilainen J.,
Mattila S., Ryder S., 2013, ApJ, 775, L38
Rigaut F. et al., 2014, MNRAS, 437, 2361
Romero-Can˜izales C. et al., 2014, MNRAS, 440, 1067
Rousset G. et al., 2003, in Wizinowich P. L., Bonaccini D., eds, Proc. SPIE
Conf. Ser. Vol. 4839, Adaptive Optical System Technologies II. SPIE,
Bellingham, p. 140
Sanders D. B., Mazzarella J. M., Kim D.-C., Surace J. A., Soifer B. T., 2003,
AJ, 126, 1607
Scalzo R. et al., 2017, Publ. Astron. Soc. Aust., 34, e030
Schirmer M., 2013, ApJS, 209, 21
Schirmer M., Carrasco E. R., Pessev P., Garrel V., Winge C., Neichel B.,
Vidal F., 2015, ApJS, 217, 33
Schlafly E. F., Finkbeiner D. P., 2011, ApJ, 737, 103
Shappee B. J. et al., 2014, ApJ, 788, 48
Strolger L.-G. et al., 2015, ApJ, 813, 93
Tody D., 1993, in Hanisch R. J., Brissenden R. J. V., Barnes J., eds, ASP
Conf. Ser. Vol. 52, Astronomical Data Analysis Software and Systems
II. Astron. Soc. Pac., San Francisco, p. 173
U V. et al., 2012, ApJS, 203, 9
Valenti S. et al., 2011, MNRAS, 416, 3138
van Buren D., Jarrett T., Terebey S., Beichman C., Shure M., Kaminski C.,
1994, in Green D. W. E., ed., IAU Circ., 5960
Xiao L., Eldridge J. J., 2015, MNRAS, 452, 2597
Zhang K. et al., 2016, ApJ, 820, 67
A P P E N D I X A : DATA R E D U C T I O N W I T H THELI
Described here are optimisations or adaptations of the procedures
from Schirmer (2013) and Schirmer et al. (2015). The focus is on as-
trometric calibration of images with low number source density re-
quired for correcting the distortion pattern present in GeMS/GSAOI
data.
A1 Calibration and background modelling
The science exposures were divided by a master flat combined from
typically 10 dome flats, and the background removed by running a
two-pass background model. In short, the first pass is a simple back-
ground model subtraction of a median combination of all exposures
without masking to remove the bulk of the signal, whereas in the
second pass objects are masked by applying SEXTRACTOR (Bertin &
Arnouts 1996) for source identification. A static background model
was used if the exposure sequence did not last longer than 1 h. In
a few cases a dynamic model was warranted using a running me-
dian of six images. In contrast to Schirmer et al. (2015) no mask
expansion factor or collapse correction was deemed necessary for
this work.
A2 Astrometric calibration
The astrometric calibration of individual dithered images prior to co-
addition in THELI enables the distortion correction of GeMS/GSAOI
data, but also requires the most attention to properly process the data.
Because of the low number source density, no single combination
of parameter settings for the packages that THELI employs (Scamp,
Swarp: Bertin et al. (2002) and SEXTRACTOR) proved to be suffi-
cient for all GeMS/GSAOI data sets, some informed adjustments
were always required. Most settings are well covered in the THELI
documentation and aforementioned procedures, but the following
adjustments were not obvious and were vital for the astrometric
calibration of our GeMS/GSAOI data.
In all of the data sets the comparatively small FOV of GSAOI
in combination with sparsely populated fields meant all-sky astro-
metric reference catalogues were insufficient for astrometric cali-
bration of the individual frames. Instead secondary reference cata-
logues were first created with THELI based on Ks-band archival data
of ground-based wide-field instruments: HAWK-I on the VLT or
VIRCAM on the VISTA telescope. These were calibrated against
typically 100–1000 2MASS sources in the FOV, which have individ-
ual uncertainties of ∼100 mas, resulting in astrometric uncertainties
of the reference images of 10 mas. After successfully construct-
ing a distortion corrected co-added GeMS/GSAOI image, subse-
quent GeMS/GSAOI epochs were calibrated against a reference
catalogue extracted from this image. Depending on field crowding
of the astrometric reference image and its spatial resolution, the
parameter DEBLEND_MINCONT in postcoadd.conf.sex in the THELI re-
duction folder needed to be adjusted (decreased when working with
a GeMS/GSAOI reference image).
The WCS header information in our raw GeMS/GSAOI data
is generally accurate within ∼5 arcsec, but this proved to be not
precise enough in fields with few matching reference sources. In
almost all cases it was necessary to update the CRPIX1/2 keywords in
the headers to match with the coordinates of the reference catalogue.
Then, as described in section 3.5.1 in Schirmer et al. (2015), a refined
median estimate of the relative array positions and orientations from
all exposures in a night is used (MOSAIC_TYPE = FIX_FOCALPLANE),
and as we have adjusted the reference pixel manually no WCS
matching is required: MATCH = N.
THELI uses SEXTRACTOR to create a source catalogue from the sci-
ence images which are then fitted to the corresponding sources
(within CROSSID_RADIUS) in the reference catalogue to obtain a dis-
tortion correction. This process is done for each array separately
and the SEXTRACTOR settings might not be appropriate for both the
array containing the galaxy (high source number density) and an
array covering a sparse field. In the case of NGC 3110 in order to
recover all four arrays it was necessary to run the source extraction
individually with appropriate settings (most importantly detection
thresholds DETECT_THRESH and DETECT_MINAREA and de-blending pa-
rameter DEBLEND_MINCONT). This can be achieved by suspending the
parallel manager script and the details are described in the THELI
documentation.12
12 https://astro.uni-bonn.de/∼theli/gui/advancedusage.html
This paper has been typeset from a TEX/LATEX file prepared by the author.
MNRAS 473, 5641–5657 (2018)Downloaded from https://academic.oup.com/mnras/article-abstract/473/4/5641/4259573
by Turun Yliopiston Kirjasto user
on 29 March 2018