NASA Technical Reports Server (NTRS) 20120016515: Seeking Counterparts to Advanced LIGO/Virgo Transients with Swift

3.7 MB·English

by  NASA Technical Reports Server (NTRS)

DRA?T VERSION SEPTEMBER 12, 2012
Preprint typeset using D-'IEX style emulatea.pj v. 12/16/11
SEEKING COUNTERPARTS TO ADVANCED LIGO/VIRGO TRANSIENTS WITH SWIIT
JONAH KANNER, JORDAN C.\MP
NASA Goddard Space Flight Center, Mail Code 663, Greenbelt, MD 20771
JUDITH RACUSIN, NEIL GEHRELS
NASA Goddard Space Flight Center, Mail Code 661, Greenbelt, AID 20771
AND
DARREN WHITE
University of Sheffield, Hicks Building, Hounsfield Road, Sheffield S3 7RH, United Kingdom
Draft version September 12, 2012
ABSTRACT
Binary neutron star (NS) mergers are among the most promis'ng astrophysical sources of gravita
. -: tional wave emission for Advanced LIGO and Advanced Virgo, expected to be operational in 2015 .
.- Finding electromagnetic counterparts to these signals will be essential to placing them in an astro
nomical context. The Swift satellite carries a sensitive X-ray telescope (XRT), and can respond to
(----, target-of-<lpportunity requeats within 1-2 hours, and so is uniquely poised to find the X-ray counter
~ parts to LIGO /Virgo triggers. Assuming NS mergers are the progenitors of short gamma-ray bursts
""';
(GRBs), some percentage of LIGO/Virgo triggers will be accompanied by X-ray band afterglows that
l--(
are brighter than 10-12 ergs/s/cm2 in the XRT band one day after the trigger time. We find that
a soft X-ray transient of this flux is bright enough to be extremely rare, and so c.ould be confidently
associated with even a moderately localized GW signal. We examine two possible search strategies
with the Swift XRT to find bright transients in LIGO /Virgo error hoxes. In the first strategy, XRT
could search a volume of space with a ,.....100 Mpc radius by observing ...... 30 galaxies over the course of
a day, with sufficient depth to observe the expected X-ray afterglow. For an extended LIGO /Virgo
horizon distance, the XRT could employ very short 100 s exposures to cover an area of '" 35 square
degrees in about a day, and still be sensitive enough to image GW discovered GRB afterglows. These
strategies demonstrate that the high X-ray luminosity of short GRBs and the relatively low X-ray
transient background combine to make high confidence discoveries of X-ray band counterparts to GW
triggers possible, though challenging, with current satellite facilities.
Subject headings: gravitational waves, galaxies: statistics, gamma rays: bursts, X-rays: general
1. INTRODUCTION the GW signal would directly probe the progenitor's dy
namics while the EM signal would carry information on
The construction of tbe Advanced LlGO and Ad
vanced Virgo 1 gravitational wave (GW) observatories the environment and allow improved parameter estima
tion. Moreover, a population of such joint EU/GW sig
is in progress, with completion expected as early as 2015
nals could be used to estimate cosmological parameters
(Harryet al. 2010; Accadia et al. 2012). These detectors
(Dalal et al. 2006; Nissanke et al. 2010).
are expected to serve as all-sky monitors for mergers of
The main challenge in the identification of an EM
binary neutron stars (NS-NS), mergers of neutron stars
counterpart to an observed GW signal "l'dU be the large
with stellar mass black holes (NS-BH), and mergers ofbi
positional uncertainty associated with current networks
nary black holes. The detectors are designed to observe
of GW observatories. The positional uncertainty for a
NS-NS mergers to an average distance of 200 Mpc, and
given event will depend strongly on a number of fac
NS-BH mergers to 400 Upc (Abadie et aI. 2010). The
tors, induding signal-to-noise ratio (SNR), position on
G W signal from such events could provide a wealth of in
the sky, available position reconstruction algorithms, and
formation, including the masses and spins of the compo
the internal state of the GW network. As designed,
nent objects, and an estimate of the luminosity distance
all three detectors in the Advanced LIGO/Virgo net
to the source. However, this network of gravitational
work would operate at similar sensitivity, leading to typ
wave detectors will have, even in the best case scenarios,
ical positional uncertainties around 20 square deg:-ees
only modest localization ability. This means that placing !"oJ
(Klimenko et al. 2011; Fairhurst 2011; Nissanke et al.
a GW observation in an astronomical context, including
2011). On the other hand, studies using data from the
identification of a host galaxy and environment, will re
quire finding 8 counterpart electromagnetic (EM) sigllal last science fUns of LIGO and Virgo have shown un
certainties for low SNR signals of 50-200 square degrees
to the merger event. A joint EM/GW observation would
(Abadie et al. 20l2b), during times when detectors dif
describe an explosive event in unprecedented detail, since
fered by a factor of ~ 2 in amplitude sensitivity. During
[email protected] the early :rears of observing with the second generation
1 https://tds.ego-gw.itjitf/tdsjfile.php?calIFile=VlR-0027A- network, evolving sensitivity levels will likely lead to va.ri-
09.pdf
2
ation in the ability to localize sources. but is typically expected to be between a few degrees and
Since there have been no certain observations of stellar a few tens of degrees, meaning that Ib is likely of order
mass compact object mergers, predicting the wavelength, 1 %. If this is the case, then there is a. major implication
flux, and duration of a possible EM signal is somewhat for L1GO /Virgo triggers: the vast majority of them will
speculative. However, some models seem promising. NS not be associated with on-axis GRBs. However, the frac
NS mergers and NS-BH mergers are both possible pro tion of GW selected events within the beam will likely
genitors for short gamma-ray bursts (GRBs) (Fox et a1. be larger than Ib due to a particular bias (Schutz 2011;
2005; Gehrels et a1. 2005), and both their prompt emis Nissanke et a1. 2010)". lnspiraling compact objects emit
sion an.d afterglow have been carefully studied. In addi gravitational wave energy preferentially in the direction
tion, theoretical considerations and simulations lead to parallel to their angular momentum axis, that is, the
the expectation of an isotropic, optical signal that re gravi tational waves are weakly beamed in the same di
sults from energy released in decays of unstable isotopes rection as the GRB jet. Schutz (2011) shows that this
following r-prOcess nUcleoeynthesis (Metzger et a1. 2010; effect will increase the fraction of G W selected compact
Roberts et a1. 2011; Piran et a1. 2012). Some considera object mergers with their beam pointed towards earth
tion has already been given to observing strategies that by a factor of 3.4 over the strictly geometric prediction.
might discover an optical or radio band counterpart to Moreover, a few short bursts have lower limits on their jet
a merger event (Metzger & Berger 2012; Coward et a1. opening angles placed above 10 degrees (Racnsin et al.
2011), and searches have been performed in a range of 2011; Fong et al. 2012; Coward et a1. 20l2). If we es.
wavelengths on low-threshold triggers from the initial timate the opening angle of short GRBs as around 10
LIGO/Virgo network (Evans et a1. 2012; Abadie et a1. degrees, then the expected fraction of LIGO /Virgo ob
2012b,a). Past searches have also sought coincident trig served mergers with earth within the jet is 5%, or one
f"V
gers using archived GRB and GW data (Abadie et al. observable GRB in every - 20 LIGO /Virgo observed NS
2012c). In this work, we focus on the X-ray band, and NS mergers.
consider the Swift satellite (Gehrels et al. 2004) as a po The above estimate of fb is highly uncertain, due to
tential instrument for discovering E.M signals following a the limited number of observations of short GRB jet
trigger from the second generation LIGO /Virgo network. breaks. On the other hand, it is possible to make a rea
Swift has an unmatched record as an engine for sonably robnst estimate of the number of short GRBs
produciug observations of GRBs and their afterglows. within the LIGO/Virgo range, based on the observed
In the typical mode of discovery, Swift's Burst Alert rates of GRBs (Chen & Holz 2012; ~letzger & Berger
Telescope (Barthelmy et a1. 2005) sweeps the sky for 2012; Abbott et a1. 2010). The main source of uncer
GRBs. When a GRB is discovered, the X-ray Telescope tainty is then the progenitor of short bursts. If we as
(XRl') (Burrows et al. 2005) and UV /Optical Telescope sume that all short GRBs are due to NS/NS mergers,
(UVOT) (Roming et a1. 2005) are automatically slewed then we expect 0.3 - 3 events in range per year with
to the estimated source position. This strategy has Advanced LIGO design sensitivity. However, the LIGO
been extremelv successful: the XRT finds soft-band X rauge for BH/NS events is greater by a factor of - 2,
ray counterparts to nearly 80% of observed short GRBs leading to 2 - 20 events per year if we ""sume all short
with prompt observations. For comparison, optical band GRBs are caused by NS/BH mergers. In the early years
counterparts are only discovered for - 30% of short of Advanced LIGO, with half the design range, we might
GRBs. If we accept the putative link between compact then expect ~ 0.1 GW /GRB coincidences per year un
object mergers and short GRBs, then the XRT band (0.3- der the NS/NS assumption, or ~ 1 such event per year
10 keY) seems a natural place to find the counterparts under the BH/NS assumption. Uncertainties include the
to compact object mergers. possible gain in reach of the GW instruments due to re
duced background with a coincident GRB observation,
or factors accounting for the less-than-perfect spatia! and
2. SOURCES
temporal coverage of both GRB and GW monitors.
2.1. GRB Afterylows
Even if a short G RB is beamed in a direction away
The collection of short GRB afterglows observed with from the earth, X-ray band emission from the afterglow
Swift and with measured redshifts has been studied by may be visible in the direction of earth. There are no
Racusin et al. (2011). The late-time X-ray afterglows de confirmed observations of such off-axis afterglows, pre
cay with a power law t-O, with a temporal index Q ,.....' 1.5 sumably due to the difficulty in reliably identifying them
and at 1 day after the trigger time, show luminosities in in all-sky survey data. However, some predicted light
the XRl' band ranging from 1042 - 104s ergs/so Plac curves from off-axis afterglows are available from sim
ing these afterglows at a luminosity distance of 200 Mpc ulations by van Eerten & MacFadyen (2011). In order
leads to fluxes of between 10-12 and 10-9 ergs/s/cm2 to compare the simulation results with observed, on-axis
(See Figure 1). The XRT routinely observes these ob light curves, we scale the results to a luminosity distance
jects out to a redshift of z - 0.5, suggesting that the of 200 Mpc, and calculate the XRl' band flux assuming
afterglows from sourcec; ,,-ithin the GW detector horizon a power law SpectrllII! with an index of 1.5 (see Figure
(~400) Mpc would be relatively bright. 1). Comparing off-ax~ light curvE'S with observations of
An important feature of afterglows in this context is GRBs (where the observer is inside the jet opening an
that they are expected to be beamed. This h"" the impli gle) suggests that the off-axis observer will encounter a
cation that only a small fraction, Ib, of NS-NS or NS-BH number of challenges seeking an X-ray counterpart. For
mergers will have a beam pointed towards Earth. For an observer located at twice the jet opening angle, there
small jet opening angles, Ib is related to the jet openiug is a brightening time of between two and twenty days, so
half-angle as Ib - OJ /2. The jet augle is highly uncertain, a wait of at least several days would be needed to observe
3
an off-a.xis afterglow in X-rays. A large time window be - 90 minutes instead of - 12 hours, and the sky coverage
tween the GW trigger and the observation of the coun is nearly total, where a ground based facility has access
terpart would make establishing a connection difficult .. to a smaller fraction of the sky.
filoremfer, the off-axis emission is considerably dimmer The combination of fast response times to target-of
than the on-axis emission, by several orders of magni opportunity (TOO) requests (around 1 or 2 hours), and
tude (See Figure 2). III addition to being more difficult a long heritage with GRB afterglows makes the Swift
to detect, the fainter emiss!on will also be more difficult satemte a natural facility to consider. Much of the fol
to separate from background variability. This suggests lowing discussion could apply to other facilities as well,
that an observation of an off-axis afterglow would require . particularly XMM-Newton and Chandra, however the
an exceptionally nearby NS-NS merger. For a merger at fast TOO response may make Swift the only practical
50 Mpc, the peak flux of an X-ray afterglow observed at choice for seeking quickly fading counterparts. With this
twice the jet opening angle would range _ 10-15 _10-13 in mind, and with the intention of making the discussion
ergs/s/em2, and so could be observable in a 10 ks XIlT as concrete as possible, we fOCllS specifically on the Swift
exposW'e. Such events are likely to be rare, perhaps once observatory.
every 10-20 years based on observed GRB rates. How
3.1. Searching the full error box
ever, they present an interesting possibility, since the
large SNR signal that they Vlould produce in the Ad The first strategy that we consider is using the Swift
vanced LIGO IVirgo network would allow for exceptional XRT to tile an entire LIGO /Virgo error box. Estimates
localization and parameter estimation. for the uncertainty associated with a LIGO/Virgo posi
tion reconstruction vary from a few tens of square degrees
2.2. Kilonovae
to over a hundred square degrees (Abadie et al. 2012b;
While observing the afterglow to a short GRB from Nissanke et al. 2011; Klimenko et al. 2011; Fairhurst
a position outside the beam will likely be challeng 2011). In fact, the precision of any particular position es
ing, another source of emission from NS-NS merg timate will depend on a number of factors, including SNR
ers may create an isotropic transient. Though lack and sky position. Certainly, any position reconstruction
ing in observational evidence, transients known as kilo with the LIGO /Virgo network will cover an area signif
novae have been treated in the literature by sev icantly larger than tbe 0.16 square degree XRT field of
eral a'lthors (Metzger et al. 2010; Roberts et al. 2011; view. ]n order to evaluate the feasibility of search strate
Li & Paczynski 1998; Kulkarni 2005; Goriely et al. gies, we consider 100 square degrees as a nominal value
2011). The model predicts that ejecta from the merger for the LIGO /Virgo position uncertainty.
will grow heavy nuclei through r-process nucleosynthesis, To characterize the ability of the XRT to quickly sur
which subsequently decay and heat the material. The vey a large area, we write the limiting flux of an obser
thermal emission leads to an optical band transient with vation as a function of observi~g time as
a peak luminosity around one day after the merger event, -r
and a dimming over the course of the next few days. The F = 6 x 10-12 (100S) ergs/s/cm2 (1)
transient may have a blue color, with U band emission
that appears brighter and peaks sooner than the R band
where T is the observing time (Moretti et al. 2007). This
emission. Because this mechanism is largely indepen
is valid in the regime of photon limited exposures (T <
dent of the environment around the merger, and leads to
104 s)) and assumes 12 counts are needed for a detection,
isotropic emission, it is possible to imagine that a large
fraction of NS-NS and NS-BH mergers are accompanied with a conversion factor between COl,mt. rate and flux of
by observable kilonovae. A kilonova at 200 Mpc is ex 5 x 10-11• If we imagine that the sought afterglow to a
GW trigger is visible for roughly one day, then we can
pected to peak around magnitude 19-22, bright enough
calculate how much observing time, and so what limiting
to be detected by the UVOT instrument on Swift. UVOT
flux, we can associate with observations covering various
is aligned in parallel to the XRT, so a search O\.-er one or
amounts of area.
two days for X-ray afterglows with Swift is a simultane
ous search for optical band kilonovae. It should be noted
T = (86,400 s) (0.12 deg') _ S (2)
that Swift's capability in searching for optical band tran
3 Area
sients is not uniquej ground-based optical survey tele
scopes can also search large areas to these magnitudes X 14 ( 2) 2
F,., 2 10- Area erg/s/cm (3)
(Metzger & Berger 2012).
0.12 deg
3. SEARCH STRATEGIES WITH SWIFT where S represents the amount of time for each slew and
The large position uncertainty associated with settle of the instrument. Equation 3 assumes the observ
LIGO /Virgo triggers matches well onto the capa ing time is much larger than the total slew time. The
!
bilities of ground based, large area survey projects factor of is to account for the fact that most observa
such as PTF, Pan-STARRS, QUEST, and SkyMapper tions cannot continue over an entire orbit, due primarily
(Metzger & Berger 2012). However, a space-based X-ray to occultation by the earth. In addition, the size of the
facility may present some unique advantages. The X-ray field of view has been reduced to 0.12 deg2 to allow some
afterglov."S to short G RBs are easily distinguished from overlap in the tiling pattern. The resulting limiting fluxes
other X-ray sources both by t.heir large flux and char are plotted in Figme 2 for a range of areas, with S set
acteristic power-law dimming. A -space-based) as com to 30 s.
pared to ground-based, facility also has the advantage Under th.se assumptions, the XRT might take a series
that wa.it time for a source to pass overhead is typically of 290 exposures, each 100 s long. This would cover 35
4
"',~::;::S;:---7=O;;==7,'5;=;;=:O:====:=il
10" .,'_ ~f ~ :~ "4- -.- Flux limit for 100 5 exposUI8
:\~ Flux limit for 1000 5 exposures
l:~ ··· ~~~~·················
i :::: ....". ... ...:' .. ;. ; ::: ..
\:~. ~.:·~,.E~r:::·
K .\.-
~ i ___ i
'"l i:wU .... .............. ~-- ..........,.,.. . _-----, ......_ -
.5 10';) . ",r
-: 10':,0
~
• 1,0.'"" , 1ern lat O20s0 ~M p c
'''1,
,<"·t============~~~~~~~~'~==~
IIf:~O!:;·--~:;---:,:::.Q·;---:,:;Q'~---::,,:r,-...:...~,,'
Tlmetllol;!$,
FIG. 1.-The grAy curves are XRT light curves for short GRBs FIG. 2.- Plot showing the potential for a large area survey
with known redshifts, scaled to 8. distance of 200 !\lpc (Enms ct al. with XRT over the course of a. single day. Enough exposures are
2007, 20(9). The green line shows XRT flux limit (or 100 8 expo taken over the course of 24 hours to CO"~W the area. shown on the
sures, which could cover ,..., 35 square degrees i.n one da.y, and the horizontal axis, under the assumption tha.t the XRT requires 30 s
green line indicatc6 the limit for a 1000 6 exposure. Some o~ of slew and settle time for each image. For comparison, the range
served afterglows quickly fa.de, and so are not observable 1 hour of fluxes of .short GRB afterglows, scaled to 1 day after the trigger
a.fter the burst. However, the ulong-lived" afterglows are generally time and e. luminosity distance of 200 Mpc, is shown as the green
bright enough to be observed 1()"'100 hours after t.be burst, even shaded region. The gray region shows the peak XRI' flux of a range
with short ( ..... 100 $) exposures. The dasbed, colored curves show of off-axis lightcurves, also scaled to 200 Mpc. The peak flux of
predicted light curves for off-:>.Xis light curves viewed at twice the the model off-axis light curves occurs several days after the GRB
jet opening angle, scaled to 200 ?fpc (van Eerten & MacFadyen trigger time.
2011). The simula.ted light curves ha.ve jet energies of 1048 &nd
1050 ~, &nd circumburst medium number densities of 1 em -a
and 10-3 em-a.
2011; Mateos et aL 2008). Demanding variability in the
source should be a strong handle for cutting this num
deg2 in ahout one day, and yield a flux limit of around ber down further. The black dashed curve uses statis
6 x 10-12 ergs/s/cm2. With the current on-hoard soft tics of observed AGN variability to forecast how many
ware, this could be accomplished with eight applications variable AGN are likely to fluctuate by a factor of 2
of the programmed 37 tile pattern. The resulting flux or more between two images (based on numbers found
limit seems to be sensitive enough to find most on-axis in Gibson & Brandt (2011)). The study represented as
afterglows; but not predicted off-axis afterglows for most a blue"triangle found a similar number density by sys
viewing angles. tematically searching the RASS data for variable sources
Clearly, this approach would be less strenuous if the with a "flare" like light-curve (Fuhrmejster & Schmitt
position uncertainty associated with a particular event 2003). The red star shows the result of another RASS
could he reduced. Some estimates do predict that 100 study that used stronger cuts to seek G RB orphan af
square degrees is a conservative estimate, and suggest terglov!s (Greiner et al. 2000)_ The three orders of mag
20-40 square degrees as more typical numbers, depending nitude in density reduction between this point and the
on underlying signal models and algorithm. For exam number of AGN was achieved by demanding that no X
ple, Nissanke et al. (2011) predict that, using a I\larkov ray source was present at the location of the transient
Chain Monte Carlo parameter estimation technique, only either before or aft<>.r the flare event. This seemed to be
36 square degrees of area need to be searched to recover a very powerful cut, and resulted in finding around one
70% of binary mergers detected with an SNR of at least source in every 10,000 square degrees, or a 1 % chance of
six at all three detector locations, or 12 square degrees finding an unrelated afterglow like source in association
could be searched for a 50% recovery rate_ The addi with the 100 square degree GW error box. When seeking
tion of a fourth GW detector to the network could also only afterglows within a limited distance range, it is pos
improve localization ability to around 10 square degrees. sible to remove the majority of flare stars by demanding
KAGRA (Kuroda & LCGT Collaboration 2010) located a spatial coincidence between the transient source and
in the Kamioka mine in Japan, could be operational by an optical galaxy. This suggests that a transient, soft
2018, and a third LIGO site in India 2 could be opera X-ray source, brighter than ~ 10-12 ergs/s/cm2 found
tional by 2022. in connection to a GW trigger is likely to be associ
An important consideration in searching a large area ated. The expectation that some LIGO /Virgo triggers
for a single event is to understand jf we can distinguish will have short-lived, soft X-ray counterparts brighter
the event from other sources. For the case where the er than anything else within the error box suggests that
ror box is out of the galactic plane, some suggestive num a wide-field, focusing instrument in this band would be
bers are shown in Figure 3. At a sensitivity of 2 x 10-12 a useful follow-up tooL
erg/s/em2, a typical area of 100 square degrees would A LIGO /Virgo trigger in temporal coincidence with a
find only a few extragalactic sources (Puccetti et aL Fermi GBM (Meegan et al. 2009) trigger would be of
great interest. GBM sees a large fraction of the sky
2 https:/jdcc.ligo.orgjcgi-binjDocDBjShowDocument?docid=91470 (~ 65%) (Meegan et aL 2009), and so this is a likely
5
•
,,' •
·>~~i.
!
H~ ",,~, ~.
::: .... -.. .............. ..... '.
~ ."~~<~...
I r.-.~X=.,.~S.. ".- ""'-y-O~.~5-~2' """.,.V,-----, 1.. .. ...
lC~ '*-"* Flttosurveydata .:. ......... ~ ... "
Aam. with no DC scuce (RASS)
10" J... Any flare (RASS)
AGN Vllriabllty
It);Q'" lO'u lcr" 19-U
S!'1uII l.rg em .... ,-. I
Error radius (degrees)
FIG. 3.- The extragalactic X-ray background number density
estima.t.ed in '.'arious ways. The black points show the ob6~ved FIG. 4.- A distribution of estimated error radii for 88 bursts
number density at high galactic latitude in the Swift serendipItous {rom the Fenni GBM Burst Catalog. The bursts'&:re selected only
survey, with a. fit shown in green (Puccetti et~. 2011). The black with the criteria. that Too < 2 s. The x-axis is in deg~. the y-axis
dashed line is an estimate of the number of variable AGN that may is the number of bursts in the bin. Many of the bursts in the 0 bin
be mist&ken as transients, 8.58uming a. reference image that is twice ha.ve position information that is known ffom some other source,
the depth of the limit shown on the x-axis Gibson & Brandt (2011). such as the Fermi Large Area Telescope or Swift·
The triangle marks the density of variable "flare like" sources in
the ROSAT All-sky survey (Fuhrmeister & Schmitt 2(03), and t~e
red star marks the density of candidate orphan afterglows found In
a. ROSAT data (Greiner et a.1. 2000). to, their host galaxies. In fact, Berger (2011) has fo~nd
that short GRBs seem to track the total mass of galIDCles.
scenario. The GBM has a large uncertainty in its posi So, in response to a LIGO jVirgo trigger. it may be possi
tion reconstruction (Briggs et al. 2009). For example, in ble to search only the locations of known galaxies within
the GBM burst catalog (Paciesas et a1. 2012), the listed a. fixed distance horizon rather than the entire eITor box.
short bursts have a median positional uncertainty of 7.5 This method was applied in efforts to search for EM af
degrees. Some of this error (~ 3 degrees) is due to sys terglows to GW triggers using the initial LIGO jVirgo
tematics in the localization process (Hurley et a1. 2011; network (Kanner et al. 2008; Abadie et al. 2012b). For
Paciesas et a1. 2012). the 2009-2010 search, a galaxy catalog was constructed
A GW trigger in coincidence with a Fenni GRB ob from publicly available information, known as the GW
servation would definitively establish the progenitor of Galaxy Catalog (GWGC)(White et a1. 2011). This ap
the burst. However, the large error radius associated proach takes advantage of the limited distance reach of
with Fermi means that a GBM trigger alone could not the GW instruments to merger events, in the sense that
provide a host galaxy identification or provide the coor the sk~T is relatively sparse in galaxies to a limited dis
dinates of the potential afterglow. Given the high degree tance range. In the limit of a GW detector that could
of interest in such an event, seeking the afterglow and observe events anywhere in the observa.ble universe, this
host galaxy s..,ms well wOIth the effort. The range of would clearly not be the case, and the density of observ
typical GB!>.1 position uncertainties (3 - 12 degrees, see able galaxies on the sky would make a galaxy catalog
Figure 4) over1aps the range of positional errors with ineffective in limiting the amount of sky area to be ob
LIGOjVirgo, with a median position uncertainty of 7.5 served. The question that we ask is: out to what distance
degrees, or 176 deg2 under the assumption of a circular· reach is a galaxy catalog still useful in this way with the
region. While these uncertainties may often be larger Swift XRT?
than the GW position uncertainties, in the early days of A similar question was addressed by Nuttall & Sutton
advanced detectors, it is possible that the thr.., GW de (2010), who found that the galaxy catalog can be an
tectors will have unequal sensitivities. This would lead to extremely useful tool in recovering the true location of a
GW error boxes that are very spread out on the sky. For GW inspiral signal out to at least 100 1Ipc, even if only
example, if only two detectors are operating at the time a few galaxies are imaged. Here, we attempt to find the
of an event, then the GW network will localize the event limiting range where searches ~lith and without a galaxy
to a ring on the sky of order 1000 square degrees. In such catalog require the XRT to observe the same amount of
a scenario, the Fermi error ellipse would be more con sky area.
straining the LIGOjVirgo uncertainty region. It is also Past counterpart searches using a galaxy catalog as
possible to imagine taking the intersection of the GBM sumed that the likelihood of a merger event in the galaxy
error ellipse and the LIGO jVirgo skymap. traces either the mass or the star formation rate of the
gaiaxy (Abadie et a1. 2012b,a). Competing models ex
3.2. Searching with a gal=u catalog ist for which types of galaxies are more likely to host
The fact that the LIGO jVirgo network is sensitive to merger events (O'Shaughnessy et al. 2010). In order to
only a limited distance can be used to dramatically re explore the implications of various models, we have used
duce the amount of area that must be searched for an E:r...l the HyperLeda database to add measurements of I-band
counterpart. Merger events should occur within, or close (near infrared) luminosity to the GWGC. This resulted
6
.. ,
10.... 10'" 10"" 10·'
Frxa"" oI'_berolgal.oxietl in the 100 Mpccat<llog FIG. 6.- The GWGC contains 53,000 galaxies within 100 Mpc
of earth. The figure assu!nes that the number of galaxies within
FIG. 5,- The GWGC contains roughly 53,000 galaxies within
100 Mpc of earth. The plot shows what fraction of the total Dumber a horizon distance r scales as rJ, and that the catalog is 60%
complete. Within a 100 Mpc horizon distance, imaging the galaxies
of gala.xies must be selected in order to obtain a target fraction
needed to contain 00% of the I-band luminosity within a 100 square
of the total luminosity or mass in the catalog. Only 10% of the
degree LIGO/Virgo error box requires an order of magnitude less
galaxies contain 50% of the, I-band luminosity. Including 90% of
pointings of the XRT than the number ,required to tile the whole
the blue light luminosity requires 40% of the number of gaJa.xies.
LIGOjVirgo error box. The limit where a galaxy catalog is no
The distribution of the color adjusted mass estima.te suggests that
longer useful seems to be around 200 Mpc. For comparison, the
even a smaller fraction of galaxies in the ca.ta.log ma.y contaln a.
design curve for Advanced UGO predicts a sky average NS-NS
given fraction of the total m8.S.d, however, this ma.y be an effect. of
average range of around 200 Mpc. Initial LIGO had a sky average
color scatter due to metalHcity.
range of around 20 Mpc for NS-NS mergers.
in a catalog with 51,136 objects with B-band measure pleteness fraction of the catalog, this means a 100 deg2
ments, 34,363 objects with I-band measurements, and LIGO /Virgo error box with a 100 ~Ipc range could be
31,732 galaxies with both I-band and B-band measure covered at the 90% level by observing ~ 90 galaxies, or
ments. The B-band luminosity is estimated to be ~ 60% at the 50% level by observing ~ 20 galaxies. This repre
complete (White et al. 2011). Conventional wisdom sug sents dramatically fewer pointings of the Swift XRf than
gests that wavelengths in the near infrared should be the 800 fields required to tile the whole error box. So,
good tracers of total stellar mass. Hov.-ever, the applica for a search with a range of 100 Mpc, the galaxy catalog
tion of II. color correction has been shown to improve mass still seems to be a useful tool. This suggests that, in the
estimates in some cases (Bell & de Jong 2001; Bell et al. early months of Advanced LIGO /Virgo, when horizon
2003). distances are expected to be below the design goals, a
Adopting the model presented in Bell & de Jong reasonable strategy would utilize exposures of roughly 1
(2001)1 we constructed a color corrected. mass estimate ks, allowing the XRT to image 2 g.alaxies per orbit, and
using the I-band magnitude and IB - I] color for each so observe 32 galaxies in a 24 hour period. Repeating
galaxy in our sample with both measurements. We ap the observations over a 2nd, not necessarily concurrent,
plied the model as day would allow image subtraction with the UVOT to
10glO(M/L) = -0.88+0.60IB- 1] (4) enable & search for kilonovae. However, it is important
to note that this scenario is limited to the assumption of
where M and L are the galactic mass and I-band lu a NS-NS merger: BH-NS mergers have horizon distances
minosity, both in solar units. Bell et al. (2003) showed that are greater by roughly a factor of two. Moreover,
that, in near infrared wavelengths, this color dependency it is important to distinguish between the sky-averaged
is likely too steep, but did not provide corrected values range distance, which we have used here, and the optimal
for I-band measurements. For this reason, it is likely that horizon distance, which differ by a factor of 2.26.
our color corrected mass estimr.te amplifies the effects of In the case ofa very nearby source « looMpc), it may
color sca.tter due to metailicity variations between galax be possible to limit observations to only a few galaxies.
ies, and so results in an artificial broadening in the distri In such cases, the search might be optimized for off-axis
bution of masses in the catalog. The effect was partially afterglows by taking longer, ~ 10ks exposures of each
mitigated by removing galaxies with colors far from the galaxy, over a span of a few dals. This would result in
center of the distribution. flux limits of around 2 x 10-1 ergs/s/cm?, and so be
In the GWGC, out to 100 Mpc there is roughly 1.3 deep enough for the more optimistic off-axis afterglow
possible host galaxies per square degree, or 130 possi models.
ble hosts for a typical LIGO /Virgo error box. How No galaxy catalog is currently complete to 200 !vIpc,
ever, if we take the position that only enough possible although there are efforts underway to obtain one by
hosts need to be imaged to make including the true host the time of Advanced LIGO (KasliwaI2012). Using the
likely, we can reduce this number further. Figure 5 shows GWGC, a geometric scaling allows us to anticipate the
that, averaged over the sky, 40% of galaxies contain 90 result for a 200 Mpc horizon distance (See Figure 6. At
% of the I-band luminosity. After adjusting for the com- 200 Mpc, if we still hope to capture the true source with
7
a 9Oo/c likelihood, and assuming a similar distribution of likely be applicable under different sets of circumstances.
galactic luminosities as those alread!, contained in the Duriug the early years of advanced gravitational wave
GWGC, then the requirement is to obsen-e around 700 detectors, around 2015-2018, Advanced LIGO and Ad
galaxies per LIGOjVirgo trigger. This number is similar vanced Virgo are likely to operate at sensitivities less
in scale to the 800 fields required to tile the whole er optimal than their design curves. The sky-average range
ror box, so we conclude that the limit of usefulness of a for NS-NS mergers will be perhaps 50 - 100 Mpc. Un
galaxy catalog for 24 arcminute fields is around 200 Mpc der these circumstances, the large (~ 100 square degree)
(See Figure 6). position uncertainty associated with a LIGOjVirgo error
Another important concern related to the application box may be dramatically reduced through the use of a
of a galaxy catalog is the possibility that NS-NS or NS galaxy catalog. The Swift observatory could then search
BH mergers will occur outside of their host galaxies. for an X-ray counterpart by imaging a few tens of galaxies
Asymmetries in the supernova that forms a compact ob over the course of a day, with exposures around a kilosec
ject can impart a net momentum, OT "natal kick" , to the ond. With this procedure, anyon-axis afterglow should
resulting NS or BH. If the kick is large enough to unbind be detectable, and so it should be possible to make a de
the system from the gravitational potential of its host tection, or else place limits on the beaming angle of short
galaxy, then the binary can drift outside the host galaxy GRBs. Moreover, while the XRI' searches for an after
before merger. In fact, Fryer & Kalogera (1997) found glo,,·, the UVOT will simultaneously obtain data across
evidence for kicks around 200 km/s in observed neutron the band where kilonova emission is expected, imaging
star binaries with separation distances small enough to down to around magnitUde 22, sufficiently deep to image
allow mergers within a Hubble time. Considering where a kilonova at 100 Mpc. Kilonovae are expected to emit
mergers may be observed in relation to their host galax isotropically, and so could be observable even for off·axis
ies, Kelley et a1. (2010) found that if even larger kicks merger events, suggesting that this observable could ac
are assumed, 360 km/s, then nearby mergers will be company LIGO /Virgo triggers more often than afterglow
rv
observed up to 1 Mpc away from their host. However, emission. During this period, the "best guess" estimate
other investigators find these large kicks to be unlikely, for LIGOjVirgo detected mergers is only a few per year,
and tend to favor models where typical mergers occur so it seems plausible to follow-up every high significance
within 1 to 100 kPc of the host galaxy (Fryer & Kalogera trigger in this manner.
1997; Brandt & Podsiadlowski 1995; Bloom et a1. 1999; As the GW detectors mature and reach their design
Fryer et a1. 1999; Belczynski et a1. 2006). Observational sensitivity, the galaxy catalog is likely to become a less
evidence also supports the notion that most mergers oc useful tool. This means that, instead of representing the
cur within 100 kpc of the host galaxy. An attempt to error box with,.... 30 pointings, it will be necessary to
match "hostless" short GRBs with nearby galaxies found use hundreds of XRT tillngs to cover the eITor box. To
that, most of the observed short GRBs with known red complete such an ambitious observing plan in a limited
shift very likely occurred within 100 kpc of their host period of time reqaires sacrificing sensitivity, and the 100
galaxy (Berger 2010). In the same work, a comparison s observations' would be only barely deep enough to de
between the distribution of these observed short GRB off tect on-axis afterglows, and would likely be unable to
sets from their host galaxies was found to be consistent detect kilonovae. On-axis afterglows are only expected
with p:edictions from models of NS-NS merger locations. to accompany ~ 10% of LIGOjVirgo detections, and the
The Swift XRT has a FOV of 24 x 24 arcminutes, and detection rate in this period oould be ~ 40 per year. Un
the UYOT has a FOV of 17 x 17 arcminutes. If a merger der these circumstances, it seems unreasonable to expect
occurs at a distance"o f 100 Mpc from earth, then an an orbiting facility to follow-up every GW trigger. On
XRT (UVOT) observation centered on the host galaxy the other hand, all-sky GRB monitors, most notably the
will observe the counterpart for any source within 300 Fenni GBM, continuously observe a large fraction of the
kpc (200 kPc) of the host galaxy. This should include sky, and so effectively select which LIGO /Virgo triggers
essentially all NS-NS mergers. Even as close to earth are most likely to have soft-band X-ray afterglows. Coin
as 50 :vIpc, is seems unlikely the merger site should be cidences between a mature Advanced LIGOjVirgo net
outside either FOV. Only cases where the kick velocity is work and GBM should occur at the one per year level, an
extremely large, or the host galaxy's gravitational well is estimate that comes from the observed GRB population,
much smaller than the Milky Way, allow mergers beyond and so is independent of large uncertainties in population
this distance. synthesis or the GRB jet opening angle (Coward et al.
2012; Chen & Holz 2012). A coincidence between GBM
4. CONCLUSIONS and the LIGOjVirgo network will be an exciting event,
The high fraction of short GRBs with X-ray band after and so obtaining the precise position, host galaxy, and
glows, and the potentially bright fluxes associated with red shift information only obtainable through an after
them, make the Swift X-ray band an attractive wave glow observation will be well worth the effort. Once the
length to seek EM counterparts to NS-NS and NS-BH LIGO /Virgo network reaches its full design sensitivity,
mergers. An imaging, wide field, soft X-ray band detec with an average NS-NS inspiral horizon of around 200
tor with a fast response to TOO requests is required. In Mpc, a sensible plan with the Swift X-ray observatory
the best case, the X-ray facility FOV would be at least 3 would be to only respond to triggers in coincidence with
degrees wide to match the scale of LI G 0 jVirgo position a GRB observation. In this era (~ 2018), it is even pos
uncertainty. However, given the full range of req uire sible that a fourth GW detector site will be operational,
ments, the Swift satellite seems to be the strongest can and so reduce the sky area that needs to be searched.
didate facility which is currently in operation. This pa
per discussed two possible search strategies, that would
8
We are grateful for fruitful discussions and feedback Universities through a contract with NASA. This work
from Lindy Blackburn, David Burrows, Thomas Dent, made use of data supplied by the UK Swift Science Data
Brennan Hughey, Susan Kassin, Takanori Sakamoto, and Centre at the University of Leicester and data obtained
Peter Shawhan. JK is supported by an appointment from the High Energy Astrophysics Science Archive Re
to the NASA Postdoctoral Program at Goddard Space search Center (HEASARC), provided by NASA's God
Flight Center, administered by Oak Ridge Associated dard Space Flight Center.
REFERENCES
Abadie, J., et a!. 2010, Classical and Quantum Gra.vity, 27, Greiner, J., Hartmann, D. H., Voges, W., Boller, T., Schwarz, R.,
173001 & Zharikov, S. V. 2000, AA, 353, 998
-, 2012&, Astronomy & Astrophysics, 541, A155 Harry, G. M., et ai. 20ID, Classical and Quantum Gravity, 27,
-. 2D12b, Astronomy & A"trophysics, 539, A124 084000
- . 2012c. ArXiv e-prints, arXiv:1205.2216 Hurley, K, et aI. 201j, Preprint, arXiv:1110.6470
Abbott, B. P., at aJ.. 2010, ApJ, ns, 1438 Kanner, J., Huard, T. L., Marks, S., Murphy, D. C., Piscionere,
Accadie., T., et at 2012, Journal of Instrumentation, 7, 3012 J., Reed, M:, & Sha.whan, P. 2008, Classical and Quantum
Ba.rthel:ny, S. D., et aI. 2005, Space Science Reviews, 120, 143 Gravity, 25, 184034
Belczynski, K., Perna, R" Bulik, T., Kalogera, Y., lvanova, N., & Kasliwal, M. 2012, Private Communication
La.mb, D. Q. 2006, ApJ, 648, 1110 Kelley, L, Z., Ramirez..Ruiz, E., Zemp, M., Diem811d, J., &
Beli, E. F., & de Jong, R. S. 2001, ApJ, 550, 212 Mandel, I. 2010, ApJL, 125, L91
Bell, E. F., McIntosh, D. H., Katz, N., & Weinberg, M. D. 2003, Klimenko, S., et al. 2011, submitted to Phys. Rev. D
ApJS, 149, 289 Kulkarni, S. R. 2005, Preprint, arXiv:astro-ph/05ID256
Berger, E. 2010, ApJ, 722, 1946 Kuroda., K., & LCGT Collabora.tion. 2010, Classical and
- . 2011, New Astronomy Reviews, 55, 1 Quantum Gravity, 27, 084004
Bloom, J. S., SigurdBson, S., & Pols, O. R. 1999, MNRAS, 305, Li, L.-X., & P8CZyDski, B. 1998, ApJL, 507, L59
763 Mateos, S., et aJ. 2008, Astronomy & Astrophysics, 492, 51
Brandt, N., & Podsiadlowski, P. 1995, MNRAS, 274, 461 Meegan, C., et al. 2009, ApJ, 702, 791
Briggs, M. S., Connaughton, V., Meegan, C. A., Wilson-Hodge, Metzger, a. D., & Berger, E. 2012, ApJ, 746, 48
a.
C., Kippen, M., & Hurley, K. 2009, in American Institute of Metzger, D., et al. 2010, MNRAS, 406, 2650
Phys:cs Conference Series, Vol. 1133, American Institute of Moretti, A., et a1. 2007, in Society of Photo-OptlcaJ
Physics Conference Series, ed. C. ~Ieegan, C. Kouveliotou, & Instrumentation Engineers (SPIE) Conference Series, Vol. 6688,
N. Gebrels, 40-42 Society of Photo-Optical Instrumentation Engineers (SPIE)
Burrow~, D. N., et aJ. 2005, Space Science Reviews, 120, 165 Conference Series
Chen, H.-Y., & Holz, D. E. 2012, Preprint, arXiv:1206.0703 Nissanke, S., Holz, D. E., Hughes, S. A., Dalal, N., & Sievers,
Coward, D., et at. 2012, ArXiv e-prints, a.rXiv:1202.2179 J. L. 2010, ApJ, 725, 496
Coward, D. M., et aI. 2011, 1-1NRAS, 415, L26 Nissanke, S., Sievers, J., Dalal, N., &. Holz, D. 2011, ApJ, 739, 99
Dalal, N., Holz, D. E., HughM, S., & Jain, B. 2006, PRO, 74, Nuttall, L. K., &:: Sutton, P. J. 20]0, Physical Review D, 82,
063OC6 102002
Evans, P. A., ct al. 2007, Astronomy & Astrophysics, 469, 379 O'Shaughnessy, R., Kalogera, V., & Belczynskl, K. 2010, ApJ,
- , 2009, MNRAS, 397, 1177 716, 615
- . 2012, ArXive-prints, arXiv:1205.1124 Paciesas, W. S., et al. 2012, ApJS, 199, 18
Fairhunt, S. 2011, Clas.sical and Quantum Gravity, 28, 105021 Piran, T., Nakar, E., &:: Rosswog, S. 2012, Preprint,
Fbng, W.-f., et at. 2012, ArXiv e-prints, arXiv:1204.5475 arXiv:1204.6242
Fox, D. B., et aI. 2005, Na.ture, 437, 845 Puccetti, S., et al. 2011, AA, 528, A122
Fryer, C., & Kalogera, V. 1997, ApJ, 489, 244 Racusin, J. L., et ai. 2011, ApJ, 738, 138
Fryer, C. L., Woosley, S. E., & Hartma.rul, D. H. 1999, ApJ, 526, Roberts, L. F., Kasen, D., Lee, W. H., & Ra.mirez-Ruiz, E. 2011,
152 ApJL, 736, L21
Fuhrmeister, B., &. Schmitt, J. H. M. M. 2003, AA, 403, 247 Roming, P. W. A., et at. 2005, Space Science Reviews, 120, 95
Gehrels, N., et aI. 2004, ApJ, 611, !005 Schutz, B. F. 2011, Classical and Quantum Gravity, 28, 125023
-.200.5, Nature, 437, 851 van Eerten, H. J., & MacFadyen, A. J. 2011, ApJ Letters, 733,
Gibson, R. R., & Brandt, W. N. 2011, ArXive-prints L37
Goriely, S., Bauswein, A., & Janka, H.-T. 2011, ApJL, 738, L32 White, D. J., Daw, E, J., & Dhillon, V. S. 2011, Classical and
Quantum Gravity, 28, 085016