The nature of the DLS fast transients

Draft version February 5, 2008

Preprint typeset using I^T^X style cmulatcapj v. 12/14/05

O
O
(N

u

a:
<

(N

>

m

o
\o
o

^«

Or
I

o

13

THE NATURE OF THE DLS FAST TRANSIENTS

S. R. KULKARNI AND A. RAU

Caltech Optical Observatories 105-24,

California Institute of Technology, Pasadena, CA 91125, USA

Draft version February 5, 2008

ABSTRACT

The discovery and study of highly transient sources, especially those which rise to high brightness
and then fade to obscurity, has been a major part of modern astrophysics. Well known examples
include supernovae and novae. A byproduct of the Deep Lens Survey was the discovery of three
transients which varied on a timescale of less than an hour. All three had faint and red counterparts,
the brightest of which was identified with an M star. However, the remaining two showed hints of an
extragalactic origin, one had a spatially extended counterpart and the other appeared in projection
on the outskirts of a bright elliptical galaxy. If these two sources were really of an extragalactic origin
then the two events represent a new class of exotic explosive transients. We undertook spectroscopic
observations with the Keck telescope and find the two counterparts are also late type Galactic dwarfs.
Our main conclusion is that flares from M dwarfs constitute a dense foreground fog and dominate over
any plausible class of extragalactic fast transients by at least two orders of magnitude. Overcoming
this fog will likely require dedicated surveys with careful optimization of target field location, filter(s)
and cadence, pre-search imaging to filter out late type dwarfs and a well planned rapid followup plan.
Subject headings: stars: flare - subdwarfs - surveys

1. BACKGROUND

In the first half of the twentieth century the search for
and study of variable sources was a major focus of optical
astronomy. This area is widely expected to undergo a
renaissance, thanks to significant technological changes
(sensors, computing, storage and dissemination). Apart
from existing wide-field imagers dedicated facilities such
as the Panoramic Sur vey Telescope fc R apid Response
System fPan-STARR.S: lKaiser et all2002ft and the Large
Synoptic Survey Telescope (LSST) are motivated in part
to explore transient phenomena in the optical sky.

An excellent precursor of the planned optical searches
for transients is the Deep Lens Survey 1 (DLS) . This sur-
vey is notab le for its comprehen sive scope and almost real
time alerts llBecker et al.ll2004|) . The Deep Lens Survey
netted three fast 1 transients. The quiescent of the bright-
est transient was identified with an M dwarf and the tran-
sient itself attributed to a flare from the M dwarf. Of the
remaining two, the quiescent counterpart of one was spa-
tially extended and the other counterpart coincided with
a bright elliptical galaxy). Becker et al. (2004) derive de-
tection rates of 14, 10 and < 5 deg -2 day -1 for transients
on timescales of about 1900 s in the B, V and i?-band,
respectively.

A byproduct of the 2.2-m MPG/ESO survey for orphan
afterglow was a search for fast transi ents over a sma ll
piece of sky (exposure of 0.24 deg 2 davs: lRau et alJ 2006'l.
From this survey we derive a 3-c upper limit of 28 events
deg -2 day -1 for any class of fast transients in the R
band (excluding flares from M dwarfs), consistent with
the DLS rate for fast transients.

The two DLS fast transients, if real, constitute a new
class of extragalactic transients. The focus of this Let-

1 http://dls.physics.ucdavis.edu/

2 Following the terminology of X-ray astronomy we refer to tran-
sients which change on timescales of an hour or less as fast tran-
sients.

ter is the investigation of fast optical transients. Here
we report on imaging and spectroscopic observations of
the two quiescent counterparts to the two DLS fast tran-
sients.

2. THE DEEP LENS SURVEY TRANSIENT SEARCH

The Deep Lens Survey was motivated by considera-
tions of weak lensing. During the period 1999-2005 five
2° x 2° fields were repeatedly observed on the CTIO-
Blanco & NOAO-Mayal 4-m telescopes with exposures
from 600 s to 900 s in the B, V, R and z'-bands. The
repeated visits to these fields made the survey attractive
t o identify fast transi ents.

Bec ker et al.l 1)20041) reported the analysis of a transient
search based on 14 runs conducted during the period
2000 November and 2003 April. As with other transient
searches these authors found that the largest number of
transients were moving solar system objects, followed by
variable stars and finally, supernovae (SNe).

Three fast transients remained after filtering out the
above mentioned contaminants. OT 20020115 was de-
tected as a bright source, -B=20.73mag, with a turn-on
time of less than 700 s. It faded rapidly following a power
law index of 1.5< a <2.4. Spectroscopy established this
to be a Galactic dM4 star, presumably a UV Ceti type
flare star at a distance of ^350 pc.

OT 20010326 rose from obscurity to 5=22.7 in less
than 700 s. It faded on a timescale of 1000 s with
0.8 < a < 1.2. At the position of the transient a compact
(not resolved in an archival HST F606W image) persis-
tent object with V ~ 24.5, R ~ 23.3 and z' ~ 21.2 was
found. This source resides (in projection) on the out-
skirts of the elliptical galaxy PKS 1358—11, a member of
the galaxy cluster A 1836 at z = 0.037. If OT 20010326
does lie in A 1836 then the quiescent counterpart could be
a bright globular cluster associated with PKS 1358—11 or
a faint galaxian member of A 1836. Radio observations
were conducted four days later resulting in a limit of

-0.1 ± 0.3 mJy in the 8.5-GHz band.

OT 20030305 rose in less than 700 s and then doubled
in flux in less than 700 s to V — 21 mag. The transient
dropped by 1.5 mag over the next 700 s, and was found
to brighten again in the _B-band by 0.5 mag where it re-
mained nearly constant for at least 1000 s. A post-cursor
was de tected at _R^24.6 and z' ~ 21.4 mag. iBecker et al.l
(2004) claimed an apparent ellipticity of the post-cursor.
If so, it is tempting to associate the quiescent object to
be a distant galaxy.

If the extragalactic nature of OT 20010326 and
OT 20030305 could be established then astronomers have
uncovered a new class of explosive transients with a stag-
gering combined annual all-sky rate of Rp ~ 10 8 yr _1 .
The red color of the putative host galaxies would mean
that the progenitors belong to the old population (cf.
novae, short hard bursts, supernovae of type la) as op-
posed to long duration GRBs which arise only in young
populations.

3. WHY ARE DLS FAST TRANSIENTS INTERESTING?

An annual all-sky rate of fast transients, R? of 10 8 yr" 1
means three events per second. The best way to appre-
ciate the potential importance of the DLS candidates is
the comparison of this rate with the local (z — 0) rates
of known transients (see Table EJ) •

TABLE 1
Local (z=0) rates of known transients.

TABLE 2
Log of the LRIS observations.

Type

Rate
[Gpc- 3 yr" 1 ]

Ref.

long soft GRBs

~30
~ 5 x 10 4
10 - 10 5

~10 8

Guctta ct al. (2005)

core-collapse SNe
short hard GRBs

CaDDellaro et al. ("19991
Nakar et al. ("20051

novae

see below

The observed rate of DLS fast transients exceeds the
z = rates if GRBs (short and long) and core-collapse
SNe by several orders of magnitude. Allowing for larger
distances, one must appeal e.g., to core collapse SNe out
to a redshift of ~ 1. However, the short duration of
OT 20010326 and OT 20030305 make little sense for a
supernova origin. The DLS transients are also unlikely
associated with long duration GRBs. The z-integrated
rate of these events is ~ 10 5 yr -1 .

We can also compare the DLS fast transient rate to
novae. These explosions have the advantage of being less
expensive (relative to the above catastrophic or one-shot
explosions) owing to recurrence.

At z = 0, the mean K s -band volume tric density is
5.7 x lO 8 /iL 0(K) Mpc- 3 {Cole et alJl2001(> : we set h, the
ratio of the Hubble's constant to lOOkms^Mpc -1 , to
0.7. The nova production rate ex pressed in the usual
units is about 3 x 10~ 10 L

HWilliams fc Shafted
suggest that the spe-

l200l . Idella Valle et~af~l Il99"l
cific incidence of novae is higher in star-forming galax-
ies. The local volumetric nova rate is at least pn ~
10 8 Gpc~ 3 yr" 1 and could be three times higher.

If the DLS transients were associated with to some
kind of unusual novae then the mean distance would be
[(3i?jr)/(47r / 9 A f)] 1 / 3 ~ 0.4 Gpc. At this mean distance
they were as bright as supernovae but with rise (or fall

Date

Mode

Setup

Exposure

[s]

Seeing

[arcsec]

OT 20010326:

2004 April 22

Sp
Sp

400/3400
400/8500 (7830 A)

2x1200
2x1200

0.9
0.9

OT 20030305:

2006 Feb 23

hll

B

5x330

1.0

Im

R

4x300

1.0

2006 Feb 24

Im

B

5x330

1.0

Im

B

10x150

1.0

Im

R

6x300

1.0

Im

I

10x150

1.0

Sp
Sp

400/3400
400/8500 (8300 A)

3 x 1830
3 x 1800

1.0
1.0

Note. — The first column is the UT date of the observation.
The mode (imaging (Im) or spectroscopy (Sp)) is indicated in the
second column. In column three we list the imaging filter (if imag-
ing mode) or the grism/grating (number of grooves/blaze angle)
and (for the red grating) the central wavelength. Golumns four
and five give the exposure sequence and the full width at half max-
imum in the R-band.

time) much shorter than supernovae and even smaller
than that of the fastest novae ever known. In this frame-
work, the DLS fast transients form an entirely new and
truly exotic class of cosmic cataclysmic events.

Unfortunately, a prosaic explanation is that
OT 20030305 and OT 20010326 are simply flares
from M dwarfs in our Galaxy. Motivated to resolve this
great ambiguity we undertook observations with the
Keck telescope of the counterparts of OT 20030305 and
OT 20010326.

4. KECK OBSERVATIONS OF OT 20010326 AND OT 20030305

Our observing program was performed with the
Low Resolution Imaging Spectrograph (LRIS: lQke et alJ
1995) mounted on the Keck-I 10-m telescope and in two
separate runs (see Tabled}. LRIS as suggested by the
name, has both an imaging and a spectroscopic mode.
The light beam is split by a dichroic and images/spectra
can be obtained in the Blue and Red channels simulta-
neously.

4.1. OT 20010326

Owing to the faintness (i?~23.3) and red color (V —
R = 1.2 mag) the blue channel (A < 5600 A) spectrum is
of little value. The red channel spectrum shows strong
absorption features which we identify with TiO and CaH
bands (Figure ^). A narrow emission feature close to the
Ha wavelength is detected. We identify this with Ha.
The line is unresolved with a full width at half maximum
(7±2 A) consistent with the instrumental line width. The
centroid is blue shifted with respect to the local standard
rest frame by 140±20 km s _1 . The measured Ha line flux
is (2.3 ±0.3) x lO-^ergcm^s- 1 .

4.2. OT20030305

We undertook deep observations of this source. The
red spectrum shows a strong emission feature which we
identify with Ha (Figure EJ. Fortified with the Ha de-
tection we were also able to detect H/3 emission in the
blue part of the spectrum (Figure [21 middle and right

4

"

i

-

•<

TiO

^

7 3

in
w

1 2

—

- CaH
TiO

TiO [

TiO

1/

/:

ao

CD

M

5 1

1

o

\kJ

wr/

-

£

i

:

6500 7000 7500 8000

Wavelength (angstroms)

Fig. 1. — LRIS spectrum of the quiescent counterpart of
OT 20010326. The Ha emission line and CaH+TiO absorption
bands are marked. The spectrum is binned by a factor of 25 to
emphasize the broad absorption features.

panels). The full width at half maximum of the Ha line
is 7 ± 0.8 A, comparable to the instrumental resolution.
The line centroid corresponds to a velocity with respect
to the local standard rest frame of —5 ± 10kms~ . The
Ha and H/3 line fluxes were measured as (2.6 ± 0.2) x

1-17,

10~ i7 crgcm~ 2 s _1 and (1.1 ±0.2) x 10 _i ' ergem - " s"
respectively.

The imaging data were calibrated with respect to the
DLS source catalog and verified using the colors of stars
in the field compared to colors derived fro m the syntheti c
stellar spectra included in the library of Pickles (1998).
The quiescent counterpart was not detected in the B-
band down to a limiting magnitude of B = 26.6 mag but
detected in the redder bands with R = 24.54 ± 0.10 mag
and / = 22.17 ± 0.07 mag. As the seeing was moderate
during our observations and the imaging data suffered
from additional instrumental distortions, we are unable
to confirm (or disprove) the claimed ellipticity of the
source.

5. DISCUSSION & CONCLUSION

In the Deep Lens Survey Transient Search three fast
transients with rise or fall time of ^1000 s were discov-
ered. Surprisingly all three were detected in the 5-band
and none were seen in the i?-band. Similarl y, all three
events had quiescent red counterparts. iBecker et al.l
(2004) performed subsequent spectroscopic follow-up ob-
servations of the brightest transient and identified it as
a flare from a Galactic M dwarf.

Here we reported Keck spectroscopic and imaging ob-
servations of the quiescent counterparts of the remaining
two events. The spectrum of OT 20010326 shows strong
TiO absorption band heads and is thus an M star. The
spectrum of OT 20030305 shows Ha and H/3 emission
lines and is consistent with a Galactic origin (no signif-
icant local standard rest frame velocity). The emission
lines and the red color of the quiescent counterpart are
best interpreted as arising from a Galactic M dwarf.

Assuming a distance of about 1 kpc for the
OT 20030305 and OT 20010326 (based on their appar-
ent R magnitude) the flare energy release is 5 x 10 33 erg.
This is comparable to the energy rel ease from the UV-
Ceti variable YY Gem l|Moffettill974|) . However as (rea-

sonably) pointed by IBecker et alJ l)2004|) one expects that
high-altitude (z) M dwarfs are not expected to possess
strong activity (owing to their older age) . It may well be
that a fraction of late type dwarfs retain activity for a
longer period.

Alternatively, the late type counterparts discussed here
may have had activity induced later in their life. Indeed
we know of several classes of detached binaries belonging
to the old Population and which are active (e.g., RS CVn;
lHal]|ll976j) . We suggest that the quiescent DLS counter-
parts are rapidly spinning owing to binarity (driven by
a close-in planet or a brown dwarf companion). Unfor-
tunately, the spectral resolution of our LRIS spectrum
pose no significant constraint on the rotation period: the
full width at half maximum merely constrain the rotation
period to be larger than the breakup period.

The key conclusion of our paper is that it is now clear
that all the fast transients found by the DLS are flares
from Galactic M dwarfs. The fast transient annual rate
in B band (obtained by using the s ky exposure a nd tran-
sient detection efficiency given in IBecker et al.ll2004^) is
Rjr ~ 10 8 yr _1 . The impressive value of -Rjr, relative
to any known category of extragalactic transients (jjjjj,
means that the "foreground fog" of flares will ensure
that the false positives outnumber genuine extragalactic
events by at least two orders of magnitude.

Indeed, current searches even with modest coverage
are already facing a steady drizzle of interlopers. Not-
ing the tendency of astronomers to report all unusual
transients as candidate GRB afterglows we scoured the
GRB literature. A large flare in the localization of the
gam ma-ray burst GB 781000 6 B turns out to be a dMe
star l|Greiner fc Motchl 119951) . iMohan et aD l|2000D re-
ported a source brightening to R — 19.5 mag in the IPN
localization of GRB 001212; the quiescent counterpar t
was later identified with a 2MASS M dwarf ljOizisl2000j) ,
During th e course of the Ca talina survey for near earth
asteroids iChristensenl ()2004fl identified a fast transient
source (fading from R = 15.9 to 19.7 mag in 1 hr) in
Lynx and later spectros copically established to be an M
dwarf (Dioreovski et al. 20 QJ). The fast transient fo und
in the CFHT Legacy Survey ( Malacrino et al. 2005) was
later i dentified wi th a 2MASS counterpart, likely a M
dwarf (|Pricfll2005fl .

Consider a specific example of a survey undertaken by
a telescope with field of view, Aft (in steradians). Let
r be the exposure time per field of view (frame time,
usually between 30 s and 300 s) . The number of flares per
frame is thus N? = Rj^tAQ/(Att) and the number over a
time T is TR^An/(4ir). For Aft = 2.8 x 10~ 3 steradian
(corresponding to 10 square degrees) and T = 0.3 yr (a
year's worth of observing) we find N? ~ 6, 600.

How can this immense fog of M dwarf flares be pen-
etrated? Spectroscopy of the quiescent counterparts
would require significant amount of time on large tele-
scopes and is thus not a feasible approach. A counter-
part which has spatial extent may simply be due to an
asterism or a gaseous nebula formed by repeated flares.
An attractive locale such as a bright galaxy or a cluster
of galaxies (cf. OT 20010326) does not necessarily imply
extragalactic origin. Finally, those transients which lie
in unfavorable regions (close to a diffraction spike, close
to a bright object) would have poorer photometry and
errors may conspire to make the transient appear more

7000 8000

Wavelength (angstroms)

9000

4840 4860 4880
Wavelength (angstroms)

6540 6560 6580
Wavelength (angstroms)

Fig. 2. — LRIS spectrum of the quiescent counterpart of OT 20030305. The right panel shows the heavily binned (factor 25) red continuum
together with the Ha emission line and CaH+TiO absorption bands marked. Close-ups of the regions around H/3 and Ha are shown in the
center and right hand panel, respectively.

interesting. Indeed, it is only transients with such inter-
esting coincidences or properties that will inevitably lure
the hapless observer into a web of incorrect conclusions.

The best line of defense is to take advantage of the
blue (UV) color of flares. On detection of a potential
fast transient follow-up observations in a band (or two)
complementary to the discovery band should be under-
taken and the color(s) thus used to discriminate against
the M dwarfs. The Rayleigh-Jeans tail of a blackbody
has a distinctly different spectrum (/„ oc v 2 ) as opposed
to say a synchrotron radiation (/„ oc v a with a ~ — 1).
Needless to say these observations should be done whilst
the flare is in progress i.e. on timescales of 100 s or so.

We close this paper with two observations. First, Pan-
STARRS and LSST have a multitude of science goals.
Discovering new classes of fast transien t s is on ly one of
them. From the work of iBecker et alJ l|2004j) and the
work presented here it is now abundantly clear that flares
from late type stars constitute a veritable foreground fog.
Surveys have to be tuned to minimize the fog: judicious
choice of directions (high Galactic and ecliptic latitude) ,
careful choice of the primary search (optical) filter, lim-
iting the target fields so that (over time) M dwarfs in
those fields are identified, undertaking a pre-search sur-

vey designed to identify suspect late type dwarfs (e.g.
deep K band observations during bright time) and/or a
well planned rapid followup strategy with complemen-
tary filters may be more successful than multi-purpose
surveys.

Next, we note that the OT yyyymmdd designation is
Y2K compliant and thus represents a slight improve-
ment over the present GRB naming convention. Op-
tical astronomers have the opportunity to continue the
(quixotic and impractical) GRB and SN convention of
appending A, B, C, ..., AA and so on (for successive
events) but, given the measured rate of flares from M
dwarfs of 10 8 yr _1 , face the risk of drowning in a bliz-
zard of alphabets.

We thank A. Becker and P. Boeshaar for construc-
tive feedback and S. Djorgovski for reminding us of the
Lynx transient which then led us to search the GCN
literature for other examples of M dwarf flares. We ac-
knowledge useful discussions with M. della Valle, E. Ofek,
E. S. Phinney, P. Price, E. Priest, M. Salvato & E. Scan-
napieco. This work is supported in part by grants from
NSF and NASA.

REFERENCES

Becker, A. C, et al., 2004, Astrophys. J., 611, 418

Cappellaro, E., Evans, R., & Turatto, M. 1999, A&A, 351, 459

Christensen, E. 2004, GCN Circular 2849

Cole, S., et al., 2001, Mon. Not. R. astr. Soc, 326, 255

della Valle, M., Rosino, L., Bianchini, A., & Livio, M., 1994, A&A,

287, 403
Djorgovski, S. C, Gal- Yam, A., & Price, P. 2004, GCN Circular

2851
Gizis, J. E. 2000, GCN Circular 901

Greiner, J. & Motch, C. 1995, Astr. Astrophys., 294, 177
Groot, P. J., et al., 2003, MNRAS, 339, 427

Guetta, D., Piran, T., & Waxman, E. 2005, Astrophys. J., 619, 412
Hall, D. S. 1976, in ASSL Vol. 60: IAU Colloq. 29: Multiple Periodic

Variable Stars, ed. W. S. Fitch, 287
Kaiser, N., et al., 2002, in Survey and Other Telescope Technologies

and Discoveries. Edited by Tyson, J. Anthony; Wolff, Sidney.

Proceedings of the SPIE, Volume 4836, pp. 154-164 (2002)., ed.

J. A. Tyson & S. Wolff, 154-164

Malacrino, F., Atteia, J., Boer, M., Klotz, A., Veillet, C,
Cuillandre, J., & Kaavelars, J. 2005, GCN Circular 2964

Moffett, T. J. 1974, Astrophys. J. Supp. Series, 29, 1

Mohan, V., Sagar, R., Pandey, S. B., & Castro-Tirado, A. J. 2000,
GCN Circular 900

Nakar, E., Gal- Yam, A., & Fox, D. B. 2005, astro-ph/0511254

Oke, J. B., et al., 1995, PASP, 107, 375

Pickles, A. J. 1998, PASP, 110, 863

Price, P. A. 2005, GCN Circular 2965

Rau, A., Greiner, J., & Schwarz, R. 2006, Astr. Astrophys., 449,
79

Williams, S. J. & Shatter, A. W. 2004, Astrophys. J., 612, 867