Serendipitous Kepler observations of a background dwarf nova of SU UMa type
Serendipitous Kepler observations of a background dwarf nova of SU UMa type
Thomas Barclay 1 2
Martin Still 1 2
Jon M. Jenkins 0
Steve B. Howell 2
Rachael M. Roettenbacher 3
0 SETI Institute/NASA Ames Research Center , Moffett Field, CA 94035 , USA
1 Bay Area Environmental Research Institute, Inc. , 560 Third Street West, Sonoma, CA 95476 , USA
2 NASA Ames Research Center , M/S 244-40, Moffett Field, CA 94035 , USA
3 Deptartment of Astronomy, University of Michigan , 830 Dennison Building, 500 Church Street, Ann Arbor, MI 48109 , USA
A B S T R A C T We have discovered a dwarf nova (DN) of type SU UMa in Kepler data which is 7.0 arcsec from the G-type exoplanet survey target KIC 4378554. The DN appears as a background source in the pixel aperture of the foreground G star. We extracted only the pixels where the DN is present and observed the source to undergo five outbursts - one a superoutburst - over a time span of 22 months. The superoutburst was triggered by a normal outburst, a feature that has been seen in all DNe superoutbursts observed by Kepler. Superhumps during the superoutburst had a period of 1.842 ± 0.004 h, and we see a transition from disc-dominated superhump signal to a mix of disc and accretion stream impact. Predictions of the number of DNe present in Kepler data based on previously published space densities vary from 0.3 to 258. An investigation of the background pixel targets would lead to firmer constraints on the space density of DNe.
accretion; accretion discs - methods; data analysis - stars; dwarf novae - novae; cataclysmic variables
1 I N T R O D U C T I O N
Cataclysmic variables (CVs) are binary star systems consisting of
a white dwarf primary and a less massive main-sequence or
postmain-sequence secondary. The two stars orbit a common centre
of mass with typical periods between 50 min and 5.7 d (Sekiguchi
1992; Downes et al. 2001; Ritter & Kolb 2003; Thorstensen, Le´pine
& Shara 2006). The small physical size of the binary system causes
the secondary to be distorted by the gravitational field of the white
dwarf and leads to the filling of its Roche lobe. Mass is transferred
through the L1 inner Lagrangian point and forms a stream of
material falling into the potential well of the white dwarf. In the case
where the white dwarf is non-magnetic, the infalling material forms
an accretion disc (Shakura & Sunyaev 1973). Viscous shearing
layers amplify a weak magnetic field within the disc, and lead to
turbulent angular momentum transport radially outwards (Balbus
& Hawley 1998). Mass can then move inwards and be accreted on
to the surface of the white dwarf (Hellier 2001). At optical
wavelengths, the disc is usually much more luminous than the white
dwarf, its companion and the impact of the accretion stream on to
The DN subclass of CVs undergo quasi-periodic outbursts (see
Warner 2003, for a review). During an outburst, the optical
brightness of the system can increase by a factor of 10 or more (e.g.
Ramsay et al. 2009), which interrupts the faint quiescent state in
which the DNe typically spend most of their time.
If the mass transfer rate from the secondary to the accretion disc
is greater than the rate of disc accretion on to the white dwarf,
M˙ (2) > M˙ (d), then the disc radius will increase over time. This
can only happen if the dimensionless disc viscosity parameter α
is low (where low is usually taken in models to be 0.03; e.g.
Smak 1984; Cannizzo 1993). The disc instability model predicts
that eventually the surface density in the disc will exceed a critical
value max, which initiates a phase transition in the disc material
(Ho¯shi 1979; Lasota 2001). At this point, the α parameter increases
(by a factor of ∼10; Smak 1984; Cannizzo 1993) and M˙ (2) < M˙ (d)
for a time until the surface density at the outer edge of the disc
decreases below some critical surface density min, and α returns
to the original lower value. This model is known as the limit cycle
accretion disc instability (Lasota 2001).
A subset of DNe known as SU UMa stars occasionally undergo
a superoutburst (SO) in addition to undergoing normal outbursts
(NO). SOs are typically a factor of 2 brighter and 5–10 times longer
than a NO (Warner 2003). Superhumps, photometric variability on
a period a few per cent longer than the orbital period of the system,
are observed during the SO. During a NO, the disc expands in
radius (Osaki 1989). After a number of these outbursts, the disc can
become large enough to grow beyond the inner Lindblad resonance,
close to the 3:1 corotation resonance with the binary orbital period
(Whitehurst 1988). Once there is enough mass at this resonance
radius, the superhump oscillation can be driven by tides.
We identified a background source in Kepler spacecraft archival
data whose light curve resembled that of a DN. In Section 2 we
describe the observations made by the Kepler spacecraft of this
source, our re-extraction of the light curve from pixel level data
using a more appropriate pixel aperture and our removal of
systematic photometric effects from the flux time series. In Section 3
we discuss observed outbursts, and in Section 4 we discuss the
implications of these observations on the nature of outbursts from
2 O B S E RVAT I O N S
Kepler has been almost continuously observing over 1.6 × 105 stars
at a cadence of 30 min (Borucki et al. 2010; Koch et al. 2010) at
millimagnitude to micromagnitude precision (Gilliland et al. 2011)
since the spacecraft’s launch on 2009 March 06. While the primary
aim of the Kepler mission is the discovery of exoplanets, the long
baseline of continual, high-precision observations has also provided
an unrivalled data set to the stellar astrophysics community.
The available bandwidth does not allow all starlight falling on
to the Kepler detectors to be downloaded at a cadence of 30 min.
Instead, the pixels to be collected must be preselected and account
for only 5.75 per cent of the total detector area available (Koch et al.
The continuity of observations are broken every month by the
satellite pointing towards Earth for data to be downloaded. Once
every three months, the satellite rolls 90◦ to keep the sunshade
pointed in the appropriate direction but has the effect of changing
the CCD that a star falls upon every quarter. Because of the change
in physical position of the instrument and because a new data set is
downloaded, the data from Kepler are divided into quarters starting
with Quarter 0 (Q0), containing commissioning observations, and
full science operations beginning in Q1.
Each observed star is extracted by selecting a pixel mask which
can change in size and shape every quarter because the position of
the star in pixel space is different. For this reason, the contamination
from background stars is different every quarter – this contamination
can be significant because part of the Kepler field covers the Galactic
plane (where the stellar density is high) and the pixels are relatively
large (3.98 × 3.98 arcsec2).
The observations described in this paper are of one particular
target aperture – that around KIC 4378554. This star was chosen to
be a target in the Kepler exoplanetary programme. Kepler observed
KIC 4378554 during Q1–Q8 excluding Q6 (see Table 1 for a log of
the observations). No planet has been found transiting this system
as yet. The Kepler Input Catalogue lists KIC 4378554 as a Kp =
14.9 star1 with colour (g − r ) = 0.57, an effective temperature of
5617 K and a surface gravity of log g = 4.615 in cgs units; these
values are consistent with the source being a main-sequence G-type
star (Brown et al. 2011). The star is only visible during three of the
four seasons of observations because it lies on Module 3 for one
quarter. Module 3 failed during Q4.
2.1 Optimal aperture observations
Two light-curve products are provided by the Kepler project: light
curves extracted using Simple Aperture Photometry (SAP) and light
curves which have undergone the additional step of Pre-Search
Data Conditioning (PDC; Twicken et al. 2010). The latter of these
is produced for use in exoplanet searches but is known to cause
problems if used for stellar astrophysics work. These problems
include the attenuation of astrophysical signal and the injection of
spurious signals; because of this, we use the SAP data. A light
curve showing Q5 data can be seen in Fig. 1. These data have not
had systematic artefacts removed, and there are several noticeable
features due to systematics such as the increase in flux level over the
final 50 d of the quarter and the thermal settling after Earth-points
at Barycentric Julian Date (BJD) 245 5307 and 245 5337. However,
there is one obvious feature that appears to be astrophysical in
nature – a 2 per cent increase in brightness with a peak at BJD
245 5342.8 lasting 2.5 d. In addition to the brightening shown here,
we identified four similar events – two in Q3, one in Q4 and one
in Q7. Events like these are not typically seen in G-type stars and
therefore warranted further investigation.
2.2 Custom light-curve extraction from target pixel files
In order to determine whether the brightness increases are due to
KIC 4378554 or a background star in the optimal aperture of KIC
1 Kp refers to approximate Kepler instrumental magnitudes. For more details
on this passband, see Brown et al. (2011).
4378554, we examined the archived target pixel files.2 These files
contain a series of images – one for each time stamp – showing
the individual pixels downloaded from the Kepler spacecraft. These
images were calibrated by the Kepler pipeline (as described by
Quintana et al. 2010). Shown in Fig. 2 is the Q5 time series for each
calibrated pixel in the aperture of KIC 4378554. The pixels shown
in grey were used to create the light curve in Fig. 1. From Fig. 2 we
can see that the brightness increase at BJD 245 5342.8 is not caused
by KIC 4378554 but by a source which appears to have its centre
on or close to pixel (1043 857) approximately 1.5 pixels south-west
of the target star. For convenience, we will refer to this source as
NIK 1 (standing for not in KIC).
We extracted a new light curve by summing only those pixels
which NIK 1 contributes flux using the KEPMASK and KEPEXTRACT
tools from the PYKE Kepler community software.3 In quarters where
we see a sudden brightening event, this is relatively straightforward.
This is the case in Q3, Q4, Q5 and Q7. For Q1 we use the Q5
aperture, and for Q8 we used the Q4 aperture because NIK 1 will
be at the same position on the CCD four quarters apart. For Q2 we
estimate the position of NIK 1 from the full-frame image taken at
the end of Q3, when a brightening event occurred and NIK 1 is
visible.4 We overlaid the full-frame image containing NIK 1 on the
pixel level data and selected the appropriate pixels. The number of
pixels we use for all quarters is listed in Table 2.
2.3 Removing systematic trends using cotrending basis vectors
The light curves we extracted from target pixel files still contain
systematic features. This is because the extracted aperture does not
2 Kepler target pixel and light curve files are distributed at http://
3 PYKE is available from http://keplergo.arc.nasa.gov/ContributedSoftware.
4 Kepler full-frame images are collected once a month and are available
contain all the flux and also contains contaminating flux from the
target star, KIC 4378554. The amount of flux in the aperture, both
from NIK 1 and from other contaminating sources, changes
continually over time due to differential velocity aberration and suddenly
due to changes in the focus after Earth-points (Christiansen et al.
The instrumental systematics which affect a single light curve are
unique. However, we see strong correlations between stars which
are on the same CCD channel. We can take those stars which are
most highly correlated, which we call the reference ensemble, and
identify the correlated features. These correlations can be
represented as linear combinations of orthogonal vectors, which can be
fitted and subtracted from each individual light curve to remove
instrumental features. The Kepler team have created a product
containing these vectors, known as cotrending basis vectors (CBVs),
and have made them available to the public5 [a description of how
they were created is given in Christiansen et al. (2011b) and Barclay
et al. (2011)].
The available files consist of the 16 vectors, of length the same
as the light curve, which provide the greatest contribution to the
variance of the reference ensemble. There is one set of 16 vectors
per quarter for every CCD readout. It is crucial that an appropriate
number of vectors are chosen to fit to the light-curve data; using too
few will leave instrumental signal in the light curve, while using too
many will cause overfitting and remove astrophysical signal.
We used the KEPCOTREND tool from PYKE to perform a linear
leastsquares fit of the CBVs to the light curve we extracted using
KEPEXTRACT and masked the regions where the brightening events occur
out of the fit because these regions are dominated by astrophysical
signal, not systematics. We selected the number of CBVs to fit by
experimenting with the fit: we assumed that there was no intrinsic
long-period variability outside of the brightening and attempted to
remove any long-term trends. Starting with two CBVs, we
iteratively increased the number of CBVs until the regions outside of
5 CBVs are available from http://archive.stsci.edu/kepler/cbv.html.
Quarter Pixels in optimal ap. Pixels in custom ap. Time range masked in CBV fit
(BJD −245 4900)
# of CBVs in fit
272.714 65–282.501 47
285.359 80–293.986 26
441.891 37–448.345 35
621.906 76–629.788 12
the brightening events was approximately constant. The number of
CBVs used and the regions masked from the fit are listed in Table 2.
The CBVs only correct for the amplitude of systematic effects
and not for constant offsets caused by differences between the
contamination from nearby sources and changes in the fraction of flux
of NIK 1 in the pixel aperture in different quarters. After
removing the instrumental systematics, there are still differences between
the median flux levels between quarters. To mitigate the effects
of this, we normalize the light curve. A representation of the true
flux level was obtained by first normalizing independently for each
Fi − med (F )
MAD (F )
where Fnorm is the vector of normalized fluxes, F is the vector of
non-normalized fluxes, Fi is the ith element of vector F, med(F)
is the median of F and MAD(F) is the median absolute deviation
from the median (MAD) of F. MAD is defined as
MAD (F ) = med (|Fi − med (F ) |) .
We show the full custom extracted and cotrended time series
containing seven quarters of data in Fig. 3. The light curve is that
of a SU UMa type DN [cf. the adapted light curve of VW Hyi
from Bateson (1977) shown in Warner (2003)]. We observe five
outbursts (labelled OB1–5 in Fig. 3): four NOs and one SO. One of
the NOs (OB1) immediately precedes the SO (OB2), while another
NO (OB3) is of much lower intensity than the other NOs and occurs
a mere 4 d after the end of OB2. Our method of normalization is
relatively crude; if, for example, there are significantly different
levels of contamination or the noise level was dramatically different
between the quarters, we would be incorrectly estimating the relative
outburst amplitudes. We gain confidence that the normalization is
appropriate for the following reasons: the outbursts in Q5 and Q7 are
of comparable intensities, and there is a relatively smooth transition
from the tail of the SO seen in Q3 and into Q4. It is possible that
we are severely underestimating the amplitude in Q4 and OB3 is
actually of comparable intensity to OB1, OB4 and OB5, but this
would require a sharp break in the decline of the SO in the data gap
between Q3 and Q4.
2.4 Archival observations
We were unable to locate NIK 1 in quiescence in Kepler
fullframe images which have a limiting magnitude of Kp ∼ 20, and
so we looked to archival data to find NIK 1 in a quiescent state.
Palomar Sky Survey I and II (POSS-I and II) observations appear
in the USNO-B1.0 catalogue (Monet et al. 2003). USNO-B1.0 lists
a source offset (in the direction of KIC 4378554) from NIK 1 ×
2.7 arcsec. The POSS-I blue magnitude of this source is given as
19.15 and the source is catalogued as USNO-B1 1294-0347652.
They again find this source in the POSS-II blue image and derive a
blue magnitude of 15.13.
POSS-I and II images are available for this region as part of the
Digitized Sky Survey (DSS). We examined the POSS-II images of
this region and found a source we estimate to be at B 17.3 based
on the brightness of nearby stars in the blue image, much brighter
than the detection limit of the full-frame images where we do not
detect NIK 1. We deduce that this source is NIK 1 in a state of
outburst. Fig. 4 shows a comparison between the POSS-II red and
blue images which were taken at different epochs. No source is
detected at the position of NIK 1 in the red image.
Despite the USNO-B1.0 listing a source at the same position in
POSS-I blue data, we were unable to locate any source. However, the
fact that the source is listed with two quite different brightness
measurements leaves us in little doubt that USNO-B1 1294-0347652
and NIK 1 are one and the same.
3 O U T B U R S T S
The Kepler data contain observations of five outbursts: four NOs
and one SO. The dates the outbursts reach the maximum and their
rise and fall durations are given in Table 3.
Quarter Type of outburst 1 2
BJD of Fmax
aThe fall time is truncated by the onset of the SO.
bThe rise time is truncated by the tail of the first NO.
3.1 Normal outbursts
The four NOs are shown in detail in Figs 5 and 6. OB4 and OB5 are
very similar looking with comparable brightnesses. However, we
caution that the uncertainty in measuring the outburst peak flux is
high because of the method used to normalize these data. The shapes
of the NOs are typical of those seen in other DNe (e.g. V344 Lyr;
Cannizzo et al. 2010) consisting of a fast rise to peak brightness of
between 0.81 and 1.25 d with a somewhat slower fall to quiescence
taking 1.86–2.57 d.
3.2 The superoutburst
The SO we detected, OB2, is shown in Fig. 5. It reached peak flux
on BJD 245 5176.86 and returned to quiescence 10.42 d later. We
note that the SO was immediately preceded by a NO, OB1, which
did not return to quiescent brightness before the onset of the SO.
OB3 then occurred only 4 d after the SO returned to the quiescent
Superhumps always occur during SOs (e.g. Vogt 1980; Warner
2003; Kato et al. 2011). In NIK 1, the superhumps start during the
decline of the NO, OB1, at the stage when the SO is likely beginning
and continue until the OB3 begins.
We performed a Fourier transform of the SO searching for periods
ranging from the Nyquist period (58.84 min) to 12 h. The strongest
frequency (FA) found is at 13.03 ± 0.03 cycles d−1 (1.842 ±
0.004 h) where the uncertainty was calculated using a Monte Carlo
technique (Lenz & Breger 2005). In addition, we detect a signal at a
frequency of 22.85 ± 0.08 cycles d−1 (= 1.05 h, FB). The first
overtone of the strongest frequency should occur at 26.06 cycles d−1
(F2). However, because the Nyquist frequency (FN) of the data
is 24.47 cycles d−1, we cannot uniquely detect F2. The value of
F2 − FN is 1.59 ± 0.06 and FN − FB is 1.62 ± 0.08; therefore, we
concluded that the frequency at 22.85 cycles d−1 is very likely to be
a reflection of the first overtone about the Nyquist frequency.
In order to better characterize the superhumps, we propagated
the time series through a boxcar high-pass filter with a cut-off
frequency of 18.0 d−1. This is shown in the top panel of Fig. 7.
We calculated a discreet short time Fourier transform (Harris 1978)
over the duration of the SO. We used the filtered data and calculated
a Fourier transform every 0.1 d with a siding window size of 2.0 d.
The resulting power spectrum is shown in the lower panel of Fig. 7.
Both FA and FB are visible for the entire duration of the SO and
persist through the interval between OB2 and OB3. The superhump
falls below a 3σ detection level at BJD 245 5190, close to the time
OB3 begins. We do not detect the two signals over the rest of the
time series with any significance.
In the upper two panels of Fig. 8 we zoom in around the two
strongest frequencies seen in Fig. 7. The peak frequency of both the
fundamental and reflection of the first overtone frequency vary over
time. The crosses show the position of the maximum power in the
sliding window power spectrum. In Q3, the period of the main peak
and the reflection of the frequency of the first overtone appear to
move in opposite directions. This is expected if the higher frequency
peak is indeed a reflection about the Nyquist frequency. In the lower
panel of Fig. 8 we show how the amplitude of both the fundamental
(red crosses) and the reflection of the first overtone (blue crosses)
change over time. The second harmonic frequency decays slower
than the fundamental. The fundamental frequency falls below the
3σ detection level 0.6 d before the first overtone frequency.
We see changes in the superhump waveform when comparing
the modulation at mid-outburst (top panel in Fig. 9), at the tail
end of the SO (middle panel), and when the system is in a fainter
state before the rise of OB3 (lower panel). The third time series is
double peaked, which explains the reflection of the second harmonic
decaying slower than the fundamental frequency in Fig. 8 – the
waveform now has twice the frequency of the initial waveform.
4 D I S C U S S I O N S
4.1 The nature of the outbursts
NIK 1 is now the fourth CV which has been seen in outburst by
Kepler (Still et al. 2010; Garnavich, Still & Barclay 2011; Kato
et al. 2011). All four have undergone a SO. In two of the three
other cases the SO has been immediately preceded by a NO (Still
et al. 2010; Kato et al. 2011). In the third case, the rise of the SO is
missed during a telescope safe mode (Garnavich et al. 2011). While
statistics are still low this raises a question: are all SOs triggered by
Bateson (1977) suggested that in the DN VW Hyi, some SOs are
triggered by a NO, and Marino & Walker (1979) posited that this
is the case for all SOs of VW Hyi. In a few other ground-based
observations of DN, the SO has been observed to be preceded by
a NO – e.g. T Leo (Kato 1997), TV Crv (Uemura et al. 2005)
and GO Com (Imada et al. 2004). Cannizzo et al. (2010) postulate
that a NO occurs due to thermal instability in the disc, and, when
there is enough mass built up from previous outbursts at large radial
distances, a SO can be triggered. However, it takes time to propagate
the heating front to large radii where this overdensity of material
in the outer disc exists. We first see the NO and then when outer
material transitions to the hot state, the SO occurs.
The limit cycle disc instability model predicts that the disc is
much smaller directly after a SO and easier to excite into an
outburst state. The intensity of outbursts should be lower and outbursts
should occur more frequently for a smaller disc (Lasota 2001).
Our observations are consistent with this pattern. We observe no
outbursts before OB1 – the system was quiescent for 208 d – and
in the remaining observations the longest gap between outbursts is
181 d, the gap between OB4 and OB5, the final two outbursts. The
NOs are more frequent after the SO and the spacing increases over
time. We also see a much smaller outburst directly after the SO,
which is consistent with theory.
We observed superhumps lasting for ∼17 d. They begin during
the fading of the NO which preceded the SO and last until the
beginning of OB3. We see superhumps for 3–4 d after the SO has
faded to quiescence. There have only been four other reports of
superhump persisting after the end of a SO: V1159 Ori (Patterson
et al. 1995), ER UMa (Gao et al. 1999), EG Cnc (Patterson et al.
1998) and V344 Lyr (Still et al. 2010). It seems likely material will
persist at large radii well after the majority of material has been
accreted by the white dwarf, and the reason superhumps have been
seen persisting so few times is because of a lack of sensitivity in
The disc instability model (Osaki 1989; Lasota 2001) predicts
two main sources driving the superhump modulation: viscous
dissipation within a flexing disc driven by the compression of the disc
opposite the secondary star, and the hotspot where the accretion stream
impacts the disc sweeping around the rim of a non-axisymmetric
disc (Wood et al. 2011). During the early superhumps, the hotspot
signal is much weaker than the signal from the disc and the phase
curve is single peaked. As the disc fades, the hotspot from the
impact of the accretion stream on to the disc will become brighter
relative to the disc because the disc has been drained of material
and so is much less luminous.
We see the sumperhump waveform change over the course of the
SO. In the folded time series shown in the upper panel of Fig. 9, the
waveform is asymmetric, and the pulse rises around twice as fast
as it falls. The shape is very similar to that seen in V344 Lyr (Still
et al. 2010) at a comparable time in the SO evolution and in the
simulations shown in fig. 1 of Wood et al. (2011) when modelling
early superhumps. In the middle panel, we see some evolution of
the waveform and we no longer see evidence for the asymmetry; the
light curve is similar to that seen in fig. 3 of Wood et al. (2011) where
they have accretion stream impact in the model. Our assumption is
that in the middle panel we are seeing no variable component from
the accretion stream.
The waveform changes in the lower panel of Fig. 9. Similarly to
the other panels there is a peak around phase 0.9 – any difference
can be attributed to the error on the period on which the data were
folded. However, there is also a secondary peak which is offset in
phase by ∼180◦ and of lower intensity than the main peak. This is
wholly consistent with fig. 2 of Wood et al. (2011) where both the
disc and the hotspot contribute to the emission for the system.
Given that orbital periods are typically a few per cent lower than
the superhump period (Warner 2003), we can place an upper limit
on the binary orbital period of 1.842 ± 0.004 h; this places NIK
1 below the period gap. This meets expectations as almost all SU
UMa stars are below the period gap (Kato et al. 2011).
4.2 DNe in the Kepler field of view
What is the possibility of finding more DNe as background sources
in the Kepler data? Using the empirical relation of Patterson (2011),
the lowest peak absolute magnitude of a DN outburst in a system
with a period longer than 90 min is MV = 5.3. This implies Kepler
can observe DNe in outburst out to a distance of 8.7 kpc if we
assume we can observe all sources with a peak brightness Kp <
20, which is approximately the source confusion limit of Kepler.
However, we cannot assume an isotropic distribution out to this
distance. Instead, we restrict our search to only sources less than
300 pc above the Galactic plane, which Patterson (2011) estimates
to be the scale height for DNe.
The centre of the Kepler field of view is 13◦.5 above the Galactic
plane and the field of view is 116 deg2 (Koch et al. 2010). Therefore,
the greatest distance an observable DN can be from Earth and still
be within our scale height constraint is 1285 pc, and our search
volume is 2.5 × 107 pc3 (this includes the entire field of view,
References. (1) Patterson (2011); (2) Ak et al. (2008); (3) Pretorius et al.
(2007); (4) Hertz et al. (1990); (5) Shara et al. (1993); (6) Kolb (1993).
not just the pixels downloaded). The period luminosity relation of
Patterson (2011) predicts all DNe outbursts in our search volume
will be brighter than V = 15.8 and easily be detected by Kepler,
even if they have high reddening.
Patterson (2011) gives a lower limits to the space density at
DNe of 1.8 × 10−6 pc−3 , but this value is calculated using only
42 systems with known distance and most likely underpredicts the
true value significantly. Other estimates from observations include
7.9 × 10−6 pc−3 (Ak et al. 2008), 1.1 × 10−5 pc−3 (Pretorius et al.
2007), 2.1 × 10−5 pc−3 (Hertz et al. 1990) and 10−4 pc−3 (Shara
et al. 1993). Theoretically, the number is thought to be around
1.8 × 10−4 pc−3 (Kolb 1993; Pretorius et al. 2007). For each of
these estimates we have calculated the inferred number of DNe
potentially visible to Kepler. These are given in Table 4 and assume
an isotropic distribution within the search range of one Galactic disc
Ak et al. (2008) measure the scale height of the DN population to
be only 150 ± 18 pc. If we use this value as the ceiling of our search,
the numbers predicted decrease by a factor of 8. These values are
also shown in Table 4.
The predicted values range from 6 to 4485, which leads us to
believe that the true space density of DNe is wholly unconstrained.
If we are considering systems that may be found as background
sources in apertures of target stars, the numbers decrease by a factor
of ∼17 because Kepler only downloads the pixels from 5.75 per
cent of the focal plane. With this restriction, the predicted number
in Kepler ranges from 0.3 to 258. With only nine known DNe in
the data, we are unable to rule out any of the models at the 3σ
level. The predictions are likely to be an underestimate of the true
number because this estimate would not include the source in this
work as it would be further than 300 pc from the Galactic equator.
A systematic search of the target pixel files may lead to a bounty of
new DNe. Such work will help constrain the true value of the space
density of DNe.
5 S U M M A R Y
We have serendipitously observed outbursts of a DN which was
captured within the aperture of a G-type star. We first extracted
the flux of this source from the target pixel files and then removed
systematic trends from the data using CBVs. We observed four NO
and one SO over 22 months. The SO was likely triggered by a NO.
We observed superhumps during the SO and saw the waveform
change from being single peaked to double peaked over the course
of the outburst, which we attribute to the appearance of the accretion
stream impact on to the disc.
Given that this DN was observed as a background source, we used
various predictions of DNe space density to predict the number of
DNe in the Kepler field of view and in data already collected. The
models vary wildly from predicting 4485 and 257.9 in the field of
view and in collected data, respectively, to 2.6 and 0.32. A detailed
study of the background pixels collected by Kepler could be used
to constrain the space density of DNe.
Kepler was selected as the 10th Discovery mission. Funding for
Kepler is provided by NASA’s Science Mission Directorate. This
paper made extensive use of the PYKE tools provided by the
Kepler Guest Observer Office for the use of the community. All of
the data presented in this paper were obtained from the
Multimission Archive at the Space Telescope Science Institute (MAST).
STScI is operated by the Association of Universities for Research
in Astronomy, Inc., under NASA contract NAS5-26555. Support
for MAST for non-HST data is provided by the NASA Office of
Space Science via grant NNX09AF08G and by other grants and
contracts. We used images from The Digitized Sky Surveys which
were produced at the Space Telescope Science Institute under US
Government grant NAG W-2166. The images of these surveys are
based on photographic data obtained using the Oschin Schmidt
Telescope on Palomar Mountain and the UK Schmidt Telescope.
The plates were processed into the present compressed digital form
with the permission of these institutions.
This paper has been typeset from a TEX/LATEX file prepared by the author.
Ak T. , Bilir S. , Ak S. , Eker Z. , 2008 , New Astron., 13 , 133
Balbus S. A. , Hawley J. F. , 1998 , Rev. Modern Phys. , 70 , 1
Barclay T . et al., 2011 , Kepler Data Release 12 Notes, KSCI-19052-001
Bateson F. M. , 1977 , New Zealand J. Sci., 20 , 73
Borucki W. J. et al., 2010 , Sci, 327 , 977
Brown T. M. , Latham D. W. , Everett M. E. , Esquerdo G. A. , 2011 , AJ, 142 , 112
Cannizzo J. K. , 1993 , ApJ, 419 , 318
Cannizzo J. K. , Still M. D. , Howell S. B. , Wood M. A. , Smale A. P. , 2010 , ApJ, 725 , 1393
Christiansen J. L. et al., 2011a , Kepler Data Characteristics Handbook, KSCI-19040-002
Christiansen J. L. et al., 2011b , Kepler Data Release 11 Notes, KSCI-19051- 001
Downes R . A., Webbink R. F. , Shara M. M. , Ritter H. , Kolb U. , Duerbeck H. W. , 2001 , PASP, 113 , 764
Fraquelli D. , Thompson S. E. , 2011 , Kepler Archive Manual , KDMC-10008- 003
Gao W. , Li Z. , Wu X. , Zhang Z. , Li Y. , 1999 , ApJ, 527 , L55
Garnavich P. , Still M. , Barclay T. , 2011 , Astron. Telegram, 3507 , 1
Gilliland R. L. et al., 2011 , ApJS, 197 , 6
Harris F. , 1978 , Proc. IEEE , 66 , 51
Hellier C. , 2001 , in Hellier C., ed., Cataclysmic Variable Stars. SpringerPraxis, Chichester
Hertz P. , Bailyn C. D. , Grindlay J. E. , Garcia M. R. , Cohn H. , Lugger P. M. , 1990 , ApJ, 364 , 251
Ho¯shi R., 1979 , Progress Theor . Phys., 61 , 1307
Imada A. et al., 2005 , PASJ, 57 , 193
Kato T. , 1997 , PASJ, 49 , 583
Kato T . et al., 2011 , preprint (arXiv:1108.5252)
Koch D. G. et al., 2010 , ApJ, 713 , L79
Kolb U. , 1993 , A&A, 271 , 149
Lasota J.-P. , 2001 , New Astron. Rev., 45 , 449
Lenz P. , Breger M. , 2005 , Commun. Asteroseismol., 146 , 53
Marino B. F. , Walker W. S. G. , 1979 , in Bateson F. M., Smak J. , Urch I . H., eds, IAU Colloquial 46 , Changing Trends in Variable Star Research. Univ. Waikato, Hamilton, p. 29
Monet D. G. et al., 2003 , AJ, 125 , 984
Osaki Y. , 1989 , PASJ, 41 , 1005
Patterson J. , 2011 , MNRAS, 411 , 2695
Patterson J. , Jablonski F. , Koen C. , O'Donoghue D. , Skillman D. R. , 1995 , PASP, 107 , 1183
Patterson J. et al., 1998 , PASP, 110 , 1290
Pretorius M. L. , Knigge C. , O'Donoghue D. , Henry J. P. , Gioia I. M. , Mullis C. R. , 2007 , MNRAS, 382 , 1279
Quintana E. V. et al., 2010 , Proc. SPIE , 7740 , 77401X
Ramsay G. et al., 2009 , MNRAS, 398 , 1333
Ritter H. , Kolb U. , 2003 , A&A, 404 , 301
Sekiguchi K. , 1992 , Nat, 358 , 563
Shakura N. I. , Sunyaev R. A. , 1973 , A&A, 24 , 337
Shara M. , Moffat A. , Potter M. , Bode M. , Stephenson F. R. , 1993 , in Regev O., Shaviv G ., eds, Ann. Israel Phys. Soc. , 10 , 84
Smak J. , 1984 , Acta Astron., 34 , 161
Still M. , Howell S. B. , Wood M. A. , Cannizzo J. K. , Smale A. P. , 2010 , ApJ, 717 , L113
Thorstensen J. R. , Le´pine S., Shara M. , 2006 , PASP, 118 , 1238
Twicken J. D. , Chandrasekaran H. , Jenkins J. M. , Gunter J. P. , Girouard F. , Klaus T. C. , 2010 , Proc. SPIE , 7740 , 77401U
Uemura M. et al., 2005 , A&A, 432 , 261
Vogt N. , 1980 , A&A, 88 , 66
Warner B. , 2003 , Warner B ., ed., Cataclysmic Variable Stars . Cambridge Univ. Press, Cambridge
Whitehurst R. , 1988 , MNRAS, 232 , 35
Wood M. A. , Still M. D. , Howell S. B. , Cannizzo J. K. , Smale A. P. , 2011 , ApJ, 741 , 105