Äîêóìåíò âçÿò èç êýøà ïîèñêîâîé ìàøèíû. Àäðåñ îðèãèíàëüíîãî äîêóìåíòà : http://star.arm.ac.uk/preprints/2010/560.pdf Äàòà èçìåíåíèÿ: Fri May 21 19:11:57 2010 Äàòà èíäåêñèðîâàíèÿ: Tue Oct 2 02:54:55 2012 Êîäèðîâêà:

Mon. Not. R. Astron. Soc. 000, 000­000 (0000)

Printed 13 May 2010

A (MN L TEX style file v2.2)

Multi-wavelength observations of the helium dwarf nova KL Dra through its outburst cycle
Gavin Ramsay1, Iwona Kotko2, Thomas Barclay1,3, C. M. Copperwheat5, Simon Rosen4, C. Simon Jeffery1 , T. R. Marsh5, Danny Steeghs5 , Peter J. Wheatley 1
2 3 4 5

5

Armagh Observatory, Col lege Hil l, Armagh, BT61 9DG Astronomical Observatory, Jagiel lonian University, Cracow, Poland Mul lard Space Science Laboratory, University Col lege London, Holmbury St. Mary, Dorking, Surrey, RH5 6NT Department of Physics and Astronomy, University of Leicester, University Road, Leicester LE1 7RH Department of Physics, University of Warwick, Coventry, CV4 7AL

13 May 2010

ABSTRACT

We present multi-wavelength observations of the helium-dominated accreting binary KL Dra which has an orbital period of 25 mins. Our ground-based optical monitoring programme using the Liverpool Telescope has revealed KL Dra to show frequent outbursts. Although our coverage is not uniform, our observations are consistent with the outbursts recurring on a timescale of 60 days. Observations made using Swift show that the outbursts occur with a similar amplitude at both UV and optical energies and a duration of 2 weeks. Although KL Dra is a weak X-ray source we find no significant evidence that the X-ray flux varies over the course of an outburst cycle. We can reproduce the main features of the 60 day outburst cycle using the Disc Instability Model and a helium-dominated accretion flow. Although the outbursts of KL Dra are very similar to those of the hydrogen accreting dwarf novae, we cannot exclude that they are the AM CVn equivalent of WZ Sge type outbursts. With outbursts occurring every 2 months, KL Dra is an excellent target to study helium-dominated accretion flows in general. Key words: Physical data and processes: accretion discs; Stars: binary - close; novae - cataclysmic variables; individual: - KL Dra; X-rays: binaries; ultraviolet: stars

1

INTRODUCTION

AM CVn systems are accreting binaries consisting of a white dwarf primary and a degenerate or semi-degenerate secondary star. Their binary orbital p eriods are extremely short (< 70 mins) and are almost entirely hydrogen-deficient (see Nelemans 2005 for a recent review). As such they are the b est sources in which to understand hydrogen-deficient accretion flows. There are currently 26 known (or candidate) AM CVn systems (see Rau et al. 2010 for the most recent discoveries) of which around half a dozen have shown optical outbursts with amplitudes of 3­4 mag. These outbursts are assumed to b e similar to those observed in the hydrogendominated accreting dwarf novae. Although the physics of helium-dominated accretion is p oorly understood, outbursts are thought to b e due to instabilities in the accretion disc (eg Smak 1983, Tsugawa & Osaki 1997 and Kotko, Lasota & Dubus 2010).
c 0000 RAS

Since many of the more recently discovered AM CVn systems are relatively faint (V 19­20) the outburst characteristics of AM CVn systems as a whole are not well understood. To b etter characterise these prop erties we have started a monitoring programme using the Liverp ool Telescop e to obtain images of the northern AM CVn systems once a week over the course of at least one year. The results of our monitoring programme will b e presented in a future pap er (Barclay et al. in prep). Here, we concentrate on one individual system, KL Dra, which was initially thought to b e a sup ernova (Jha et al 1998). Subsequent observations showed it was an AM CVn system with an orbital p eriod close to 25 mins (Wood et al. 2002). Our initial observations made using the Liverp ool Telescop e showed two outbursts from KL Dra within 2 months (Barclay, Ramsay & Steeghs 2009). This makes KL Dra an excellent system with which to gain a b etter understanding of outbursts in AM CVn binaries.


2

Figure 1. The optical observations of KL Dra made using the Liverpool Telescope (LT), Isaac Newton Telescope (INT) and Nordic Optical Telescope (NOT). The first observation was taken on 2009 July 28th. The dot-dashed line helps guide the eye in the two best sampled outbursts.

In this pap er we present results from our recent observing campaign of KL Dra using the Swift satellite, the Liverp ool Telescop e, the Isaac Newton Telescop e, the Nordic Optical Telescop e, the William Herschel Telescop e and the Gemini North Telescop e.

2

OPTICAL PHOTOMETRIC OBSERVATIONS

The strategy for our observations made using the 2.0m Liverp ool Telescop e (LT) was to obtain a single 180 sec image of KL Dra using the RATCAM imager (Steele et al. 2004) in the g band filter approximately once every week. During outbursts we increased the sampling rate to once every few days. Images which had b een bias subtracted and flat-fielded using the LT automatic pip eline were typically downloaded the afternoon after the observation had b een made. We also obtained supplementary images using the Isaac Newton Telescop e (INT) and the Nordic Optical Telescop e (NOT). Since KL Dra is 5.7 from the nucleus of an anonymous galaxy (Jha et al. 1998 and Fig 2 of Wood et al. 2002), differential imaging would have b een an option for obtaining the photometric light curve of KL Dra. However, since the images were taken using a numb er of telescop es (giving different image scales etc), we decided to use ap erture photometry. Care was taken to exclude as much of the galaxy as p ossible and to ensure a star/galaxy free background. The difference in magnitude b etween two comparison stars was constant to within a few 0.01 mag. To place our photometry onto the standard system, we obtained an image of the field and several standard stars in Oct 2009 using the INT. This allowed us to determine the g band magnitude of several local comparison stars.

We show in Figure 1 our optical light curve of KL Dra. Our first observation made on 2009 July 28 showed KL Dra in a bright optical state (g 16.2). One week later the system had faded by around 0.5 mag, after which no observations were obtained for 5 weeks. However, 63 days after our first observation, KL Dra was again observed in a bright optical state (g 16.0) with an outburst amplitude of 3 mag. The rise to p eak brightness (maximum) was short (<2 days) with a short duration drop in brightness 5 days after maximum. Two weeks after maximum there was another sharp drop in brightness giving an overall duration for the outburst of 15 days.

A third burst was detected from KL Dra 61.5±2.0 days after the preceding one. Since KL Dra shortly came too close to the Sun to b e observable, we obtained no ground-based optical observations for nearly 9 weeks. However, an outburst was again seen at a time which is consistent with KL Dra showing outbursts on a rep eating interval of 60 days. Very recently we detected the fifth outburst of KL Dra (2010 Apr 1st) which took place at least 54 days after the start of the previous outburst. Although we find no evidence for other p eriods in our optical data a more comprehensive dataset is needed to test this more thoroughly.

During its low optical variable which is probably riod (Wood et al. 2002). It short duration brightenings occasions (ie MJD55092).

state, KL Dra app ears rather associated with its orbital p ealso shows what app ears to b e (up to 1.5 mags) on several

c 0000 RAS, MNRAS 000, 000­000


The outburst cycle of KL Dra

3

Figure 2. Top panel: the mean optical spectrum of KL Dra obtained using the WHT and the ISIS blue arm when it was in outburst (2009 Oct 10, MJD=55114). Lower Panel: the mean optical spectrum of KL Dra obtained using the Gemini North Telescope when it was in a low state (2007 July 17).

3 3.1

OPTICAL SPECTRA High state spectra

We were fortunate to b e observing on the 4.2m William Herschel Telescop e on La Palma when KL Dra was in outburst. We obtained a series of 100 sec exp osures using ISIS on the night of 2009 Oct 10 which lasted 67 mins. We used the R300B (giving wavelength coverage b etween 3500­5300°) A and R158R (5200­10000°) gratings together with a 0.8 A arcsec slit. Based on the FWHM of the arc lines, the resolution of the sp ectra was determined to b e 3 and 6 ° for A the blue and red sp ectra resp ectively. We reduced these data using optimal extraction (Horne 1986) as implemented in the Pamela1 code (Marsh 1989) which also uses the Starlink2 packages Figaro and Kappa. The wavelength calibration was linearly interp olated from copp er-argon arc lamp exp osures bracketing our observations. We show the mean sp ectrum obtained in the blue arm in the upp er panel of Figure 2. Broad absorption lines of He A I (FWHM30­35°, corresp onding to velocities of 2000 km/s) are detected, which is similar to the discovery sp ectrum of KL Dra (when it was in an outburst) shown in Wood
1

et al. (2002). Unlike the discovery sp ectrum we detected an emission line at 4686 ° (He I I). The red sp ectra are remarkA ably featureless with only a p ossible detection of the He I ° 5876 ° and He I I 7177 A lines in absorption. We searched A for radial velocity shifts in the absorption lines but found none, with an upp er limit of 40 km/s. This is consistent with a disc origin for the lines and a mass ratio typical for AM CVn systems. We also searched for variability in the ° flux of the 4686 A line. However, the low signal-to-noise did not allow us to place any significant constraints on this question.

3.2

Low state spectra

PAMELA was written by T. Marsh and can be found at http://www.warwick.ac.uk/go/trmarsh 2 The Starlink Software Group homepage can b e found at http://starlink.jach.hawaii.edu/starlink c 0000 RAS, MNRAS 000, 000­000

We observed KL Dra on the night of 17 July 2007 with the GMOS instrument in long-slit sp ectroscopy mode on the Gemini North Telescop e at Mauna Kea. The acquisition image allowed us to determine its brightness (g =19.0) using the local comparison stars mentioned in §2 and showed it was in a low optical state. We obtained 60 consecutive 180 sec exp osures and used a 1.0" slit and the B1200 grating with binning factors of 4 (spatial) and 2 (sp ectral), giving a wavelength range of ° A 3800 - 5300A with a disp ersion of 0.46° p er binned pixel. Weather conditions for these observations were good, with photometric transparency and variable seeing around 1 . The low-state sp ectra were reduced in the same manner as the high-state sp ectra. We combine the results into a


4
single averaged sp ectrum which we plot in the lower panel of Figure 2. In contrast to the sp ectrum taken in the high optical state, the low-state sp ectrum is dominated by weak emission lines and are typically double-p eaked and indicative of emission from optically-thin regions in the heliumdominated accretion disc. Moreover, the widths of the lines are very broad (reaching +/- 1800 km/s in the case of the He 4686 ° line) with the presence of an enhancement at veA locities close to zero km/s. This could b e connected to the central `spike' that is seen more prominently in the AM CVn system GP Com which is thought to originate close to the accreting white dwarf (Marsh 1999). flux on a consistent scale (the implied flux in the UVW1 filter) we convolved the effective area curves for the different UVOT filters with white dwarf atmosphere models of different temp eratures (Koester, private communication). For a white dwarf with T =16000 K, this implies a correction factor of 0.68 for the UVW2 data, 1.82 for the U data and 1.23 for the UVM2 data (the results were were similar when we used blackb odies). The correction factors for individual filters are comparable within a temp erature range of ±4000 K. If the temp erature of the UV comp onent changes significantly in the outburst then the fluxes will b e more uncertain. We show the UV light curve in the middle panel of Figure 3. The corrected fluxes in the different UVOT filters now give consistent fluxes during the low optical state. The first p oint to note is that the optical outburst seen at MJD 55173 (2009 Dec 08) is also seen at UV wavelengths and the increase in flux is very similar (a factor of 20). Whilst this result was exp ected, it is the first time (to our knowledge) that an outburst of an AM CVn system has b een seen in b oth optical and UV wavebands. It gives confidence that an outburst seen at UV wavelengths would also b e seen at optical wavelengths. The data is consistent with the UV and optical flux starting to increase at the same time (to within half a day). In contrast, in the case of SS Cyg (which with an orbital p eriod of 6.6 hrs will have a much larger disc than KL Dra) one outburst was observed to b e delayed by 1.5­ 2.0 days at extreme UV energies compared to the optical (Wheatley, Mauche & Mattei 2003). We also note that the initial rise to maximum brightness is relatively short and there is a significant decrease in the UV flux (which is similar to the drop in the optical flux seen at MJD55118, Figure 1), after which there is an increase in the UV flux. We note a brief UV brightening seen at MJD55159 which was also marked by a small increase in the optical brightness. Given that our ground-based optical observations gave some indication that the outbursts rep eated on a timescale of 60 days, we obtained a further set of Swift observations starting on 2010 Feb 09. As exp ected, KL Dra was in a bright UV state (cf, Figure 3). The data folded on a p eriod of 61.5 days (Figure 4) shows a very similar profile to that of the optical data.

4

SWIFT OBSERVATIONS

We obtained target of opp ortunity observations of KL Dra using the NASA Swift satellite (Gehrels et al. 2004) once every two days for two months. The goal of these observations was to characterise the UV and X-ray flux of KL Dra over the course of an outburst cycle. Because Swift is in a low Earth orbit, each observation sequence (which makes up an `ObservationID') is made up of typically 2­4 separate p ointings. The exp osure time of each UVOT image is generally 600 sec in duration, while the total exp osure time of the Xray observation in each ObservationID is typically 2­3 ksec. Observations commenced on 2009 Nov 08 (MJD=55146).

4.1

UVOT observations

The Ultra-Violet/Optical Telescop e (UVOT) on b oard the Swift satellite has a 30 cm primary mirror and 6 optical/UV filters (Roming et al. 2005). The main goal of our UVOT observations was to determine how the UV flux changed over the course of an outburst. Since Swift op erates a `filter of the day', we were not able to pre-define the filter, but images were obtained in either the U (central wavelength 347 nm, and a full width half maximum of 79 nm), UVW1 (251 nm, 70 nm), UVM2 filter (225 nm, 50 nm), or the UVW2 (188nm, 76 nm) filters. To determine the UV flux of KL Dra we used the Swift tool uvotmaghist which is part of the HEASoft3 package of software. This tool takes into account effects such as coincidence loss and converts the count rate to flux based on observations of white dwarfs made as part of the Swift calibration process (Poole et al. 2008). As recommended in the UVOT software guide4 we used a source radius of 3 . We chose a background region free from stars and galaxies. We note that while the nearby galaxy is much brighter at optical bands compared to KL Dra, in the UVW1 filter KL Dra is much brighter than the galaxy and in the UVW2 filter the galaxy is very faint. We found that in the low optical state the UV flux was dep endent on the UV filter which was used ­ eg the flux in the UVW2 filter was consistently higher than the other UV filters. This is due to the fact that the filters are sampling different parts of the sp ectrum of KL Dra. To put the UV
3

4.2

XRT observations

The X-ray Telescop e (XRT) (Burrows et al. 2005) on-b oard Swift has a field of view of 23.6â23.6 arcmin with CCD detectors allowing sp ectral information of X-ray sources to b e determined. It is sensitive over the range 0.2-10keV and has an effective area of 70 cm2 at 1keV (for comparison the Rosat XRT had an effective area of 400 cm2 at 1keV). The XRT has a numb er of modes of op eration (designed to observe gamma-ray bursts at various stages of their evolution) but here we concentrate on the `photon counting' mode which has full imaging and sp ectroscopic information. The data are processed using the standard XRT pip eline and it is these higher level products which we use in our analysis 5 .
5

http://heasarc.gsfc.nasa.gov/docs/software/lheasoft/ http://swift.gsfc.nasa.gov/docs/swift/analysis/ UVOT swguide v2.pdf
4

http://swift.gsfc.nasa.gov/docs/swift/analysis/ xrt swguide v1 2.pdf c 0000 RAS, MNRAS 000, 000­000


The outburst cycle of KL Dra
NH kT Fluxo Fluxu 0.05+0.11 â 1020 (cm-2 ) -0.05 2.8+4.1 keV -1.0 7.7+2.1 â 10-14 (ergs s-1 cm-2 ) (0.1-10keV) -1.8 10.4+2.8 â 10-14 (ergs s-1 cm-2 ) (0.01-100keV) -2.5 2 =1.12 (8 dof )

5

Table 1. The spectral fits to the X-ray spectrum of KL Dra. We used the tbabs absorption model and the vmekal thermal plasma model in XSPEC. Fluxo refers to the observed flux while Fluxu refers to the unabsorbed flux.

To determine the count rate of KL Dra at each ep och we combined the X-ray events from each ObservationID into a corresp onding image using xselect6 . This image was input to the HEASoft tool XIMAGE and the routine SOSTA (which takes into account effects such as vignetting, exp osure and the p oint spread function) to determine the count rate and error at the p osition of KL Dra. We show the count rates for the individual observations in the lower panel of Figure 3. Since the count rates are low, we created images using more than one observation and determined the count rate from these. We also show these results in the lower panel of Figure 3 as thicker symb ols. In the lower panel of Figure 4 we show the X-ray data folded on a 61.5 day p eriod. We created an image from all the X-ray data taken in the low optical state and also the high optical state. The mean count rates (0.00258±0.00029 ct/s for the low optical state and 0.00190±0.00036 ct/s for the high optical state) are over-plotted as thicker lines in Figure 4. Although the mean X-ray flux during the low optical state is higher than that of the high optical state the difference is not significant: we find no evidence that the X-ray flux changed significantly b etween the high and low optical states. We also tested whether there was a change in the soft/hard (0.1­1keV/1­10keV) ratio b etween the low and high optical states ­ there was none. Using XSELECT we initially extracted an X-ray sp ectrum of KL Dra over a time interval when it was in a low optical state. In addition, we extracted a background sp ectrum from a source free region. We used the appropriate resp onse matrix from the Swift calibration files and created an auxiliary file using the HEASoft tool xrtmkarf. We fitted the X-ray sp ectrum of KL Dra in a low optical state using the vmekal thermal plasma model and the tbabs neutral absorption model. Keeping the metallicity fixed at solar (but with the hydrogen abundance fixed at zero) we found a good fit to the data (2 =0.84, 6 degrees of freedom). The hydrogen column density determined from our model fits is consistent with the total hydrogen column density to the edge of the Galaxy in the direction of KL Dra ( 7.4 â 1020 cm-2 . Dickey & Lockman 1990). Since, we found no evidence that the X-ray flux varied b etween the low and high optical states we created a second sp ectrum using all the X-ray data. Using a model in which the metallicity was fixed at solar we obtained a fit with 2 =1.12, 8 dof. We give the fits with associated errors in Table 1.

Figure 3. The light curves of KL Dra made over the course of the Swift observations. Top panel: Ground-based g band observations made using the LT, INT and NOT. In order to make a suitable comparison with the UV and X-ray data we converted the g band mag to flux using the conversion quoted in Holberg & Bergeron (2006). Middle Panel: The UV flux (in units of 10-16 A ergs s-1 cm-2 °-1 ) where we have normalised the flux in the different filters to match that expected in the UVW1 filter (see text for details). Bottom Panel: The X-ray count rates as determined using the Swift XRT.

5

DISTANCE AND LUMINOSITY

6 http://heasarc.nasa.gov/do cs/software/lheasoft/fto ols/ xselect/xselect.html

Ramsay et al. (2006) determined the X-ray and UV luminosities for 8 AM CVn systems: of those, 5 had distances determined using parallax measurements. The X-ray luminosity of those 5 systems decreases as the orbital p eriod increases. Based on this trend, we predict that a system with an orbital p eriod of 25 mins should have an X-ray luminosity of LX 5 â 1030 ergs s-1 . This is very similar to CR Boo (5.2â1030 ergs s-1 , Ramsay et al. 2006) with has an orbital p eriod very close to KL Dra. Our findings imply that KL Dra lies at a distance of 550­850p c (using the standard error on the unabsorb ed b olometric X-ray flux, Table 1). The UV luminosity is more uncertain since it is difficult to constrain the temp erature of the UV comp onent (which derives from the primary white dwarf plus the accretion disc). However, taking the lead from Ramsay et al. (2006), we fix a single blackb ody with a range of temp erature

c 0000 RAS, MNRAS 000, 000­000


6
star is required b efore the predicted mass accretion rate can b e usefully compared with observations.

6

MODELLING THE LIGHT CURVES

Figure 4. The light curves of KL Dra made at three wavebands folded on a period of 61.5 days with a To = MJD 55173.4.

(10000­40000K), and fix the normalisation of the blackb ody so that it matches the measured UV flux (we used the uvred absorption comp onent in XSPEC, Arnaud 1996). Setting the UV flux near the UV maximum (6 â 10-15 ergs s-1 cm-2 , Figure 3) we find a UV luminosity of 3­10â1033 ergs s-1 for a distance of 700p c and a temp erature range 10000­40000K. For a low optical state the UV luminosity reduces by a factor of 20 to 1 - 5 â 1032 ergs s-1 , which is still greater than the X-ray luminosity by a factor of 20. During the low state, the UV flux originates from the accreting white dwarf and the accretion disk. It is therefore difficult to determine the mass accretion rate (which is less than the mass transfer rate, cf. Schoembs & Hartmann 1983) during the low state. However, if we use the high state UV luminosity and L = GM1 Macc /R1 , we obtain Macc 3 - 10 â 1016 g/s (=4.7 - 15.7 â 10-10 M /yr, assuming M1 =0.6 M , and a distance of 550-850 p c). Deloye et al. (2007) predict the mass accretion rate as a function of the mass and structure of the mass donor star and whether it is irradiated and find Macc 5 - 20 â 1015 g/s (which is slightly lower compared with that determined using the high state UV luminosity) for an orbital p eriod of 25 mins. A greater understanding of the origin of the UV emission during the outburst cycle and the nature of the mass donor

We model our optical lightcurve of KL Dra in the framework of the Disc Instability Model (hereafter DIM, e.g. Hameury et al. 1998) which has b een adapted for helium discs by Lasota, Dubus & Kruk (2008). (See Lasota 2001 for a review of the DIM). We calculated model lightcurves for several different sets of parameters: C (the viscosity parameter of the disc in the cold state), H (the hot state), and the mass transfer rate, Mtr . In our simulations we fixed the mass of the accreting white dwarf at 0.6 M . We also allowed the inner radius of the accretion disc, rin , to approach the radius of the white dwarf. The outer radius of the disc, rout , was allowed to vary around a mean of rout = 1.2 â 1010 cm. For our initial simulations we took Mtr = 3 â 1016 g/s and we show a simulated light curve in Figure 5 using C =0.025 and H =0.026. It shows outbursts which rep eat on a 60 day timescale, an amplitude of 3 mag and bright state lasting 2 weeks. In the simulations there is an increase in flux of 0.5 mag b etween the end of one outburst and the start of the next: this is a known deficiency of the DIM (see comments in Lasota 2001). In contrast, if we assume Mtr = 1 â 1017 g/s in our simulations then this low-state flux increase is much greater. Further, they give more asymmetric outburst profiles (with the decline b eing more extended). We also simulated a set of light curves where we assumed that the secondary star was irradiated by X-rays and UV photons emitted in the accretion region and from the white dwarf. Irradiation of the secondary can cause an enhancement of the mass transfer rate. We found that although this gave values for the viscosity parameters similar to that of hydrogen-dominated accreting dwarf novae (eg C 0.02 and H 0.1­0.2, Hameury et al. 1998, Smak 1999) the simulated light curves gave a large increase in brightness (over 1 mag) b etween the end of an outburst and the next. A more detailed investigation can help elucidate how different conditions (different critical values of and T , smaller disc sizes) could affect cold and hot viscosity parameters.

7 7.1

DISCUSSION KL Dra as a helium dwarf nova

The AM CVn systems CR Boo and V803 Cen have b een known for several decades and hence their optical light curves have b een well studied ­ they also have orbital p eriods close to that of KL Dra. The light curves of all three systems have shown a photometric signal on the `sup erhump' p eriod (a signature of the precession p eriod of the accretion disc) when the system is in a high optical state and is a few p ercent longer than the orbital p eriod. On longer timescales, b oth V803 Cen (Patterson et al. 2000) and CR Boo (Patterson et al. 1997) have shown a prominent modulation (1 mag) on a p eriod of 19­22 hrs at various ep ochs: this has b een called the `cycling state'. Patterson et al. (2000) noted that the amplitude and p eriod of these modulations placed these systems on the short
c 0000 RAS, MNRAS 000, 000­000


The outburst cycle of KL Dra

7

known outbursts) also shows the enhancement of X-rays at the start of the outburst (Byckling et al. 2009). One of the key reasons for obtaining Swift observations was to determine how the X-ray flux of KL Dra varied over the course of the outburst cycle. It is clear that X-rays are detected from KL Dra b oth during the low and the high optical states. However, there is no evidence that the X-ray flux changed significantly during the course of the outburst (Figure 3).

8

CONCLUSIONS

Figure 5. The simulated light curve using the Disc Instability Model which uses a helium accretion flow.

p eriod end of the Kukarkin-Parenago p eriod-amplitude relationship of normal dwarf novae outbursts (Warner 1987). They concluded that these near day-long modulations were similar to hydrogen-rich dwarf novae outbursts (in particular the short-p eriod ER UMa systems). On even longer timescales, evidence has b een found for a p eriod of 77 days in V803 Cen (Kato et al. 2000a) and a p eriod of 46 days in CR Boo (Kato et al. 2000b). Both binaries sp end around half the time in a bright state. Whilst our observations of KL Dra are less extensive than those of V803 Cen and CR Boo, they show characteristics which are clearly different to these two AM CVn systems. Indeed, with a p ossible outburst cycle of 60 days and with a duty-cycle of 1/5, KL Dra closely resembles the outbursts seen in the SU UMa or U Gem typ e dwarf novae. However, at this stage we cannot exclude that the outbursts seen in KL Dra are not the AM CVn equivalent of WZ Sge outbursts (dwarf novae which recur on a timescales of years and have outburst amplitudes of 6­9 mag, eg Patterson et al. 2002). As such it makes KL Dra a prime target to investigate the similarities and differences b etween hydrogen-rich and hydrogen-deficient accretion flows.

Our observations of KL Dra show that it undergoes frequent optical outbursts which are also seen at UV wavelengths. Although our coverage is by no means complete, our data are consistent with KL Dra undergoing outbursts on a p eriod of 60 days. The amplitude of the outbursts (3 mag) and duration of the outburst (2 weeks) are very similar to the typical outbursts that seen in the SU UMa or U Gem sub-class of (hydrogen) accreting dwarf novae. As such we encourage other observers, whether they are amateur astronomers who have suitable equipment or team memb ers of survey telescop es which have suitable spatial resolution (to resolve KL Dra from its `nearby' galaxy), to obtain a more complete long term coverage and hence verify the 60 day outburst p eriod. The fact that KL Dra shows such regular outbursts makes it an ideal system with which to investigate helium accretion flows in detail.

9

ACKNOWLEDGEMENTS

7.2

The X-ray flux over the outburst cycle

The hydrogen-accreting dwarf novae have b een much studied at various wavelengths over several decades and show regular outbursts on timescales ranging from a few weeks to months or even years. Many systems show characteristics similar to SS Cyg which showed in one outburst that the increase in the extreme UV emission was delayed by 1.5­2.0 days with resp ect to the optical emission (Wheatley, Mauche & Mattei 2003). In contrast, the hard X-rays, after an initial burst, were suppressed during the remainder of the outburst. To complicate matters, U Gem has shown at least one outburst where the hard X-rays follow the optical and UV flux (Mattei, Mauche & Wheatley 2000). It is not clear if this is due to the relatively high binary inclination of U Gem. More recently GW Lib (which has only had two
c 0000 RAS, MNRAS 000, 000­000

The Liverp ool Telescop e is op erated on the island of La Palma by Liverp ool John Moores University with financial supp ort from the UK Science and Technology Facilities Council. We thank the LT Supp ort Astronomer Chris Moss for his assistance in scheduling our observations and the Swift PI, Neil Gehrels, together with the Swift science and op erations teams for their supp ort of these observations. We also thank Pasi Hakala and Tiina Liimets for obtaining several images using the Nordic Optical Telescop e which is op erated on the island of La Palma jointly by Denmark, Finland, Iceland, Norway, and Sweden. The Isaac Newton and William Herschel Telescop es are op erated on the island of La Palma by the Isaac Newton Group. All three telescop es are located in the Spanish Observatorio del Roque de los Muchachos of the Instituto de Astrofisica de Canarias. Observations were also obtained at the Gemini Observatory, which is op erated by the Association of Universities for Research in Astronomy, Inc., under a coop erative agreement with the NSF on b ehalf of the Gemini partnership: the National Science Foundation (United States), the Science and Technology Facilities Council (United Kingdom), the National Research Council (Canada), CONICYT (Chile), the Australian Research Council (Australia), Ministrio da Cincia e Tecnologia (Brazil) and Ministerio de Ciencia, Tecnologa e Innovacin Productiva (Argentina). We thank Jean Pierre Lasota for useful comments on a previous draft of this pap er. DS acknowledges a STFC Advanced Fellowship.


8
REFERENCES
Arnaud, K. A., 1996, Astronomical Data Analysis Software and Systems V, ed G. Jacoby & J Barnes, ASP Conf Series, 101, 17 Barclay, T., Ramsay, G., Steeghs, D., 2009, ATel, 2203 Burrows, D. N., et al., 2005, SSRv, 120, 165 Byckling, K., Osborne, J. P., Wheatley, P. J., Wynn, G. A., Beardmore, A., Braito, V., Mukai, K., West, R. G., 2009, MNRAS, 399, 1576 Deloye, C. J., Taam, R. E., Winisdoerffer, C., Charbrier, G., 2007, MNRAS, 381, 525 Dickey, J. M., Lockman, F. J., 1990, ARAA, 28, 215 Gehrels, N., et al., 2004, ApJ, 611, 1005 Hameury, J.-M., Menou, K., Dubus, G., Lasota, J.-P., Hure, J-M., 1998, MNRAS, 298, 1048 Holberg, J. B., Bergeron, P., 2006, AJ, 132, 1221 Horne K., 1986, PASP, 98, 609 Jha, S., Garnavich, P., Challis, P., Kirshner, R., Berlind, P., 1998, IAUC, 6983 Kato, T., Stubbings, R., Monard, B., Pearce, A., 2000a, IBVS, 4915 Kato, T., Nogami, D., Baba, H., Hanson, G., Poyner, G., 2000b, MNRAS, 315, 140 Kotko, I., Lasota, J.-P., Dubus, G., 2010, AN, 331, 231 Lasota, J.-P., 2001, NewAR, 45, 449 Lasota, J.-P., Dubus, G., Kruk, K., 2008, A&A, L523 Marsh, T.R., 1989, PASP, 101, 1032 Marsh, T.R., 1999, MNRAS, 304, 443 Mattei, J., Mauche, C., Wheatley, P., 2000, JAAVSO, 28, 160 Nelemans, G., 2005, The Astrophysics of Cataclysmic Variables and Related Ob jects, Proceedings of ASP Conference Vol. 330. Edited by J.-M. Hameury and J.-P. Lasota. San Francisco: Astronomical Society of the Pacific, 2005, p.27 Patterson, J., et al., 1997, PASP, 109, 1100 Patterson, J., Walker, S., Kemp, J., O'Donoghue, D., Bos, M., Stubbings, R., 2000, PASP, 112, 625 Patterson, J., et al., 2002, PASP, 114, 721 Poole, T. S., et al., 2008, MNRAS, 683, 627 Ramsay, G., Groot, P. J., Marsh, T., Nelemans, G., Steeghs, D., Hakala, P., 2006, A&A, 457, 623 Rau, A., Roelofs, G. H. A., Groot, P. J., Marsh, T. R., Nelemans, G., Steeghs, D., Salvato, M., Kasliwal, M. M., 2010, ApJ, 708, 456 Roming, P. W. A., 2005, SSRv, 120, 95 Schoembs, R., Hartmann, K., 1983, A&A, 128, 37 Smak, J., 1983, AcA, 33, 333 Smak, J., 1999, AcA, 49, 391 Steele, I., et al., 2004, SPIE, 5496, 679 Tsugawa, M., Osaki, Y., 1997, PASJ, 49, 75 Warner, B., 1987, Ap&SS, 130, 3 Wheatley, P. J., Mauche, C., Mattei, J., 2003, MNRAS, 345 49 Wood, M. A., Casey, M. J., Garnavich, P. M., Haag, B., 2002, MNRAS, 334, 87

c 0000 RAS, MNRAS 000, 000­000