CATACLYSMIC VARIABLES OBSERVED DURING K2 CAMPAIGNS 0 AND 1

This CV was named 1RXS J0632+2536 (hereafter J0632+2536) in SIMBAD. Ritter & Kolb (2003) and Watson et al. (2015) have classified this target as a DN with an unfiltered magnitude range of 12.4–15.3 in the updated catalog of cataclysmic binaries and the AAVSO International Variable Star Index, respectively, based on the reports from the vsnet-alert archives (e.g., de Miguel et al. 2012a, 2012b). It is also listed in several other catalogs in the infrared, optical, UV and X-ray bands, such as the 2MASS all-sky catalog of point sources (Cutri et al. 2003), WISE All-sky Data Release (Cutri et al. 2012, 2014), the USNO-A2.0 and B1.0 catalogs (Monet 1998; Monet et al. 2003), the SDSS DR7 (Adelman-McCarthy et al. 2009), the GALEX-DR5 (GR5) sources from AIS and MIS (Bianchi et al. 2012), and the ROSAT All-sky Survey Faint Source Catalog (Voges et al. 2000). Despite these detections, it is still a poorly studied CV. GALEX measurements of FUV = 18.42 ± 0.05 mag, NUV = 17.90 ± 0.03 mag and the ROSAT X-ray count rate of 0.02 cts s⁻¹ are typical for CVs. Several DN outbursts in 2009 March and 2012 January are reported in the literature (e.g., Korotkiy & Sokolovsky 2012; Masi 2012; Ohshima 2012a) and are the basis for its classification as a dwarf nova. The two discovery images obtained by a small telescope (13.5 cm) in Ka-Dar Observatory located at the Mountain station of Kazan State University show obvious position deviation from the other observations (Korotkiy & Sokolovsky 2012). Both discrepant observations may be caused by their underestimated errors. All K2 measurements are of the brighter source in the DSS finding chart.

Considering that the SFF corrected light curve has only a negligible improvement of 3.6% for this target, we only analyzed our reduced K2 light curve which is shown in the top panel of Figure 1. Besides a long-term periodic variation clearly seen in the zoomed light curve, there is a sharp decrease of system light occurring around BJD 2456790 with an amplitude of ∼0.25 mag. After the normalization by removing the systematic variations from the original K2 light curve, a periodogram analysis clearly indicates two significant peaks at the period of 0.1621 day (3.89 hr) and 0.3240 day (7.78 hr). The method of phase dispersion minimization (PDM; Stellingwerf 1978), confirms the result of the periodogram. The top plot in Figure 4 shows that the period of 3.89 hr has more power than that of 7.78 hr. However, the phased light curve of the 35 day K2 data folded on the 7.78 hr period, shown in the top panel of Figure 5, proves that this period is the correct one, showing two distinct peaks and dips during the orbit. A photometric period of ∼0.32521 day (7.81 hr) was observed during a previous photometric campaign launched by the VSNET during an outburst that occurred in 2012 from late January to early February (e.g., de Miguel et al. 2012a, 2012b; Dubovsky et al. 2012; Dubovsky 2012). The period of 3.89 hr was also detected in the vsnet-alert report proposed by Kato (2012) while other VSNET reports during the decline from outburst claim that it was variable (Ohshima 2012b, 2012c). The 35 day K2 data clearly demonstrate the stability of the period of 7.78 hr. It is very likely that this is the orbital period of the system and the longer period of 7.81 hr was a superhump period.

Zoom In Zoom Out Reset image size

Zoom In Zoom Out Reset image size

Several spectra were obtained over a timespan of one hour on 2015 February using the MODS spectrograph (Pogge et al. 2010) on the Large Binocular Telescope (LBT). The spectra cover 3400–7400 Å using two gratings with a resolution of about 2000. The summed spectra are shown in Figure 6 while the red spectra at the beginning and end of the timespan are shown in Figure 7. These spectra show many absorption features yielding an approximate spectral type of K5V, which is consistent with a long orbital period system. The strong doubled nature of the Balmer emission lines indicates a peak-to-peak separation of ∼620 km s⁻¹. During the hour timespan, the doubled lines changed to a single feature, as is typically viewed in DN as a disk structure such as a hot spot moves through the emission lines.

Zoom In Zoom Out Reset image size

Zoom In Zoom Out Reset image size

Ohshima (2012a) first regarded that the dips in the light curve are caused by eclipse, but the wide widths (about half an orbital cycle) and shallow depths (0.3 and 0.1 mag) are not consistent with eclipses. The light curve is similar to that of the long period (7.3 hr) system TT Crt (Szkody et al. 1992), where the modulations can be explained by ellipsoidal variations due to the K secondary if the mass ratio M_wd/M_sec is >1 and the inclination is close to eclipsing (∼50°–65°). The stream flow over the disk for the high mass transfer rates expected for long period systems can create hot spots that contribute to the large amplitude of the ellipsoidal variation. However, it is surprising that the maximum brightness comes after the deepest minimum, whereas a hot spot is normally expected prior to the deep minimum. Spectroscopy throughout the orbit can provide confirmation of the correct scenario.

This is a well-known SU UMa-type DN (former name of T Leo) with the explicit detection of several superoutbursts and their corresponding superhumps (e.g., Lemm et al. 1993; Kunjaya et al. 1994; Kato 1997; Patterson et al. 2005; Kato et al. 2009). Shafter & Szkody (1984) and Kato et al. (2009) obtained its orbital and superhump periods of 0.05882 day (1.41 hr) and 0.0604 day (1.45 hr), respectively. There are also other proposed identifications as a WZ Sge-type star (Howell et al. 1999) or IP. The latter identification was suggested by Shafter & Szkody (1984) and Vrielmann et al. (2004) because of a 414 s period observed in the XMM-Newton X-ray light curve that could be the spin period of the white dwarf. Wenzel (1983) derived a mean cycle length of 420 day for its superoutbursts based on four early eruptions reported in the literature. There are five known superoutbursts in 1993, 2005, 2007, 2008, and 2009 observed in the optical band (Kato et al. 2009), and one superoutburst in 1988 observed in the UV band (Hamilton & Sion 2004). Compared with other typical WZ Sge-type stars, the supercycle of QZ Vir is shorter, and more coherent.

During the 80 day K2-C1 observations in short cadence, there was no outburst event, but a secular decline with an amplitude of ∼1 mag. AAVSO data show that a superoutburst began May 13 and lasted until about June 18 (about 16 days after the start of the K2-C1 observations). So, the decline is consistent with the end of the superoutburst. Quadratic and linear functions were used to normalize the original K2 light curve before and after the data gap of K2-C1, respectively. The normalized light curve is shown in Figure 2. Considering the long-cadence SFF corrected light curve had an improvement percentage of less than 10%, the following discussions are based on our deduced light curve in short cadence, which shows quasi-periodic modulations with a mean amplitude of ∼0.6 mag. Note that this amplitude can occasionally reach up to over 1 mag. Moreover, a special flare-like transient with a duration of ∼8 day and an amplitude of ∼0.8 mag marked by the dash rectangles in Figure 2 occurs around BJD 2456878. The zoomed light curve of this transient shown in the top panel of Figure 8 further reveals that this event is also surrounded by other low-level modulations. Although the periodicity of this modulation cannot be visibly seen as in J0632+2536, the DFT result plotted in the second panel of Figure 4 using the program Period04 definitely indicates two significant peaks at the close periods of 0.0588 day (1.41 hr) and 0.0597 day (1.43 hr), which are nearly consistent with the previously derived orbital and superhump periods, respectively. The other peak at the period of 0.0294 day (0.71 hr) shown in this periodogram is the harmonic of the orbital period. In addition, we found that a group of peaks at the central period of 7.92 hr may be related to the pseudo 6 hr period.

Zoom In Zoom Out Reset image size

Since the decline at the beginning of the K2-C1 light curve is the end of a superoutburst, the fact that the detected period of 1.43 hr derived from the whole K2 data set is a little less than that obtained in the literature can be reasonably explained by the transition from the long superhump period to the short orbital period after the superoutburst. Since the data gap basically separates this K2 light curve into the decline and quiescent parts, the two periodograms for the two parts of light curves shown in the top two panels of Figure 9 further verify that the decline is the end of the superoutburst. The top panel indicates that the decay part of the superoutburst only shows the superhump period. The middle panel illustrates that the highest peak has been shifted from the superhump to the orbital period in quiescence. Besides these two periods, the quiescent light curve also includes many unknown periods (sum of these periods may be beat periods), which result in the complicated and irregular variations. Thus, we cannot obtain the distinct orbital and superhump modulations. Note that the peak at the superhump period does not disappear or become very weak in this plot. This may imply that the superhump period of 1.43 hr in QZ Vir is permanent and coexists with the orbital period. Although many periods were detected in quiescence, we cannot detect the spin period of the white dwarf at 414 s and its harmonics, as found by Vrielmann et al. (2004). The shortest period with a considerably weak power is 18.6 minutes. Additionally, the periodogram for the flare-like transient shown in the panel (c) of Figure 9 reveals a curious feature in that both orbital and superhump periods nearly vanished, but the higher harmonic of the orbital period is significant. Furthermore, there is another notable peak at a new period of 2.67 hr, which is not displayed in the other two panels.

Zoom In Zoom Out Reset image size

Wolf (1919) first classified RZ Leo as a possible nova in 1918 with a maximum at 10.5 mag. However, the second outburst in 1984 and the strong double-peaked emission lines suggest that RZ Leo should be a WZ Sge-type star (Cristiani et al. 1985; Mateo et al. 1985). Then, Green (1985), Howell & Michael (1988), and Ishioka et al. (2001) further proposed that RZ Leo should be an SU UMa star due to the detection of superoutbursts lasting 14 days and corresponding superhumps with a period of 0.07616(21)–0.0806(2) day and an amplitude of ∼0.15–0.2 mag. The secondary of RZ Leo is regarded as a normal red dwarf instead of a brown dwarf as implied by the typical WZ Sge stars (Mennickent et al. 1999; Ishioka et al. 2001). In the literature, there were seven recorded outbursts in 1918, 1935, 1952, 1976, 1984, 1989, and 2000 (see Ishioka et al. 2001 and references therein). Hence, Stubbings (2000) estimated that the supercycle of RZ Leo is approximately 10 years. In quiescence, the photometric period of 0.0756(12) day derived by Mennickent et al. (1999) is a little smaller than the spectroscopic period of 0.07651(26) day (Mennickent & Tappert 2001). Patterson et al. (2003) obtained similar photometric and spectroscopic periods with higher precision, which are 0.0760383(4) day and 0.0761(2) day, respectively. Both periods are smaller than the derived superhump period of 0.07868(19) day (Patterson et al. 2003; Kato et al. 2009).

In K2-C1, RZ Leo is another one of three CVs observed in short cadence. The SFF method failed to improve its K2 light curve due to the negative improvement percentage listed in Table 1. The nearly flat K2 light curve in short cadence shown in the middle panel of Figure 2 never shows any long-term variation, but does have regular and low-amplitude oscillations (∼0.5 mag on average) in quiescence. The Lomb–Scargle periodogram plotted in the middle panel of Figure 4 indicates two notable periods of 0.03801 day (0.91 hr) and 0.07603 day (1.82 hr). Note that besides these two periods, a shorter period of 220.7 s out of this periodogram can be detected. This periodogram is similar to Figure 8 of Patterson et al. (2003). Additionally, the PDM method also confirmed this result. Hence, like J0632+2536, the period of 0.07603 day with less power should be an authentic period, which agrees with the previous result derived from the double-hump photometric light curves (Patterson et al. 2003). Furthermore, based on the 82 day K2-C1 data in short cadence, this period can be improved by the O-C technique due to the high time resolution of the K2 data. Considering that the irregular profile of the second hump seriously distorts the two dips, the maximum time of primary hump should be the best periodic signal for our O-C analysis. The first maximum time and the period obtained from the periodogram were set to be an initial epoch and period, respectively. A primary ephemeris based on a time baseline over 1000 cycles can be used to rephase the short-cadence K2 data. However, when binning this phased light curve with a phase resolution 0.001, there is a small zero-drift in phase ∼0.015 due to the incorrect epoch. The new ephemeris after removing this zero-drift effect was deduced as

$$\begin{eqnarray}&&{T}_{\mathrm{max}}=\mathrm{BJD}\;2456808.23662(2)+\;0.07602997(4)E,\end{eqnarray} \tag{ 2 }$$

with a variance of $7.4\times {10}^{-7}$ day. The precision of the period shown in Equation (2) is about one order of magnitude higher than the period derived by Patterson et al. (2003). Figure 10 displays all the O-Cs of the maximum times of primary hump based on the ephemeris Equation (2). By using the new derived period listed in Equation (2) to fold the linearly normalized K2 light curve, we found that the phased light curve shows very large scatter, ∼0.43 mag on average. Compared with the phased light curve derived by Patterson et al. (2003), our smooth binned light curve with higher phase resolution shown in the middle panel of Figure 5 distinctly illustrates more details of the orbital light curve including a round bottom at the primary dip and the round tops at the two peaks. A flat bottom at the second dip lasting phase 0.05 (∼5.5 minutes) is a consistent feature.

Zoom In Zoom Out Reset image size

This phased light curve illustrates a brightening transient event with an amplitude of ∼0.6 mag, which is also displayed in the middle panel of Figure 8. Both plots clearly demonstrate that this event takes place at BJD 2456845.1, which is coincident with the original second hump. In the phased light curve, the top part of this transient event between phase 0.3 and 0.9 obviously changes the normal second hump. The complete profile of this transient shown in the plot of time series indicates that its duration is ∼1.88 hr, which is a little longer than the orbital period, and it is composed of a rapid rise with a rate of ∼1.6 mag hr⁻¹, a flat peak with a rate of ∼0.05 mag hr⁻¹, and a slow decline with a rate of ∼0.5 mag hr⁻¹. This event looks like a mini DN outburst with a faster and lower amplitude. In addition, there is a weak rebrightening event occurring in the decay with a typical luminosity level compared with the other primary peaks. This event is consistent with the ephemeris of Equation (2). So it is a normal primary hump and is marked by a short vertical line in the middle panel of Figure 8.

LBQS 1144+0111, an alias of this target, was first observed in the Large, Bright QSO Survey (LBQS, Foltz et al. 1987, 1989), and classified as a DA+dMe binary, which is composed of a DA-type white dwarf and a normal M dwarf star with strong emission features (Berg et al. 1992). Additionally, it was listed in a catalog of spectroscopically identified white dwarfs in the first data release of the SDSS (g = 18.09 mag; Kleinman et al. 2004). Although Tappert et al. (2004) did not detect any variability in later time-series photometry, it was claimed as a CV named WD 1144+011 in SIMBAD. However, its optical spectrum obtained by Berg et al. (1992), shows many absorption lines and a low blue continuum, which is atypical for a normal CV. Thus, its CV identification is still doubtful.

The continuous K2 light curve in long cadence covering 80 days shows an orbital modulation but no outburst or transient event. There is a long-term rise of the system light after the data gap of K2-C1. Therefore, we used linear and quadratic functions to normalize the K2 light curves before and after the data gap, respectively. Note that a small transient event marked by the dash rectangle in the bottom panel of Figure 2 appears at the end of the K2 light curve around BJD 2456888. The zoomed light curve of this transient shown in the bottom panel of Figure 8 illustrates that it is a short-timescale event lasting ∼0.5 day, which involves a temporary disappearance of the orbital modulation. Since there is no notice in the data release notes for K2-C1 or any similar variation in the other CVs in K2-C1, this appears to be a real feature of WD 1144+011. The orbital modulation with an average amplitude of 0.03 mag reappears after this transient event. By using the Lomb–Scargle periodogram and PDM methods, a highly significant period of 0.40859 day (9.806 hr) and its harmonics are found based on the normalized data. Since the K2 light curve cannot be used for a reliable O-C analysis like RZ Leo due to the sparse data points, we attempted to use a sinusoidal function to fit the normalized light curve according to the Levenberg–Marquardt algorithm. The best least-square fitting curve shown in Figure 11 can be deduced as follows:

$$\begin{eqnarray}{M}_{\mathrm{nor}} & = & -9.9650(2)\times {10}^{-4}+1.905555(2)\\ & & \times {10}^{-2}\;{SIN}\left[2\pi \displaystyle \frac{{T}_{\mathrm{BJD}}}{0.408680655(1)}-2.524317(2)\right],\end{eqnarray} \tag{ 3 }$$

where M_nor and T_BJD are the normalized magnitude and the modified BJD (i.e., T_BJD = BJD − 2456000), respectively. The best-fit period of 0.408680655(1) day in Equation (3) basically accords with the primary period shown in the periodogram. Since the SFF improvement percentage is 10.6%, we attempted to phase the SFF corrected light curve based on the period in Equation (3) and obtained a nearly equal phase gap at a frequency of 20 cycle/day (i.e., ∼19 minutes interval) shown in row (a) of Figure 12. Although our normalized light curve folded by the same period shown in row (c) presents a similar gap, the zoomed-in light curve in the right-side panel of row (c) indicates that each section of the phased light curve is smoother than that of row (a). In addition, by using the DFT method (Period04 program), we obtained a new period of 0.408741 day (9.81 hr), which only has a small difference 5.2 s from the period listed in Equation (3). In spite of this small discrepancy, the new phased SFF corrected light curve shown in row (b) illustrates a distinctly different pattern of a sine-like modulation with a phased period of 0.02 day (i.e., ∼29.4 minutes) superimposed on the original phased light curve without any gap. This modulation is totally invisible in the new phased light curve for the normalized data, which only displays smooth scatter. Note that the modulation period of 29.4 minutes presented in row (b) is exactly the same as the sampling rate of the K2 in LC and one twentieth of the folding period of 9.81 hr. Therefore, this sine-like modulation derived from the SFF corrected data is likely not real since it could be removed after a simple normalization. The 19 minutes phase gaps shown in rows (a) and (c) may be caused by the inaccurate folding period. The binned phased light curve shown in the bottom panel of Figure 5 has a slow rise and fast decline. The long period and the light curve shape imply this system is likely a pre-CV with the modulation caused by the changing view of the M star that is heated by the hot white dwarf. Time resolved spectroscopy throughout the 9.81 hr period should show noticeable changes in the emission lines.

Zoom In Zoom Out Reset image size

Zoom In Zoom Out Reset image size

This object named USNO-B1.01144-00115322 (hereafter USNO 01144) was first listed in the USNO-B1.0 catalog (Monet et al. 2003). Kryachko et al. (2009) claimed that it showed a DN outburst with a duration of >7 day and an amplitude of >4 mag as observed at KSU Astrotel Observatory in 2009 November. Its optical spectra on 2009 November 19 obtained by Pejcha (2009) show broad Balmer absorption lines with a narrow emission core and a weak Hα emission line in the blue and red wavelengths, respectively. In addition, the continuum in the blue end is obviously stronger than that in the red end.

During the observations with K2-C0 in long cadence, a brightening event was detected around BJD 2456770. Although the K2 light curve shown in Figure 1 was incomplete due to missing the beginning part of the outburst, it was enough to confirm that this is a typical normal outburst with an outburst amplitude of ∼3.5 mag, which further supports the previous classification of DN. However, the SFF corrected light curve with the significant improvement percentage 209.6% never recorded this brightening event since the SFF data just began from BJD 2456772.109, close to the end of this normal outburst. Based on the systematic difference between our normalized SAP flux and the SFF corrected data, the SFF corrected light curve was replotted in the middle panel of Figure 1 with the same scale as our deduced K2 light curve. We found that the large scatter with an amplitude of ∼1 mag has been almost eliminated by the SFF method. In particular, the sudden introduction of large scatters at the end of the smooth decay of normal outburtst can be naturally explained by its faint quiescence. The plot of the differences between the two light curves illustrate an obvious deviation lasting ∼5 days at the end of the decay section, which is caused by the very flat SFF light curve. Considering that the diagram of the differences between the two light curves in quiescence is basically flat as shown in the middle panel of Figure 1, the normal outburst with a smooth and consecutive profile presented in our deduced K2 light curve is authentic. Thus, a new and complete K2 light curve of this object can be composed of our deduced part in outburst and the SFF corrected part in quiescence. A period search based on this complete K2 light curve only gives the null result.

The three superoutbursts recorded in 2003, 2008, and 2011 accompanied with superhumps definitely classify UV Gem as an SU UMa-type star with a long superhump period (e.g., Kato & Makoto 2001; Kato et al. 2009). The O-C diagram of the superhumps derived by Kato et al. (2013) show a downward quadratic curve covering these three superoutbursts with an extremely negative decline rate of −5.34 × 10⁻⁴ on the basis of the superhump period of 0.0936 day derived from the superoutburst in 2003. Kato & Makoto (2001) estimated that the cycle length of the normal outbursts is ∼13.5 day, but there is not enough data to derive the supercycle length (Otulakowska-Hypka & Olech 2013). Until now, only one quiescent spectrum with a weak Hα emission line and typical continuum covering 4600–9000 Å was obtained by Zwitter & Munari (1994). Thus, its orbital period is still unclear (Ritter & Kolb 2003).

Fortunately, a complete normal outburst lasting 5 days with an amplitude of around 3 mag occurring around BJD 2456788 was detected in K2-C0. Considering that the SFF method successfully improved the quiescent light curve of USNO 01144, the SFF corrected light curve was checked but there was a very low improvement percentage 0.8% for UV Gem. In order to clearly demonstrate the coherence of the two light curves, they are replotted into the same frame and their almost flat differences are shown in the bottom panel of Figure 1 except for a systematic shifting ∼0.1 mag from BJD 2456783 to BJD 2456794, which totally covers the whole outburst. Like USNO 01144, the flat differences imply that the normal outburst detected in our deduced light curve is also real. In addition, the large scatter and the quasi-periodic jitters at the peak and decay parts of the outburst light curve clearly appeared in the SFF corrected light curve. The two Lomb–Scargle periodograms for our deduced data and the SFF corrected data shown in the UV Gem panels (a) and (b) of Figure 13 demonstrate that the known 6 hr pseudo periodic signal caused by the regular thrusting event of the Kepler satellite is not eliminated in the SFF data, but is removed from our deduced data. Additionally, the two significant peaks at the period of 0.086089 day (2.07 hr) and 0.088342 day (2.12 hr), respectively, appeared in the UV Gem (a) panel of Figure 13. Both periods are less than the orbital period of 0.0895 day (2.15 hr) listed in the updated CVs catalog (RKcat 7.21, Ritter & Kolb 2003), which was estimated from the superhump period of 0.0935 day (2.24 hr) using an empirical relation given by Stolz & Schoembs (1984). Hence, the K2 data allow the first determination of a plausible orbital period for UV Gem. Moreover, there is still a weak but analogous frequency feature displayed in the UV Gem (b) panel of Figure 13. Considering that this periodogram is similar to QZ Vir, the longer 2.12 hr period with less power may be the remaining superhump period after an unrecorded superoutburst. However, the phased light curve cannot be obtained due to the sparse K2 data available in long cadence.

Zoom In Zoom Out Reset image size

In SIMBAD, this target, named 1RXS J111236.7+002807, is classified as an X-ray source as it first appeared in the ROSAT All-sky Survey Faint Source Catalog (Voges et al. 2000), and has a typical CV count rate of 0.02368. However, it was regarded as a galaxy (g = 22.46 mag) in the SDSS DR 9 in spite of a position deviation of ∼14.8 arcsec compared with the coordinates in SIMBAD. Recently, the CRTS classified this target as a CV due to the detection of several transient variations with a nearly 4 mag amplitude (i.e., 17–21 mag) in low time resolution data obtained during the last 10 years (Drake et al. 2009). Hence, the faint g-band magnitude listed in the SDSS catalog corresponds to quiescence during the SDSS observations.

A typical superoutburst lasting 17 days with a weak precursor and an amplitude of ∼5.6 mag was detected in the long cadence K2-C1 data. The converted K2 magnitude is consistent with the previous observations of the CRTS and the SDSS. Although the scatter in the quiescent light curve is not apparent in the upper panel of Figure 3, it is obviously present in the magnitude plot (i.e., the log graph) since quiescence may reach the detection limit of the satellite. Considering that the SFF method only provides a negative improvement percentage −19.5%, we cannot expect that the scatter in quiescence can be reduced by the SFF method. In order to visually display the superhumps on the plateau of the superoutburst, we used the dotted lines to connect the adjoining data points due to the low time resolution in long cadence. The zoomed plots in SAP flux and Kp₂ are shown in panels (a) and (b) of Figure 14, respectively. Both diagrams further demonstrate that it is hard to reliably detect the superhump from such sparse time series, which only has 4–5 data points in each superhump cycle at most. In spite of this, we attempted to use a linear function to normalize the plateau in magnitude, which can be described as follows:

$$\begin{eqnarray}&&{\mathrm{Kp}}_{2}=-59.813(\pm 0.499)+\;0.088639(\pm 0.000574){T}_{\mathrm{BJD}}.\end{eqnarray} \tag{ 4 }$$

This can be converted into an exponential function for describing the plateau in SAP flux by using Equation (1). However, the normalized plateau in magnitude shown in panel (c) of Figure 14 shows a quasi-sinusoidal variation with an amplitude of ∼0.05 mag and a period of ∼7.6 day except for the small deviation at the end of the plateau. After a further normalization by subtracting this sine curve, a flat plateau shown in panel (d) of Figure 14 was obtained. Based on this final normalized data, a Lomb–Scargle periodogram shown in the bottom panel of Figure 4 reveals two significant peaks with comparable power at 0.060137 day (1.44 hr) and 0.30068 day (0.72 hr). However, the PDM method cannot find any period from the same data. Considering that the minimum orbital period for hydrogen-rich CVs is ∼70 minutes (Warner 2003), it is reasonable to conclude that the smaller period of 0.30068 day is only a harmonic, but the superhump period of this object is the other period of 0.060137 day due to the lack of peaks at longer periods. This implies that this object is a short-period SU UMa-type star. Although the superhump period derived from this sparse K2 data may be not very accurate, the K2 light curve in long cadence can definitely confirm the detection of a typical superoutburst and the corresponding superhump with a plausible period of 1.44 hr. This means that CSS 1112+0028 is an SU UMa-type DN close to the period minimum. Note that the known 6 hr pseudo period is also not visible in the periodogram. Further followup observations in higher time resolution are necessary to precisely certify its superhump and orbital periods.

Zoom In Zoom Out Reset image size

This has been well known as a UG Gem-type DN since 1932 (O'Connell 1932). Shafter (1983) first obtained its orbital period of 4.38 hr, the orbital inclination 43° and the mass ratio of 0.44. Ak et al. (2002) claimed that there were 7872 observations including 62 normal outbursts for TW Vir from 1955 to 1997. Based on these observations, they determined that the average outburst duration and quiescent interval of TW Vir should be 6 and 22 days by setting the two reference lines at 12.2 and 13.5 mag. Besides this optical photometry, there were some detailed UV and IR observations. The UV spectrum in outburst taken by the IUE shows a P Cygnic profile of the CIV 1550 line, which indicates a strong accretion disk wind (Córdova & Mason 1982; Hamilton et al. 2007). Szkody (1985) analyzed other IUE spectra when TW Vir switched to quiescence, and found that the CIV line returned back to a normal emission line with a larger equivalent width (145 Å) than the Balmer lines. Additionally, Mateo et al. (1985) obtained a sinusoidal IR light curve with a period of 2.2 hr (i.e., one-half the orbital period) and a half amplitude of 0.063 mag. They estimated that the secondary of TW Vir is a M3V red dwarf of 0.4 ${M}_{\odot }$.

Like QZ Vir and RZ Leo, TW Vir was observed in short cadence with K2-C1. The K2 light curve shown in the bottom panel of Figure 3 recorded two incomplete normal outbursts and one completed superoutburst lasting 16 days with a typical precursor on the rise to superoutburst. The amplitude of this superoutburst is ∼4.5 mag, which is 0.5 mag larger than the following two normal outbursts. The quiescent interval between them is less than 26 days, which is consistent with the average result derived by Ak et al. (2002). The amplitude and the duration of this superoutburst of TW Vir are 1.1 mag lower and 1 day shorter than those of CSS 1112+0028, respectively. Considering that the improvement percentage by the SFF method is 16.9% for the light curve in long cadence, the SFF corrected data and our reduced data were replotted in the same diagram like USNO 01144 and UV Gem. Inspection of the bottom panel of Figure 3 indicates that the obvious deviations between the two light curves are at the beginning of the time series including the plateau of the superoutburst, where the SFF data present larger scatter. But, the remaining parts of the two light curves are similar. The two panels illustrate that the SFF light curve undergoes an unexpected break-off point near the end of the plateau. A periodogram for the plateau of the superoutburst based on the SFF data is shown in the TW Vir panel (b) of Figure 13. A peak at the period of 6 hr is the only significant feature, and there is not any signal around the orbital period. For comparison, the periodogram of our deduced data in the same data range shown in the TW Vir panel (a) indicates that the plateau also seriously suffers from the effect of the pseudo 6 hr period. But, there are some weak peaks around the orbital period of 4.38 hr.