PARALLAX AND DISTANCE ESTIMATES FOR TWELVE CATACLYSMIC VARIABLE STARS (2024)

Table 1.Journal of Observations

Notes. Overview of the data included in the parallax solutions. N_ref is the number of reference stars used to define the plate solution, N_meas is the total number of stars measured, and N_pix is the number of images used. The epochs represent different observing runs, and the numbers in parentheses are the number of images included from each run.

Download table as: ASCIITypeset image

Table 2.Positions, Magnitudes, Parallaxes, and Proper Motions

Notes. Parameters for all measured stars in all the fields. The program star in each field is marked with an asterisk. The celestial coordinates are from mean CCD images and are referred to the USNO A2.0 or UCAC2, which are aligned with the ICRS; the epochs of the images used are typically around 2005. Coordinates should be accurate to ∼03 external and somewhat better than this internally. A 1 or 0 in the next column indicates whether a star was used as a reference star. The next column gives the scatter around the best astrometric fit (see text); very large scatter indicates some kind of problem with the centering of that particular star (caused, e.g., by saturation on some images). The V and V − I colors come next, with typical external uncertainties of 0.05 mag. Next come the fitted parallaxes, proper motions in X and Y, and the uncertainty in the proper motion (per coordinate).

Only a portion of this table is shown here to demonstrate its form and content. Machine-readable and Virtual Observatory (VO) versions of the full table are available.

Download table as: Machine-readable (MRT)Virtual Observatory (VOT)Typeset image

Table 3.Parallaxes, Proper Motions, and Distances

Notes.Summary of the parallaxes, proper motions, prior information, and distance estimates. Two lines are given for each star, as follows. Line 1. Column 1: Name of star. Column 2: relative parallax and its uncertainty; the first uncertainty given is from the formal error of the fit and the estimated external error is given in square brackets. Column 3: adopted absolute parallax and its error (in parentheses). Column 4: relative proper motion in X (i.e., eastward) and Y(northward), and the formal error per coordinate in parentheses. Column 5: distance implied by the absolute parallax, and its uncertainty, with no further corrections applied. Line 2.Column 1: left blank. Column 2: 50th percentile of the distance probability distribution after the Lutz–Kelker correction is applied. Column 3: 50th percentile distance estimate using the parallax and proper motion information, and a model for the intrinsic CV velocity distribution. The uncertainties given are the 16th and 84th percentile points, equivalent to ±1σ. Column 4: prior distance estimate. Where two numbers are given, they are the distance modulus m − M and its assumed uncertainty; where there are three numbers, they are an apparent magnitude m, an absolute magnitude M, and an uncertainty in m − M, respectively. Column 5: the 50th percentile Bayesian distance estimate, including all the prior information. Uncertainties are again 16th and 84th percentile points. ^aThe photometric distance constraint for KT Per is from the spectroscopic parallax of the common-proper-motion companion; see text for details.

Download table as: ASCIITypeset image

DW Cnc is a remarkable CV, unique in that the radial velocities of its very strong emission lines are modulated on two distinct, incommensurate periods, 86.1015(3) and 69.9133(10) minutes, each of these periods having similar velocity amplitude. The light curve is modulated at the beat frequency of the two velocity periods. The three periods are most readily explained as the orbital period P_orb, the rotation period of a magnetized white dwarf, and the orbital sideband (Patterson et al. 2004; Rodríguez-Gil et al. 2004).

Although the parallax is determined with good precision, the relative error σ_π/π is substantial, leading to a sizeable Lutz–Kelker correction. Because the system is unusual, we unweighted the distance prior in the Bayesian analysis by assuming a large uncertainty in luminosity. We find a proper motion [μ_X, μ_Y] = [ − 26, + 5] mas yr⁻¹, consistent with the USNO B value of [ − 22 ± 5, + 0 ± 2] (Monet et al. 2003). Our final distance estimate, 257(+79, − 52) pc, puts DW Cnc near the peak of prior distance probability based on the proper motion alone, which in turn is based on the observation that most CVs have modest space velocities (Paper I).

Patterson et al. (2004) reported extensive photometry giving V = 14.8 for DW Cnc, which corresponds to M_V = 7.8 at 250 pc. This is within the range of values predicted by the Warner (1987) relation for an outbursting dwarf nova at P_orb = 86 minutes; literal agreement would require an inclination i ∼  83°. An inclination this large is ruled out by the lack of eclipses. The Warner (1987) relation was derived for dwarf novae at maximum light, and DW Cnc is not a dwarf nova; nonetheless, DW Cnc is clearly much more luminous than short-period dwarf novae at minimum light. It is evidently a rare example of a nova-like variable with P_orb < 2 hr.

HT Cas is a well-studied, deeply-eclipsing dwarf nova of the SU UMa subclass, with P_orb = 106 minutes. Feline et al. (2005) presented recent high-speed photometry and summarized some of the earlier work on this object. Because HT Cas eclipses, there is a long history of detailed time-resolved studies aimed at mapping the disk emission and constraining the white-dwarf radius, which has led to several estimates of the distance. Marsh (1990) detected the secondary star, classified it as M5.4 ± 0.3, and measured the secondary's contribution to total light at eclipse. Horne et al. (1991) derived dynamical constraints using high-speed photometry of the eclipses; from these, and the secondary's contribution, they estimated a distance of 135 pc. Vrielmann et al. (2002) gave a critical discussion of previous distance estimates, and using a different mapping technique, derived a distance of 207 ± 10 pc.

The parallax measurement decisively favors the shorter distance; the Vrielmann et al. (2002) distance is essentially excluded. While the ∼130 pc distance favored by Marsh (1990) is a little longer than our astrometric distance, it lies well within the uncertainties. Our final Bayesian distance estimate includes prior distance information; for this, we adopted a rough average value of previous estimates and assumed an uncertainty generous enough to include both the longer and shorter distances, namely m − M = 6.1 ± 1, or 166(+97, − 61) pc. The parallax is accurate enough such that this assumption increases the distance estimate by only 8 pc.

The proper motion found here, [μ_X, μ_Y] = [ + 27, − 14] mas yr⁻¹, agrees well with that listed in the Lick NPM2 catalog, [ + 28, − 10] mas yr⁻¹ (Hanson et al. 2004). Assuming that the distance is at the far end of our 1σ error bar, the transverse velocity referred to the local standard of rest (LSR) is only ∼6 km s⁻¹. Due to the faintness of the secondary, we were unable to find a precise systemic radial velocity γ in the literature, but emission-line studies suggest that γ is some tens of km s⁻¹ or less (Marsh 1990; Young et al. 1981). The kinematics of HT Cas are clearly those of the disk.

MQ Dra was discovered in the Sloan Digital Sky Survey (SDSS; Szkody et al. 2003a), in which it was called SDSS J155331.12+551614.5. The spectrum shows a white dwarf with an ∼60 MG magnetic field, and only weak emission features, indicating a very low accretion rate (Szkody et al. 2003b). It is, therefore, a fine example of a low accretion rate polar (LARP; see, e.g., Schmidt et al. 2005). It has P_orb = 4.39 hr, and the secondary star's spectral type is ∼M4.5 (Harrison et al. 2005). The secondary's contribution to the light leads to a distance estimate of 130–180 pc (Schmidt et al. 2005; Szkody et al. 2008).

Although the field of MQ Dra is somewhat sparse, the parallax is measured with a formal uncertainty of only 0.7 mas; the scatter of apparent field star parallaxes suggests that this is a realistic uncertainty, but we adopt 1.0 mas to be conservative and to allow for the unmodeled uncertainty in the correction from π_rel to π_abs. Our measured relative proper motion [μ_X, μ_Y] = [ − 31, + 6] mas yr⁻¹ agrees well with the proper motion listed in the USNO B, which is [ − 28 ± 5, − 8 ± 3] in the same units (Monet et al. 2003). The parallax is accurate enough so that the various corrections do not strongly affect the distance estimate. For the final Bayesian estimate, we include a secondary-based distance modulus corresponding to d = 150(+60, − 40) pc. Combining this with the parallax and proper motion yields d = 162(+27, − 21) pc for the final estimate.

IR Gem is an SU UMa-type dwarf nova with P_orb = 98.5 minutes (Feinswog et al. 1988; Fu et al. 2004). The parallax is so small that it is barely detected, despite the large and favorable set of comparison stars available at IR Gem's Galactic latitude. The proper motion is sizeable; we measure [μ_X, μ_Y] = [ + 53.9, − 22.3] mas yr⁻¹, in good agreement with the UCAC2 catalog (Zacharias et al. 2004), which gives [ + 51.9 ± 3.4, − 28.8 ± 3.4] mas yr⁻¹. Because of the small parallax, the Lutz–Kelker correction depends critically on the uncertainty adopted, which we conservatively estimate using the scatter in the fitted parallaxes of other stars in the field (see Paper I).

In the case of IR Gem, π_abs/σ_π < 4, so there is no local maximum in the Lutz–Kelker parallax probability function; the parallax alone does not formally give a distance. With such weak parallax evidence, the Bayesian priors steer the distance estimate. For a distance prior, we use the Warner (1987) relations. The GCVS (Kholopov et al. 1999) lists V = 10.7 at maximum light. Since superoutbursts are ∼1 mag brighter than the normal outbursts for which the Warner (1987) relation is calibrated, we adopt V = 11.7 in computing the distance prior. Applying the relation requires an inclination estimate; Feinswog et al. (1988) noted that the radial velocity amplitude is low, which suggests an orbital inclination not too far from face-on. At low inclinations, the correction is fortunately nearly constant; we arbitrarily adopt i = 30°, leading to a prediction of M_V(max) = 4.6 for P_orb = 98 minutes. The resulting distance is ∼250 pc. We adopt an uncertainty of ±1 mag in the distance modulus for the Bayesian prior.

The Bayesian results are as follows. The large proper motion, interpreted using the space velocity probability density assumed in Paper I, eases the Lutz–Kelker problem and results in a 50th percentile distance near 410 pc. The prior distance estimate decreases this to 358(+104, − 86) pc. At 358 pc, the transverse velocity corresponds to 98 km s⁻¹ referred to the LSR; including a systemic radial velocity γ = +60 km s⁻¹ (estimated from Feinswog et al. 1988) gives a space velocity of ∼115 km s⁻¹. This puts IR Gem out on the high-velocity tail of the assumed CV velocity distribution.

V396 Hya is an AM CVn-type system, in which the secondary is evidently an evolved degenerate star with no hydrogen remaining. It was discovered as a high proper motion star in the Calan-ESO survey (Ruiz et al. 2001b), and designated CE 315 (Ruiz et al. 2001a). Its orbital period, 65 minutes, is the longest of all known AM CVn systems. Systems with degenerate secondaries tend to evolve toward longer orbital periods, so V396 Hya is likely to be among the most highly-evolved AM CVn stars known. Bildsten et al. (2006) found that, for these relatively long-period AM CVn stars, the luminosity arises largely from the accreting white dwarf, and predicted a system luminosity. They used a preliminary version of our parallax measurement to show that the luminosity of V396 Hya agrees well with theory. At V = 17.6, V396 Hya is just little too faint for the HST FGS.

In addition to our parallax determination, we have some spectra of V396 Hya taken with the Hiltner telescope and modular spectrograph; the instrumentation, procedures, and reductions were essentially as described by Thorstensen et al. (2004). We took 35 exposures, most of them 5 minutes, on the nights of 2001 May 14, 15, 16, and 17 UT. The average spectrum appeared very similar to that shown by Ruiz et al. (2001a). We measured radial velocities of the wings of the He i λ5876 emission line using a double-Gaussian convolution algorithm (Schneider & Young 1980a; Shafter 1983), and searched for periods using a "residual-gram" algorithm (Thorstensen et al. 1996). The Ruiz et al. (2001a) period was clearly detected as the strongest among several aliases (which are inevitable because the observations are clustered modulo 24 hr). By fitting the velocities from all the nights, we refine P_orb to 65.07 ±  0.08 minutes. The emission line profiles in our average spectrum are mostly triple-peaked, as described by Ruiz et al. (2001a). In an effort to constrain the systemic radial velocity γ, we fit the tops of the central peaks with Gaussians using the interactive splot command in IRAF. The different lines gave velocities ranging from +30 to +70 km s⁻¹ (heliocentric); from a weighted average, we estimate v = +45 ± 20 km s⁻¹ for the line core. It is entirely possible that this does not accurately reflect the systemic velocity.

The parallax is determined to good relative accuracy and dominates the Bayesian estimate of distance. We find a proper motion of [ − 277, − 52] mas yr⁻¹, significantly smaller than the Calan-ESO value, but in reasonable agreement with the USNO B, which lists [ − 264, − 30] mas yr⁻¹ (Monet et al. 2003). This large proper motion more than compensates for the Lutz–Kelker effect in the Bayesian calculation, so the 50th percentile distance of 92 pc is essentially equal to 1/π_abs. The final estimate is slightly larger than the 77 pc preliminary estimate used by Bildsten et al. (2006), but their conclusions should be essentially unchanged. Some insight into the kinematics of this system can be had by computing the tangential velocity, referring it to the LSR, and rotating into Galactic coordinates; for a distance of 90 pc, this yields [U, V, W] = [ + 79, − 68, 0] km s⁻¹—oddly enough, nearly parallel to the plane of the Galaxy. If we adopt γ = +45 km s⁻¹, this becomes [ + 57, − 95, + 28] km s⁻¹. In any case, these are not the kinematics of the thin disk (though the system apparently does not stray far from the plane). It is interesting to note that another AM CVn system, GP Com, evidently also has a high space velocity (Thorstensen 2003). It may be significant that both V396 Hya and GP Com, two of only ∼18 known AM CVn stars (Roelofs et al. 2007b), also belong to the small handful of CVs that were first discovered because of their proper motions. This suggests that AM CVn systems are fairly common, or that many are members of an old population, or both. However, Roelofs et al. (2007a) suggested that the space density is not particularly large, based on the HST parallaxes of a sample of AM CVn stars.

KT Per is a Z Cam-type dwarf nova with P_orb near 3.90 h (Ratering et al. 1993; Thorstensen & Ringwald 1997). We chose it for parallax determination because of its substantial proper motion in the UCAC2, [μ_X, μ_Y] = [ + 63, − 5] mas yr⁻¹ (Zacharias et al. 2004). Also, Thorstensen & Ringwald (1997) estimated a distance of 245 ± 100 pc based on the M3 secondary star, which is close enough for parallax to provide a useful check.

Our astrometry shows that the V = 15.19 field star, located 15.4 arcsec west and 2.3 arcsec north of KT Per, is evidently a physical companion, sharing with KT Per a nearly identical large proper motion and parallax. We obtained spectra of this star with the Hiltner 2.4 m telescope and calibrated them using observations of flux standards. We have a library of spectra of K dwarfs classified by Keenan & McNeil (1989), taken with the same instrument. Using these for comparison, we estimate the spectral type of the companion to be K4.5 ± 1 subclass. Pickles (1998) tabulated mean colors as a function of the spectral type. We measure V − I = +1.59 for this star, which, by his table, would indicate a K7 to M0 star, significantly later than observed. However, La Dous (1991) estimated E(B − V) = +0.2 for KT Per, using the strength of the 2200 Å extinction feature in archival IUE spectra. Combining this with E(V − I)/E(B − V) = 1.35 (He et al. 1995) yields (V − I)₀ = 1.32, which is nicely consistent with the spectral type. As a check, we note that the reddening maps of Schlegel et al. (1998) give E(B − V) = 0.32 for the total Galactic extinction in this direction (l = 130°.2, b = −11°.3); it is not implausible that KT Per is extinguished by a substantial fraction of this total. Taking Pickles' mean value of M_V = 7.64 for this star, and taking A_V = 3.2E(B − V), yields a spectroscopic distance modulus (m − M)₀ = 6.9, corresponding to 240 pc, in close agreement with the somewhat cruder estimate from the secondary star. Using cross-correlation methods described in Thorstensen et al. (2004), we find the companion's heliocentric radial velocity to be −2 ± 10 km s⁻¹.

As Table 3 shows, the parallax is small, but clearly detected, and the distance derived from the parallax, while broadly consistent with the photometric estimate, is a bit shorter. The compromise Bayesian distance estimate is 180(+36, − 28) pc; at 180 pc, the space velocity with respect to the LSR is 38 km s⁻¹, not too large for a disk population object.

Liebert et al. (1982) called attention to PG 1550+191, later named MR Ser, as the only AM Her star discovered in the Palomar-Green ultraviolet survey. Mukai & Charles (1986) detected TiO absorption bands and estimated the spectral type of the secondary as M5-M6. Using the assumptions that the secondary fills its Roche lobe and obeys a zero-age main-sequence mass–radius relation, they estimated d = 142 pc from the secondary; Schwope et al. (1993), using similar methods, estimated d = 139 ± 13 pc independently. As with VV Pup, Araujo-Betancor et al. (2005) detected the white dwarf in HST ultraviolet spectra, and again fitted one- and two-component atmospheres, yielding, respectively, 180(+23, − 30) and 160(+18, − 26) pc. Combining all of these, we adopt 155 ± 25 pc as the prior distance estimate. Once again, the astrometric distance—113(+14, − 12) pc—is somewhat shorter than the prior estimate. This suggests that the mass of the white dwarf is fairly high, resulting in a smaller radius, and, hence, that the solid angle inferred from the atmosphere fits corresponds to a shorter distance. The secondary star's surface brightness may also have been overestimated. In any case, the prior estimate increases the best-guess distance to 126(+14, − 12) pc from its purely astrometric value.

The transverse velocity at the nominal distance is 44 km s⁻¹ referred to the LSR; Schwope et al. (1993) found a mean velocity γ = 7 ± 18 km s⁻¹ for the sodium IR doublet, which tracks the secondary's photosphere and, hence, should give a trustworthy reading of the systemic velocity. Including this gives an LSR space velocity of 49 km s⁻¹.