«HST FINE GUIDANCE SENSOR ASTROMETRIC PARALLAXES FOR THREE DWARF NOVAE: SS AURIGAE, SS CYGNI, AND U GEMINORUM Thomas E. Harrison1, Bernard J. ...»
HST FINE GUIDANCE SENSOR ASTROMETRIC
PARALLAXES FOR THREE DWARF NOVAE: SS AURIGAE,
SS CYGNI, AND U GEMINORUM
Thomas E. Harrison1, Bernard J. McNamara
Astronomy Department, New Mexico State University, Las Cruces, NM 88003
Department of Astronomy, University of Washington, Seattle, WA 98195
Barbara J. McArthur, G. F. Benedict
McDonald Observatory, University of Texas at Austin, Austin, TX 78712 Arnold R. Klemola University of California Observatories/Lick Observatory, University of California, Santa Cruz, California, 95604 and Ronald L. Gilliland Space Telescope Science Institute, 3700 San Martin Drive, Baltimore, MD 21218 Received ; accepted –2– ABSTRACT We report astrometric parallaxes for three well known dwarf novae obtained using the Fine Guidance Sensors on the Hubble Space Telescope. We found a parallax for SS Aurigae of π = 5.00 ± 0.64 mas, for SS Cygni we found π =
6.02 ± 0.46 mas, and for U Geminorum we obtained π = 10.37 ± 0.50 mas.
These represent the ﬁrst true trigonometric parallaxes of any dwarf novae. We brieﬂy compare these results with previous distance estimates. This program demonstrates that with a very modest amount of HST observing time, the Fine Guidance Sensors can deliver parallaxes of unrivaled precision.
Subject headings: Astrometry — novae, cataclysmic variables — stars:
individual (SS Aurigae, U Geminorum, SS Cygni) Based partially on observations obtained with the Apache Point Observatory 3.5-meter telescope, which is owned and operated by the Astrophysical Research Consortium.
1. INTRODUCTION Like many astronomical objects, the distances to cataclysmic variables are imprecisely known. Cataclysmic variables (CVs) are interacting binaries composed of a white dwarf and a main sequence companion. The secondary star ﬁlls its Roche Lobe, and matter is transferred to the white dwarf through the inner Lagrangian point. In non-magnetic systems, the mass transfer occurs through an accretion disk. Most, but not all, CVs exhibit outbursts where the system suddenly brightens (see Warner 1995 for a complete review of the behavior of the diﬀerent subclasses of the CV family). These eruptions range from the luminous classical novae, where the outbursts are due to a thermonuclear runaway on the white dwarf (see Starrﬁeld et al. 1998), and theexplosion releases Etot ∼ 1045 erg over the lifetime of the outburst. To the dwarf novae (DN), where the outburst energy is more modest, Etot ∼ 1040, and is generated by an accretion disk instability cycle (see Cannizzo et al. 1998, and references therein). But our knowledge of these energies and other fundamental parameters of CV systems suﬀers due to the lack of accurate distance measurements.
Berriman (1987) has compiled a list of distance estimates for CVs that used a variety of techniques, including reports of astrometric parallaxes, and spectroscopic parallaxes that relied on the photometric parameters of the CV secondary stars. This latter technique may be the most reliable, but its application is made uncertain by the complex spectral energy distribution of CVs at minimum light, where emission from the hot white dwarf, the accretion disk, the irradiated secondary star, and features that arise from the accretion stream and its impact with the accretion disk (the “hot spot”) contaminate the systemic luminosity. To evaluate the accuracy of the various secondary distance estimators requires direct measurements of the parallaxes of a number of well known CVs. Of the dwarf novae with astrometric parallaxes compiled by Berriman, only that for SS Cyg has a signiﬁcance –4– greater than three sigma: 50 ± 15 pc (Kamper, 1979). We show below that even this measurement is incorrect, and we conclude that until now, no dwarf nova has truly had its trigonometric parallax measured. In this paper we report the ﬁrst high precision parallaxes for three well known dwarf novae: SS Aurigae, SS Cygni and U Geminorum. These parallaxes were obtained using the Fine Guidance Sensors (FGS) on the Hubble Space Telescope (HST).
2. FGS ASTROMETRIC OBSERVATIONS
The FGS were designed to provide exceptional pointing and tracking stability for the science instruments on the HST no matter where the telescope was pointed. As such, the FGS were designed to have a large dynamic range and a large ﬁeld-of-view. For normal guiding operations, only two FGS are employed. This frees the third FGS (“FGS3”) to make astrometric measurements. There are two modes of astrometric operation: “position” and “transfer”. For obtaining parallaxes and proper motions, position mode is used.
For resolving close binaries, transfer mode is used. Astrometry using the FGS has been fully described elsewhere (e.g., Benedict et al. 1994, Benedict et al. 1992, Bradley et al. 1991), however, the most comprehensive source for understanding the nuances of obtaining high precision parallaxes with the FGS is found in the Fine Guidance Sensor Instrument Handbook. Since our observing program followed the prescription outlined within the Handbook, only the most important details relevant for parallax measurement will be addressed here. We present a more complete discussion of the FGS astrometric data analysis for the program CVs, including their proper motions, and comparison of their astrometric and infrared spectroscopic parallaxes, in Harrison et al. (1999).
separated epochs when the parallax factors approach unity. Each FGS has a ﬁeld-of-view (FOV) that consists of a quarter annulus of inner and outer radii of 10 and 14 arcmin, respectively. This entire area, known as a “pickle”, is accessible to the interferometer. Not all of the area of the pickle ends up being accessible in the typical astrometry project because of the diﬀerent roll angles of the HST found at diﬀerent epochs of observation. The large FOV remains one of the greatest strengths of the FGS: It allows astrometry of objects in regions where the density of potential reference stars is low. It is important to note, however, that the astrometric precision is a function of the location of the object within the FOV of the FGS. Objects nearer the center of the pickle are more precisely located than those near an edge. This variation arises from optical distortion across the FOV, known as the Optical Field Angle Distortion (“OFAD”), which is greater, and less-well calibrated, near the edges of the pickle (see Jeﬀreys et al. 1994). A dedicated calibration program, called the Long Term Stability Test (the “LSTAB”, see McArthur et al. 1997), is periodically run throughout each HST cycle to assess the stability of the OFAD. The LSTABs detect any changes in plate scale and are vital for astrometric data reduction.
Another important aspect of the FGS is their dynamic range: Positions of objects with
8.0 V 17.0 can be measured without changing the instrument conﬁguration. This is especially important for the dwarf novae in our program, as at outburst they approach V ∼ 8, but at minimum have V ∼ 14.5. Thus, we were able to ignore the variability of our targets in planning the observational program. The astrometric precision to which a target can be measured is a function of the brightness of the target. For objects with V 15.0, the single measurement precision is ≈ 1 mas, for fainter objects the precision is ≈ 2 mas.
The entire error budget for a minimalist FGS parallax program is ≈ 1 mas. A carefully planned program with multiple observational epochs, like that detailed below, can achieve parallaxes with sub-mas precision.
3. THE OBSERVATIONAL PROGRAM
An ideal FGS parallax program would consist of ﬁve or more reference stars having V 15.0, all within a few arcminutes of each other, and spread uniformly around the target.
Such ﬁelds are not always available for objects of astrophysical interest, and therefore the choice of potential astrometry targets is limited. We based our selection of CVs on four criteria, 1) their astrophysical signiﬁcance, 2) their minimum brightness (V 15.0), 3) the availability of a good set of reference stars, and 4) the likelihood that their parallaxes would have a precision of ≤ 10%. We also conﬁned our selection of CVs to U Gem type DN–objects that have similar outburst characteristics and where the secondary star is clearly visible at minimum light. This latter criterion was imposed to allow us to evaluate the accuracy of the technique of spectroscopic parallax. Clearly, parallaxes for a large sample of CVs would be desirable, but given the scarcity of HST time, a modest program was proposed to conﬁrm that high-precision parallaxes for several such objects could be obtained with a modest amount of observing time.
Three target DN that met the above criteria were SS Aur, U Gem, and SS Cyg. U Gem and SS Cyg are very well known, having been observed continuously for more than 100 yr. SS Aur is probably less well known outside of the CV community, but it is a regularly studied DN (c.f., Shafter and Harkness 1986, Tovmassian 1987, and Cook 1987).
[For further information on these and other DN systems, as well as outstanding problems in CV research, the reader is directed to Warner (1995).] Using the HST Guide Star Catalogue, potential reference stars were identiﬁed that fell within the pickle centered on the target DN. We then selected the brightest, and best positioned to serve as reference stars.
As stated above, ideally ﬁve or more reference stars are desired for an FGS astrometry program. But the actual number of reference stars used must be balanced vs. the time within a single HST orbit available for reference star measurement. The fainter the object, –7– the longer the time required for a position measurement. For U Gem and SS Aur, only four reference stars were used due to the overall faintness of the reference stars and the targets.
For SS Cyg, a brighter target in a well populated reference ﬁeld, ﬁve reference stars were used. The positions, VRI photometry, spectral types, visual extinction, and spectroscopic parallaxes of all thirteen reference stars employed in our program are listed in Table 1.
An observational sequence consists of slewing to the target ﬁeld, acquisition of the target DN, and then repeated measurement of the position of the target and each of the reference stars. As described in the Handbook, the pointing of the HST exhibits a small drift throughout an orbit. To account for this drift, one (or more) of the reference stars is chosen as a “drift check star”. This star is measured more frequently than the other reference stars allowing this drift to be modeled, and removed during data reduction. A typical observing sequence for SS Aur was: SS Aur, Ref. #2 (drift check star), Ref. #12, Ref. #21, Ref. #9, SS Aur, Ref. #2, Ref. #12, Ref. #21, Ref. #9, SS Aur, Ref. #2, Ref. #12, Ref. #21, Ref.
#9, Ref. #2. In the case of SS Aur, each star was observed for 60 seconds. The acquisition of each object has an associated overhead, and the combined exposure times and overheads for the observational sequence described above consumed an entire 54 min HST “orbit”.
Using the Handbook guidelines, we estimated that two such sequences were necessary at each epoch to achieve the program goal of parallaxes with a precision of ≤ 1 mas.
During a single exposure time, a large number of independent samples of the target’s position are collected. For objects with V = 14, the number of samples obtained in a 60 s exposure is ≈ 600. These samples are averaged to determine a single position in FGS coordinates. At the end of a observational sequence, the x and y positions in FGS coordinates for all of the targets are obtained. These positions then go thorough an extensive calibration process that accounts for the OFAD, the pointing drift during the observation, and the diﬀerential velocity aberration caused by the motion of the HST –8– through space. The result is a set of positions for one epoch of observation. To measure a parallax, observations at a minimum of two epochs are necessary. To account for the proper motions of the target and reference stars, observations over as long a timeline as possible are desired. To secure a set of observations approaching those needed for a classical parallax measurement, we obtained data on three epochs. Each observational epoch occurred at the season of the maximum parallax factor for the target DN. Originally, these three epochs were separated by six months, but due to the diﬃculties with NICMOS, and the subsequent changes in HST proposal priorities, our third epoch observations were delayed by an additional six months. Thus, from start to ﬁnish, the observational program spanned two years.
4. PARALLAXES OF THE DWARF NOVAE
After the observations were obtained, and processed, astrometric solutions were sought for each of the targets. As stated earlier, two observational sequences were obtained at each of the three epochs. Thus, six independent sets of measurements were used in the astrometric solution, performed by the Space Telescope Astrometry Team (STAT) at the University of Texas (see Benedict et al. 1994). A master plate was constructed using a six parameter plate solution that is simultaneously solved for translation, rotation, scale, and terms for independent scales on the x and y axes. The solutions were robust for SS Cyg and U Gem, but less so for SS Aur. During the ﬁrst observational epoch for SS Aur, the FGS could not lock on to one of the reference stars (Ref. #8), apparently because it was much fainter than estimated in the Guide Star Catalogue. Subsequently, for epochs two and three, we replaced this reference star with another (Ref. #21). Therefore, the solution for SS Aur was not as well constrained, and the resulting precision was slightly poorer than found for the other two DN (see Harrison et al. 1999).