SPECTROSCOPY FROM THE HUBBLE SPACE TELESCOPE COSMIC ORIGINS SPECTROGRAPH  OF THE SOUTHERN NOVA-LIKE BB DORADUS IN AN INTERMEDIATE STATE

1.1. Accretion Rates and WD Temperatures in CVs

1.2. The VY Scl Nova-like CV BB Doradus

Data from the American Association of Variable Star Observers showed that BB Dor dropped from its usual high state (${m}_{{\rm{v}}}\approx 15$) into a low state in 2013 mid-January, reaching a magnitude of $\sim 19$–19.5, thereby triggering the HST/COS observations of it. However, the luminosity of the system started to increase on January 19, and as a consequence, at the time of the COS observations (2013 February 13, ∼25 days after its deep minimum), BB Dor was found in an intermediate state, with a magnitude of 17.5–18.0. It took another ∼25 days to return to its high state.

The COS instrument was set with the detector in the FUV channel configuration, with the G140L grating centered at a wavelength of (CENWAVE=) 1105 Å, covering the wavelength range 1110–2150 Å, with a resolution of $R\approx 3000$. The photons were collected through the Primary Science Aperture with a diameter of 2.5 arcsec. The COS data consist of a total of 7272 s of good exposure in TIME-TAG mode. We used the four FP-POS positions to improve the signal-to-noise ratio and minimize the effects of flat-field artifacts. The data (LC1VVD010) consisted of five subexposures, because one of the FP-POS positions spread over two consecutive spacecraft orbits. A final spectrum was then constructed by co-adding the five X1D files generated by CALCOS (v3.1; Hodge et al. 2007; Hodge 2011).

The longer wavelengths beyond 2000 Å were found to be very noisy and that portion of the spectrum had to be discarded. Consequently, the final spectrum starts at 1117 Å and ends at 2000 Å. The spectrum is presented in Figure 1.

Zoom In Zoom Out Reset image size

Strong emission lines from low and high ionization species have been identified and are annotated in Figure 1. The wavelength region 1180–1250 Å does not show any discernible hydrogen Lyα profile/wings (from the WD and/or accretion disk) due to the many emission features in that region. In fact, the spectrum is contaminated with strong and broad geocoronal emission of Lyα (1216 Å) and O i (1302 Å). At the longer wavelengths, $\sim 1850$–1900 Å, we identify emission lines from Al iii (1857 and 1862 Å), and there seem to be additional emission features beyond 1900 Å that remain unidentified. Because of this, we ignore the portion of the spectrum in the longer wavelengths ($\lambda \gt 1850\,\mathring{\rm A}$) in our modeling (our spectral modeling does not include emission lines).

Of all the emission lines from the source, the C iv line is the strongest and best resolved. We therefore use it to assess the inclination of the system because this line is believed to be forming in the hottest region in the disk, namely, the very inner disk $r\sim 1$–2R_wd. The C iv (1u) line is made of a doublet with rest wavelengths at 1548.20 and 1550.78 Å. The modeling of the doublet with a double Gaussian gives an FWHM of 6.85 Å (for each Gaussian), and twice as much for the maximum broadening at their base. This translates into a velocity of $\sim 650$ km s⁻¹ for the FWHM and ∼1300 km s⁻¹ for the maximum broadening. Assuming that the maximum broadening is due to the largest Keplerian speed in the disk, obtained at $r\sim {R}_{\mathrm{wd}}$, one obtains an inclination of ∼20° for the disk/binary for an average CV WD mass of $0.8\,{M}_{\odot }$with a maximum Keplerian speed of ∼3900 km s⁻¹. Assuming a larger radius for the region where the C iv line forms, $r\approx 2{R}_{\mathrm{wd}}$, gives an inclination of ∼30°.

There are also some absorption lines, mostly at short wavelengths. All the absorption lines are much narrower than the emission lines. We present the lines in Figure 2. We tentatively identify the following lines: Si iv (1122.5 and 1128.3 Å), Si iii (1140.6, ∼1142, and ∼1145 Å), C iii (∼1175 Å), S iii (1190.2, 1194.2, and 1201 Å), Si iii (1206.5 Å), Si ii (1260 Å), C ii (1334.5 and 1335.7 Å), Si iii (∼1500 Å), and Si ii (1526 Å). The nitrogen line N i (1134 Å) is most probably terrestrial in origin and has often been observed in FUSE spectra of CVs. Since the absorption lines might potentially originate in the WD photosphere or the accretion disk (when seen at low inclination), we further discuss them in Sections 4 and 5.

Zoom In Zoom Out Reset image size

A comparison of the continuum flux level of the COS spectrum to that of the archival FUSE spectrum (taken on 2007 July 9) shows that the COS continuum flux level is lower than the FUSE continuum flux level by a factor of 20. Namely, in the overlap region $\sim 1110$–1185 Å the FUSE continuum reaches $\sim 2\times {10}^{-13}$ erg s⁻¹cm⁻² Å⁻¹, while the COS continuum reaches $\sim 1\times {10}^{-14}$ erg s⁻¹cm⁻² Å⁻¹. Though the system was not caught in a deep low state, this fortunate occurrence gives us the unique opportunity to study possible photospheric emission from the WD, because the mass accretion rate must have dropped substantially, thereby possibly exposing the WD.

In order to assess the contribution of the WD to the total flux, we use the TIME-TAG data to generate three light curves. A first light curve is generated by integrating all the data over the entire wavelength coverage except for the known airglow emission lines of Lyα, O i, and N i; a second light curve is made with the C iv line (1535–1564 Å); and a third light curve is made using the continuum in the range 1424–1523 Å. In Figure 3 we show the three light curves, with each normalized to one. The continuum varies in the same way as the C iv emission, which we know does not come from the WD. Therefore, if the continuum is the sum of the disk and WD, the fact that the variability (∼20% of the total flux) is produced by the disk indicates that the WD cannot contribute more than ∼80% of the flux.

Zoom In Zoom Out Reset image size

In our analysis we model the COS spectrum of BB Dor with WD models, disk models, and composite models that include the contribution from both the WD photosphere and an accretion disk. Our main goal is to derive the WD effective surface temperature and the mass accretion rate of the system, to add one data point to the handful of NL systems with known ${T}_{\mathrm{eff}}$. Our spectral analysis technique is described in the next section.

In order to derive the mass accretion rate $\dot{M}$ and the WD temperature ${T}_{\mathrm{eff}}$, we carry out a quantitative comparison (a "fit") of the observed spectrum to theoretical disk and WD spectra for a wide range of parameters.

The synthetic model spectra of the WD stellar atmosphere are generated using the FORTRAN suite of codes TLUSTY, SYNSPEC, and ROTIN (Hubeny 1988; Hubeny & Lanz 1995). The basic input parameters are the stellar effective surface temperature, surface gravity, and chemical composition. In a first step, a converging stellar atmospheric structure is computed by successive runs of TLUSTY. The TLUSTY model atmospheres have hydrogen and helium taken explicitly, i.e., their selected levels are treated in non-LTE, and the bound–bound and bound–free transitions between these levels determine the total opacity and emissivity. Carbon, nitrogen, and oxygen are treated implicitly, i.e., they contribute only to the particle and charge conservation, assuming LTE. Next, a run of the spectral synthesis code SYNSPEC, using the output structure of the model atmosphere from TLUSTY as input is done. In this step, all opacity sources, i.e., the lines of all other elements, are considered (unless it is specifically required not to). The short code, ROTIN, is then used to account for the projected stellar rotational velocity (and/or instrumental) broadening, including the effect of limb darkening.

In this manner, we generate WD stellar photospheric models with effective temperatures ranging from ∼15,000 to ∼60,000 K in increments of 500–1000 K. The NLTE treatment is turned on for the hot WD models (T_eff > 35,000 K), using the approximate NLTE treatment of lines option in SYNSPEC (Hubeny & Lanz 1995). We chose here a value of $\mathrm{log}(g)=8.4$, which matches a WD mass of $0.80\,{M}_{\odot }$, a rounded value of the average WD mass in CVs of $0.83\pm 0.23\,{M}_{\odot }$ (Zorotovic et al. 2011). Since the spectrum has emission lines and thus a disk component, we use only solar composition (set up in SYNSPEC) because it would be impossible to derive chemical abundances for such a spectrum. We vary the stellar rotational velocity ${V}_{\mathrm{rot}}\sin (i)$ from 100 to 500 km s⁻¹ in steps of $50$ km s⁻¹. The WD rotation rate (${V}_{\mathrm{rot}}\sin (i)$) is determined by fitting the WD model to the spectrum while paying careful attention to the line profiles. The mass–radius relation from Wood (1990), for carbon–oxygen and non-zero-temperature WDs, is used to obtain the radius of the WD.

The accretion disk spectra were generated assuming that the disk is made of a collection of annuli, where each annulus has a temperature T(r) and gravity $\mathrm{log}(g(r))$ given by the standard disk model (Shakura & Sunyaev 1973; Pringle 1981), for a given central mass ${M}_{\mathrm{wd}}$ and accretion rate $\dot{M}$. For each annulus, an atmosphere model was generated using TLUSTY, followed by a run of SYNSPEC to obtain a spectrum of the  annulus. The contributions of all the annuli were combined using the code DISKSYN, and a final disk spectrum was obtained for any given inclination angle (we assume the disk to lie in the orbital plane of the binary). In the present work we use the public grid of disk models from Wade & Hubeny (1998), in which a detailed explanation of the entire numerical procedure is given.

In our analysis, we use a grid of models of synthetic spectra of WDs and accretion disks covering a wide range of values of the WD temperature ${T}_{\mathrm{eff}}$, gravity $\mathrm{log}(g)$, projected rotational velocity ${V}_{\mathrm{rot}}\sin (i)$, inclination i, mass accretion rate $\dot{M}$, and abundances. The grid of disk models has the following inclination angles: $i=18^\circ$, 41^◦, 60^◦, 75^◦, and 81^◦. Since the inclination of the system is low (Chen et al. 2001; Schmidtobreick et al. 2012), and based on our assessment in the previous section, we use $i=18^\circ$, the lowest value in the grid of models. Furthermore, as the flux level in the COS spectrum is 20 times lower than in the FUSE spectrum, we expect our COS spectral analysis to reveal a mass accretion rate lower than that derived from the FUSE spectrum, namely ${\dot{M}}_{\mathrm{COS}}\lt {10}^{-9}\,{M}_{\odot }\,{\mathrm{yr}}^{-1}$.

The distance to the system is not known accurately, but the results of the FUSE analysis implied a distance $d\lt 1\,\mathrm{kpc}$ (Godon et al. 2008). Rodríguez-Gil et al. (2012) estimated a distance of d ∼ 1–2 kpc based on a WD temperature of 30,000 K and a secondary spectral type of M3–M4. The spectrum exhibits few absorption lines, and, as stated earlier, we decided to use solar abundances in our modeling. With our assumption of a standard CV WD mass of $0.8\,{M}_{\odot }$, we keep $\mathrm{log}(g)$ constant at 8.4. Therefore, the parameters that we vary in our model fits are the temperature of the WD, its rotation rate, and the mass accretion rate of the disk.

Before carrying out a fit of the spectrum we mask portions of the spectrum with emission lines, since our synthetic spectral code does not generate emission lines.

After having generated grids of models for the COS spectrum of BB Dor, we use FIT (Press et al. 1992), a ${\chi }^{2}$ minimization routine, to compute the reduced ${\chi }_{\nu }^{2}$ (${\chi }^{2}$ per number of degrees of freedom ν) and scale factor (which gives the distance) for each model fit. While we use a ${\chi }^{2}$ minimization technique, we do not blindly select the models with least ${\chi }^{2}$, but we examine the models that best fit some of the spectral features and that are most consistent with the values of the system parameters.

4.1. Single WD Modeling

Despite the facts that the hydrogen Lyα absorption feature is not detected in the spectrum (Pala et al. 2016), and that the WD cannot contribute more than ∼80% of the flux, we decide to first model the spectrum with a single WD alone. Such a modeling will confirm that a WD alone is not enough to account for the observed flux and will help us identify the lines forming in the WD photosphere (if any).

We first try to fit the slope of the continuum, and we find that it agrees well with a 23,000 K WD (our model #1, see Table 1), but the Lyα region of the model forms a large broad absorption feature not seen in the observed spectrum. The distance obtained for this model is also far too short, only 291 pc, and ${\chi }_{\nu }^{2}=3.07$.

Table 1.  Results of Synthetic Spectral Modeling

Model	Twd	$\mathrm{Log}(\dot{M})$	WD/Disk	d	${\chi }_{\nu }^{2}$	Figure
Number	(103 K)	(M⊙ yr−1)		(pc)
1	23.0	⋯	WD	291	3.07	⋯
2	27.0	⋯	WD	412	1.98	⋯
3	30.0	⋯	WD	515	1.92	4,5
4	35.0	⋯	WD	692	1.94	6
5	40.0	⋯	WD	864	1.98	⋯
6	50.0	⋯	WD	1052	2.17	⋯
7	⋯	−8.0	disk	6808	1.36	⋯
8	⋯	−8.5	disk	4288	1.32	7
9	⋯	−9.0	disk	2545	1.36	⋯
10	⋯	−9.5	disk	1356	1.87	8
11	⋯	−10.0	disk	679	4.66	⋯
12	⋯	−10.5	disk	322	12.14	⋯
13	30.0	−8.0	0.6/99.4	6828	1.36	⋯
14	50.0	−8.0	2.4/97.6	6890	1.37	⋯
15	30.0	−8.5	1.5/98.5	4319	1.32	⋯
16	50.0	−8.5	6/94	4415	1.33	⋯
17	30.0	−9.0	4/96	2598	1.35	⋯
18	50.0	−9.0	15/85	2754	1.34	⋯
19	25.0	−9.5	6/94	1400	1.85	⋯
20	35.0	−9.5	21/79	1524	1.49	9
21	48.0	−9.5	37/63	1693	1.31	⋯
22	60.0	−9.5	46/54	1817	1.31	⋯
23	35.0	−10.0	53/47	967	1.47	⋯
24	40.0	−10.0	65/35	1094	1.21	10
25	45.0	−10.0	70/30	1175	1.19	⋯
26	50.0	−10.0	74/26	1244	1.23	⋯
27	27.0	−10.5	66/34	516	2.79	⋯
28	30.0	−10.5	76/24	602	1.68	⋯
29	35.0	−10.5	86/14	761	1.23	11
30	40.0	−10.5	91/9	920	1.32	⋯

Download table as:  ASCIITypeset image

As we increase the temperature, we find a better agreement in the Lyα region, but the model becomes too blue as its slope becomes steeper. The distance also increases, and it reaches ∼1 kpc for a 50,000 K WD (model #6), with ${\chi }_{\nu }^{2}=2.17$.

The model with least ${\chi }^{2}$ is obtained for a 30,000 K WD (model #3), with a distance of 515 pc. This model is shown in Figure 4. Even for this model the Lyα region is also too broad and too deep and the continuum is too steep. It is likely that the Lyα region (1170–1250 Å) is entirely dominated by broad emission lines. We note that setting the stellar rotational velocity to 150 km s⁻¹ in this model gives a reasonable fit to the Si ii 1260 Å line. The short-wavelength region is shown in detail with line identification in Figure 5. However, the Si ii 1260 Å line is the only line that can be matched with this WD model. The expected Si ii 1251 and 1265 Å lines, as seen in the theoretical spectrum, are not observed. The C iii (1175 Å) line is contaminated with broad emission, but its absorption feature on top of the emission is too shallow compared to the theoretical spectrum.

Zoom In Zoom Out Reset image size

Zoom In Zoom Out Reset image size

We find that we can fit some set of lines by changing the WD temperature, but each temperature provides a fit to only a few lines. For example, in Figure 6, we set the WD temperature to T = 35,000 K, with a projected rotational velocity of 300 km s⁻¹, and obtain a relatively good fit to the Si iv (1122.5 and 1128.3 Å) and Si iii (1140.6, 1141.6, 1142.3, and ∼1145 Å) lines. However, the Si ii (1260 Å) line is not reproduced in this model.

Zoom In Zoom Out Reset image size

Similarly, a WD temperature of T = 40,000 K with a projected rotational velocity of 150 km s⁻¹ is needed to fit the C ii 1324 and ∼1335 Å absorption lines; however, none of the other lines are fitted.

These are clear indications that the observed lines do not form in the stellar WD photosphere. Overall, the single WD model does not provide a satisfactory fit. The results for the fits to the single WD model are recapitulated in the first part of Table 1 (models #1–6).

4.2. Single Disk Modeling

As the absorption lines cannot satisfactorily be fitted with single WD models, we first check whether the disk alone can provide such a fit. We find that, in order to fit the narrow absorption lines in the COS spectrum, the inclination of the system would have to be extremely small ($\sim 3^\circ$). That is, the disk would have to be almost face-on to minimize the broadening of the lines from the Keplerian velocity in the disk. Since there is no indication that this is the case, we deduce that the lines do not form in the disk either. The width of the C iv emission line implies an inclination of $\sim 20^\circ$; we therefore set the inclination to $i=18^\circ$ (the smallest value in our grid of disk models). We vary the mass accretion rate from $\dot{M}={10}^{-8.0}\,{M}_{\odot }\,{\mathrm{yr}}^{-1}$ down to $\dot{M}={10}^{-10.5}\,{M}_{\odot }\,{\mathrm{yr}}^{-1}$ in logarithmic steps of 0.5 (models #7–12 in Table 1, second part). Here, we make no further attempt to fit the absorption lines, but rather we concentrate on the continuum.

The best fit is for $\dot{M}={10}^{-8.5}\,{M}_{\odot }\,{\mathrm{yr}}^{-1}$ (model #8), with ${\chi }_{\nu }^{2}=1.32$, but it gives an unacceptably large distance (more than 4 kpc) and the spectral features at the shorter wavelengths are not reproduced by the disk model. This model is presented in Figure 7.

Zoom In Zoom Out Reset image size

The slope of the observed spectrum agrees better with a disk model with $\dot{M}={10}^{-9.5}\,{M}_{\odot }\,{\mathrm{yr}}^{-1}$ (#10, Figure 8), which has a distance (1356 pc) in better agreement with the ≈1–2 kpc quoted above. However, the ${\chi }_{\nu }^{2}$ value (1.87) of this model is significantly larger than for the model with $\dot{M}={10}^{-8.5}\,{M}_{\odot }\,{\mathrm{yr}}^{-1}$. In the vicinity of the Lyα region (around 1170–1250 Å) the fit is rather poor.

Zoom In Zoom Out Reset image size

Overall, the single disk model provides only a marginal improvement over the single WD model. However, due to the limits on the distance to the system, the mass accretion rate in the single disk model is constrained to be in the range 10^−10.5 M_⊙ yr⁻¹ < $\dot{M}$ < 10⁻⁹ M_⊙ yr⁻¹.

4.3. Combined WD + Disk Modeling

Since neither the single WD nor the single disk models fit the spectrum adequately, we next attempt models with both components. We run each single disk model in combination with a WD with a temperature varying from 20,000 to 60,000 K, in steps of 1000 K, a total of almost 200 model fits. We present here only the best-fit models.

Since the disk models with a large mass accretion rate gave an unacceptably large distance, we present here only the disk + WD models with lower mass accretion rate. For completeness we list some disk + WD models with large mass accretion rate in Table 1 (models #13–18).

As the mass accretion rate is set to $\dot{M}={10}^{-9.5}\,{M}_{\odot }$ yr⁻¹ the disk model is improved with the addition of a hot WD. In Figure 9 we present such a model with ${T}_{\mathrm{wd}}={\rm{35,000}}$ K, giving a distance of 1.5 kpc, and where the WD contributes ≈1/5 of the FUV flux and the disk contributes the remaining 4/5. The projected rotational velocity of the WD is ${V}_{\mathrm{rot}}\sin i=200$ km s⁻¹. This model (# 20 in Table 1) has ${\chi }^{2}=1.49$. As the WD temperature is further increased, to ${T}_{\mathrm{wd}}={\rm{48,000}}$ K (model #21) and above (60,000 K, model #22), the model provides a small improvement with ${\chi }_{\nu }^{2}\approx 1.3$.

Zoom In Zoom Out Reset image size

The fit can be further improved to ${\chi }_{\nu }^{2}\approx 1.2$ by decreasing the mass accretion rate further: $\dot{M}={10}^{-10}\,{M}_{\odot }\,{\mathrm{yr}}^{-1}$ with T_wd = 40,000–50,000 K. Models #24–26 are all very similar and we present model #24 in Figure 10. These models give a distance of the order of 1–1.2 kpc.

Zoom In Zoom Out Reset image size

As we further decrease $\dot{M}$ to $\dot{M}={10}^{-10.5}\,{M}_{\odot }\,{\mathrm{yr}}^{-1}$ ($\approx 3\times {10}^{-11}\,{M}_{\odot }$ yr⁻¹), the lowest ${\chi }_{\nu }^{2}$ (1.23) for that mass accretion rate is obtained with ${T}_{\mathrm{wd}}={\rm{35,000}}$ K—model #29. This model is presented in Figure 11 and gives a distance of 761 pc. However, this model has a WD contributing more than 80% of the COS flux and therefore it has to be rejected. Other models with $\dot{M}={10}^{-10.5}$ M_⊙ yr⁻¹ have either a distance that is too short (when ${T}_{\mathrm{wd}}$ is decreased) or a WD contributing more than 80% of the flux (when ${T}_{\mathrm{wd}}$ is increased).

Zoom In Zoom Out Reset image size

The best fits obtained for the combined WD + disk components have a mass accretion rate $\dot{M}={10}^{-9.5}\mbox{--}{10}^{-10}\,{M}_{\odot }\,{\mathrm{yr}}^{-1}$ and WD temperature ${T}_{\mathrm{wd}}\sim {\rm{35,000}}$ K or larger.

We found that the absorption lines in the COS spectrum of BB Dor do not match the WD photosphere lines, and cannot originate in the disk. Absorption lines were also observed in the FUSE spectrum of BB Dor taken during its high state, when the disk dominated the FUV. Because of that the FUSE absorption lines could not originate in the WD photosphere, and, because of their width, they could originate in the disk only if $i\lt 10^\circ$ (Godon et al. 2008). Since the inclination is probably larger, i.e., $\sim 20^\circ$, we conclude that the absorption lines observed in the FUV spectra of BB Dor during high and intermediate states must come either from material above the disk and WD, or from circumbinary material, or from the interstellar medium (ISM).

The ISM lines in the FUSE spectra of CVs are usually very narrow with a broadening $\ll 1\,\mathring{\rm A}$, consisting of low ionization species such as Fe ii, S i, C ii, Si ii, and Ar i (Godon et al. 2006, 2012), while the absorption features in the COS spectrum of BB Dor have a larger velocity broadening $\gt 1\,\mathring{\rm A}$. Nonetheless, broader ISM lines have been observed in the FUV spectra of some CVs (Maucher et al. 1988), which enable us to tentatively identify the following lines as being interstellar in origin: Si ii (1190.42, 1193.29 Å), N i (1199.55, 1200.22, 1200.71 Å), Si iii (1206.51 Å), S ii (1250.59, 1253.81, 1259.52 Å), Si ii (1260.42 Å), and C ii (1334.53, 1335.70 Å).

No hydrogen Lyα absorption profile was discernible due to the broad emission in the spectral region 1170–1250 Å. Other regions of the spectrum also exhibit some broad and strong emission lines. As a consequence, and also due to the unknown mass and uncertain distance estimate, the modeling of the COS spectrum of BB Dor proved to be challenging.

In our modeling we assumed an inclination $i=18^\circ$, a standard CV WD mass of $0.8\,{M}_{\odot }$, solar abundances, and an upper limit of 2 kpc for the distance (Rodríguez-Gil et al. 2012). The combined WD plus accretion disk models with a disk with moderately low mass accretion rate and a hot WD appear to be the best fits as far as the lowest ${\chi }_{\nu }^{2}$ value is concerned.

These results have to be further considered while taking into account that the system was in an intermediate state with a flux 20 times lower than during the FUSE observation, when the system was in its high state (Godon et al. 2008). The FUSE spectrum of BB Dor had a continuum flux level decreasing below 1000 Å, implying that the mass accretion rate had to be $\approx {10}^{-9}\,{M}_{\odot }\,{\mathrm{yr}}^{-1}$ at the time of the FUSE observation, because models with larger mass accretion rate ($\dot{M}\gt {10}^{-9}{M}_{\odot }\,{\mathrm{yr}}^{-1}$) were hotter and contributed more flux at the shorter wavelengths ($\lt 1000\,\mathring{\rm A}$) of FUSE. The FUSE spectral fit also gave a distance estimate of $\approx 700$ pc. At the time of the COS observation, BB Dor must have had a lower mass accretion rate, namely $\dot{M}\leqslant {10}^{-9.5}\,{M}_{\odot }\,{\mathrm{yr}}^{-1}$ (i.e., three times lower than during the FUSE observations), and more likely $\dot{M}\approx {10}^{-10}\,{M}_{\odot }\,{\mathrm{yr}}^{-1}$ (i.e., 10 times lower than during the FUSE observations).

As the WD cannot contribute more than a maximum of ∼80% of the continuum, which would be achieved in the extreme case where the disk luminosity itself varies by 100%, we reject models #29 and #30, and due to the distance constraint models #27 and #28 also have to be rejected. In other words, during the COS observations, the mass accretion rate must have been of the order of $\dot{M}={10}^{-10}\,{M}_{\odot }\,{\mathrm{yr}}^{-1}$ or larger.

We therefore conclude that BB Dor, during its COS observation, most probably had a mass accretion rate $\dot{M}\sim {10}^{-10}\,{M}_{\odot }\,{\mathrm{yr}}^{-1}$ and a WD temperature of ∼35,000 to ∼50,000 K, as presented in Figure 10. This is typical of VY Scl NL systems, and is higher than expected for CVs just above the period gap. We note, however, that the distance obtained from the fit changes dramatically in the solutions, while its estimate is poorly constrained (from ∼700 pc to about 2 kpc). Anticipated Gaia data will soon provide distances to many CVs, and should help resolve the best solution for BB Dor and many other CVs. A short distance ($\lt 1$ kpc) would imply a low mass accretion rate ($\lt {10}^{-10}{M}_{\odot }\,$yr⁻¹) with a WD temperature of ∼35,000 K, while a large distance (∼2 kpc) would imply larger mass accretion rate $(\gt {10}^{-10}\,{M}_{\odot }\,$ yr⁻¹) and WD temperature ($T\sim {\rm{50,000}}$ K).

Only a few VY Scl NL systems have been observed in an intermediate state, and the modeling of their FUV spectra has shown mixed results. AC Cnc seems to reveal a disk and WD, though the mass accretion rate and WD temperature depend strongly on the unknown distance (Bisol et al. 2012). BZ Cam has an FUV spectrum consistent with a 12,500 K blackbody (Prinja et al. 2000), but it is a unique peculiar object exhibiting an emission bow-shock nebula (Ellis et al. 1984). MV Lyr was one of the few VY Scl systems that were modeled in a low state (Hoard et al. 2004; Godon et al. 2012), but the spectrum of its intermediate state could not be modeled with a standard disk model and part of the disk had to be set to a low isothermal temperature (Linnell et al. 2005). However, these mixed results are characteristic of NL systems in high state as well. While some NLs have been successfully modeled with standard disk spectra (e.g., IX Vel, V3885 Sgr, V794 Aql, RZ Gru; Godon et al. 2007; Linnell et al. 2007, 2009; Bisol et al. 2012) others appear impossible to model with standard disk spectra (e.g., RW Sex, V751 Cyg, V380 Oph, UX UMa, QU Car; Linnell et al. 2008a, 2008b, 2010; Zellem et al. 2009). The results of the statistical study of Puebla et al. (2007) suggest that the temperature profile of the standard disk model be revised at least in the innermost parts of the disk in order to properly model some of the disk-dominated CV systems. With, so far, no existing alternative disk model to the standard one, the current results for BB Dor are at least consistent with typical WD temperatures in VY Scl NLs, and improved disk models, as well as accurate distances obtained by Gaia in the near future, will provide better constraints on the physical parameters of this system.

We would like to thank Knox Long and Christian Knigge for reading and commenting on an early version of this manuscript. P.G. wishes to thank Bill Blair for his kind hospitality at the Rowland Department of Physics and Astronomy in the Johns Hopkins University, Baltimore, Maryland. Special thanks to the Austral Variable Star Observer Network (AVSON), and the American Association of Variable Star Observers (AAVSO) for their monitoring of BB Dor. This research is part of the GO program 12870 based on observations made with the NASA/ESA Hubble Space Telescope, obtained at and funded by the Space Telescope Science Institute, which is operated by the Association of Universities for Research in Astronomy Inc., under NASA contract NAS 5-26555. The research leading to these results has received funding from the European Research Council under the European Union's Seventh Framework Programme (FP/2007–2013)/ERC Grant Agreement n.320964 (WDTracer). D.d.M. acknowledges financial support from ASI INAF contract I/037/12/0.