1: \documentclass[12pt,preprint]{aastex}
2: \begin{document}
3: \title{{\it GALEX}, Optical and IR Light Curves of MQ Dra: UV Excesses at Low Accretion Rates}
4:
5: \author{Paula Szkody\altaffilmark{1},
6: Albert P. Linnell\altaffilmark{1},
7: Ryan K. Campbell\altaffilmark{2},
8: Richard M. Plotkin\altaffilmark{1},
9: Thomas E. Harrison\altaffilmark{2},
10: Jon Holtzman\altaffilmark{2},
11: Mark Seibert\altaffilmark{3},
12: Steve B. Howell\altaffilmark{4}}
13: \altaffiltext{1}{Department of Astronomy, University of Washington,
14: Box 351580,
15: Seattle, WA 98195, szkody@astro.washington.edu,plotkin@astro.washington.edu,linnell@astro.washington.edu}
16: \altaffiltext{2}{Department of Astronomy, New Mexico State University, Box 30001, MSC 4500, Las Cruces, NM 88003, cryan@nmsu.edu,tharriso@nmsu.edu,holtz@nmsu.edu}
17: \altaffiltext{3}{Observatories Carnegie Institute of Washington, 813 Santa Barbara St., Pasadena CA 91101, mseibert@ociw.edu}
18: \altaffiltext{4}{National Optical Astronomy Observatories,
19: 950 N. Cherry Avenue, Tucson, AZ 85726, howell@noao.edu}
20:
21: \begin{abstract}
22: Ultraviolet light curves constructed from NUV and FUV detectors on {\it GALEX}
23: reveal large amplitude variations during the orbital period of the Low Accretion Rate
24: Polar MQ Dra (SDSSJ1553+55). This unexpected variation from a UV source is similar
25: to that seen and discussed
26: in the Polar EF Eri during its low state of accretion, even
27: though the accretion rate in MQ Dra is an order of magnitude lower than
28: even the low state of EF Eri.
29: The similarity in phasing of the UV and optical light curves in MQ Dra
30: imply a similar
31: location for the source of light.
32: We explore the possibilities of hot spots and cyclotron emission
33: with simple models fit to
34: the UV, optical and IR light curves of MQ Dra.
35: To match the {\it GALEX} light curves with a single temperature circular
36: hot spot requires different sizes of spots for the NUV and FUV,
37: while a cyclotron model that can produce the optical harmonics with
38: a magnetic field near 60 MG
39: requires multipoles with fields $>$ 200 MG to match the UV fluxes.
40:
41: \end{abstract}
42:
43: \keywords{binaries: close ---
44: novae, cataclysmic variables --- stars: individual (MQ Dra, EF
45: Eri) --- ultraviolet:stars --- white dwarfs}
46:
47: \section{Introduction}
48:
49: The cataclysmic variable MQ Dra was
50: discovered in the Sloan Digital Sky Survey as SDSSJ155331.12+551614.5 (Szkody
51: et al. 2003; S03). Its peculiar optical spectrum showed narrow, highly polarized
52: cyclotron emission features that could be explained with an origin from
53: a white dwarf with
54: a magnetic field near 60 MG. These
55: features underwent large periodic changes during time-resolved
56: photometric and spectroscopic
57: observations, providing an orbital period of 4.39 hrs. Aside from the
58: cyclotron features, the red portion of the spectrum revealed TiO bands, indicating
59: an M4-M5V secondary star whose brightness
60: provides a system distance of 130-180 pc (S03, Schmidt et al. 2005; S05,
61: Harrison et al. 2005).
62: Modeling the continuum after subtraction of the secondary indicated
63: a cool white dwarf $<$10,000K, but the flux at the blue end of the spectrum
64: was larger than from a white dwarf alone, suggesting an additional hot spot (S05).
65: X-ray observations (Szkody et al. 2004) also hinted at an upturn
66: at soft energies, possibly indicating a hot component. Both the appearance of
67: the cyclotron features and the X-ray fluxes were consistent with a cool plasma
68: (kT$\sim$1 keV) and extremely low accretion rate (10$^{-14}$M$_{\odot}$ yr$^{-1}$).
69: The exceptionally low accretion rates in these types of systems resulted in the
70: identification of a new category of magnetic cataclysmic variables
71: called Low Accretion Rate Polars (LARPs; Schwope et al. 2002). The fact
72: that MQ Dra has a relatively long orbital period and a very cool white dwarf
73: led to the speculation that this could be a system which is a pre-Polar i.e. the
74: white dwarf has never been heated by any active accretion and the cyclotron
75: radiation is provided only by wind accretion from the secondary. In contrast,
76: a Polar has known states of active accretion via a mass transfer stream
77: from the secondary.
78:
79: This interpretation is complicated by the fact that most Polars spend much of their
80: lives in low states of accretion (Ramsay et al. 2004; Araujo-Betancar et al. 2005)
81: that can reach low accretion rate values near those for LARPs.
82: Active accretion episodes could produce heated areas that remain hot for long
83: periods of time after the accretion has stopped. This is borne out by UV studies
84: of systems during low states (Araujo-Betancar et al. 2005; G\"ansicke et al. 2006)
85: where 30,000-70,000K hot spots on the white dwarfs during low states were needed
86: to match the observed fluxes. A {\it GALEX} study of the Polar EF Eri four years after
87: it entered a low state (Szkody et al. 2006; S06)
88: showed a 20,000K spot was needed on a 9500K white dwarf to
89: come close to producing the large amplitude modulations visible in the UV light curves.
90: However, simple black bodies for the white dwarf and spots could not reproduce both
91: the amplitude of the modulations and the spectral energy distributions.
92: Schwope et al. (2007; Sw07) had better success with the spectral energy distribution
93: using actual white dwarf models
94: but they did not show UV light curves with which to compare UV observations. Since the
95: white dwarfs in known active Polars observed during low states are generally hotter
96: than those found in LARPS, the influence of the active accretion episodes
97: is not clear.
98: To obtain further insight into the heating of the white dwarf in LARPS, we undertook
99: {\it GALEX} observations of MQ Dra, which has a cooler white dwarf with a higher
100: magnetic field compared to the white dwarf in EF Eri, and an accretion rate
101: that is even lower than EF Eri during its lengthy low state.
102:
103: \section{Observations}
104:
105: The {\it GALEX} satellite obtains images with a FUV detector (1350-1750\AA) and a NUV
106: detector (1750-2800\AA) (Martin et al. 2005). MQ Dra was observed during two
107: intervals, one in June 2005 and the other in July 2006. The first set of observations
108: was done when the FUV detector was not operational, while both detectors were
109: used in 2006. The times of observation are listed in Table 1. Calibrated images
110: in 240s intervals were generated from the observations, phased according to the
111: ephemeris below and then magnitudes inside a 9 pixel radius
112: aperture were measured with
113: the IRAF\footnote{IRAF (Image
114: Reduction and Analysis
115: Facility) is distributed by the National Optical Astronomy Observatories, which
116: are operated by AURA,
117: Inc., under cooperative agreement with the National Science Foundation.} routine
118: {\it qphot}. An annulus of 3 pixels around the aperture was used to obtain and
119: subtract a background. The magnitudes were converted to flux from the values given
120: in the {\it GALEX} online documentation\footnote{See http://galexgi.gsfc.nasa.gov/tools/index.html} (FUV m$_{0}$=18.82=1.40$\times$10$^{-15}$ergs cm$^{-2}$ s$^{-1}$ \AA$^{-1}$
121: and NUV m$_{0}$=20.08 =2.06$\times$10$^{-16}$ ergs cm$^{-2}$ s$^{-1}$ \AA$^{-1}$).
122:
123: Since the previous available ephemeris determined from 5 nights of observations
124: (S03) was
125: not sufficient to phase the {\it GALEX} data to a cyclotron or geometric phasing,
126: we obtained further optical data during 22 nights in the interval from 2007 Feb 25 - May 4.
127: These data were obtained on the NMSU 1m telescope at Apache Point Observatory,
128: using a CCD and $BVRI$ filters with integration times of 90, 60, 60, 60s
129: respectively. Differential light curves were constructed using
130: nearby stars on the same frames and calibrated with the magnitudes for
131: the comparison stars that were used in S03.
132:
133: As shown by the phase resolved spectra in S03 and S05, the optical light curves
134: of MQ Dra are dominated by the prominent cyclotron peaks at 4600, 6200 and
135: 9200\AA, which come into the $BVRI$ bandpasses. Since these harmonics
136: change amplitude due to the changing viewing angle of the cyclotron region
137: during the orbit, a distinct sinusoidal variation is produced (S03).
138: Although this is a broad feature, we could combine all the nights with good
139: $R$ filter coverage (16 nights)
140: to
141: obtain a better period using the IRAF routine {\it pdm}. The past spectroscopic
142: and polarimetric data of S05 showed that the blue to red crossing
143: of the H$\alpha$ emission line was
144: close to
145: the peak amplitude of the cyclotron harmonics. Since the H$\alpha$ emission
146: had the same phasing and amplitude as the TiO bandheads, it allows a definition
147: of phase 0 as inferior
148: conjunction of the secondary (and the peak of the $R$ band light curve). Thus,
149: we obtained the following ephemeris to phase the {\it GALEX} data:
150:
151: phase 0 = HJD 2,454,156.9138 + 0.182985$\pm$0.000005E
152:
153: Since the NUV light curves from 2005 and 2006 were similar, this phasing
154: allowed us to combine the 2005 and 2006 NUV datasets to maximize the
155: S/N.
156:
157: Infrared photometry of MQ Dra was also obtained on the KPNO 2.1 m telescope
158: with the Simultaneous Quad
159: Infrared Imaging Device
160: (SQIID)\footnote{See http//www.noao.edu/kpno/sqiid/sqiidmanual.html},
161: which obtains $JHK$ images simultaneously.
162: Data were obtained on two consecutive nights, 2007 May 28 and 29. On 28 May,
163: the first exposure began at 03:42:01 UT and data were obtained over the
164: next 3.3 hrs, while on May 29 the observations started at 03:31:57 and
165: continued for the next 2.7 hrs. The images were reduced with standard
166: procedures, then
167: differential photometry was performed with respect to several field stars.
168: The differential magnitudes were
169: then corrected to absolute $JHK$ photometry using data from the 2MASS Point
170: Source Catalog, and the data were phased with the above
171: ephemeris.
172:
173: \section{MQ Dra Light Curves}
174:
175: The {\it GALEX} NUV and FUV light curves are shown in Figure 1, along with the
176: phased $BVRI$ light curves from 2007. All the light curves have large
177: variability, with peak-to-peak amplitudes of 0.7,0.8,0.5,1.0,0.9 and 0.2
178: mags for FUV, NUV, $B,V,R,I$ respectively.
179: The $JHK$ light curves are shown in
180: Figure 2. The $K$-band light curve of MQ Dra reveals classic
181: ellipsoidal modulations with one minimum being slightly deeper than the other.
182: This difference can be explained if the secondary star fills (or almost
183: fills) its Roche lobe and is distorted so that the white dwarf-facing hemisphere
184: (viewed at phase $\phi$ = 0.5) is further from the center of the secondary star and therefore
185: cooler. The $H$-band
186: light also appears to be dominated by ellipsoidal variations, but a
187: non-sinusoidal light source, presumably cyclotron emission, appears to
188: contaminate the light curve. The $J$-band light curve is even more complex, and
189: only seems to bear a small imprint of the ellipsoidal variations. The spectral
190: type of the secondary star for MQ Dra was previously determined by S05
191: to be M5V. The $K$-band spectrum for this source (see Harrison et
192: al. 2005), however, suggests a spectral type closer to M4V due to the depth
193: of the Ca I triplet (at 2.26 $\mu$m). In the light curve modeling below, we
194: assume that the secondary star is an M4V.
195:
196: The IR light curve modeling was accomplished with the Wilson-Divinney code
197: WD2005\footnote{WD2005 is an updated version of WD98, and can be obtained at
198: this website maintained by J. Kallrath:
199: http://josef-kallrath.orlando.co.nz/HOMEPAGE/wd2002.htm}, which includes
200: reflection effects. For the modeling,
201: we assumed a white dwarf primary with T$_{\rm eff}$ = 8,000 K, an M4V
202: secondary (T$_{\rm eff}$ = 3,200 K), and used the square root limb darkening
203: coefficients from Claret (2000b), the gravity darkening value of 0.3 (Claret
204: 2000a) and an albedo of 0.60 (Nordlund \& Vaz 1990). The best fitting model for the $K$-band
205: has an orbital inclination of $i$ = 35$^{\circ}$, and is plotted in red
206: in Figure 2. It is clear from this figure that the light from just the
207: secondary star is inadequate to completely explain either the flux level or
208: variations seen in the $J$ and $H$-bands. Since cyclotron emission
209: from the $n$ = 1 (at 1.8 $\mu$m) and $n$ = 2 (at 0.9 $\mu$m) harmonics from a
210: B = 59 MG field is expected to be present in both the $J$ and $H$-bands, this
211: emission likely creates the observed discrepancy between the model and observed
212: light curves.
213: The value for the inclination derived assuming a contact binary is much
214: lower than that preferred by the cyclotron modeling (i = 60$^{\circ}$, see section
215: 6). If the assumption of a contact binary is relaxed, the infrared light
216: curves can be modeled with larger orbital inclinations. To investigate this
217: possiblity, we ran WD2005 in non-contact binary mode, but starting with a
218: model where the secondary star essentially filled its Roche lobe
219: (``R$_{contact}$''), and using stellar parameters that allowed us to reproduce
220: the light curves found for the contact case with i=35$^{\circ}$. We then proceeded
221: to shrink the radius of the secondary to create light curves for a system that
222: is further and further from contact. We were able to create models where
223: the light curves are well modeled using higher orbital inclinations. The
224: results are listed in Table 2. Note, however, that in this mode, the
225: differences in the observed minima in the model light curves are so small
226: as to be undetectable given the error bars on our photometry. This is true
227: even for the 94\% radius case. But the observed minima $do$ differ. It is
228: unclear whether this is a deficiency in the light curve modeling program, or
229: not, but further investigation will require stronger constraints on the masses
230: of the two components as well as a better estimate of the temperature of the
231: white dwarf.
232:
233: The optical light curves are similar
234: in amplitude and appearance to those shown in S03 (which used arbitrary
235: phasing), while the increased
236: coverage in 2007 allows for a better resolution of the features. In particular,
237: the double-humped structure in the $B$ light curve is confirmed. S03 had
238: speculated this could be the result of cyclotron beaming with a grazing
239: eclipse of the accretion area at the lowest dip (phase 0.55 in Figure 1).
240: It is intriguing that there is a slight dip in the FUV light curve near
241: phase 0.05 (which is near the secondary minimum in $B$) but the poor
242: time resolution of the FUV and the large error bar on this single point does
243: not permit any conclusive link.
244:
245: As in EF Eri (S06), the UV light curves of MQ Dra show a prominent
246: sinusoidal modulation, with
247: the peak UV flux close to the peak cyclotron times, thus strongly suggesting a
248: link between the UV and cyclotron sources (the magnetic pole area).
249: The amplitude of the modulation in the FUV in MQ Dra is
250: comparable to that in EF Eri, while the NUV amplitude in MQ Dra is even larger than
251: its FUV and about three times larger than the NUV of EF Eri. While EF Eri and
252: MQ Dra are known to have very different field strengths (13 vs 60 MG)
253: and white dwarf temperatures (9500K vs 8000K), it is obvious that a single
254: temperature white dwarf (i.e. as evidenced by a constant UV light curve)
255: cannot explain the observed light curves in either system.
256:
257: Two possible explanations come to mind: 1) there is a hot spot on the
258: white dwarf near the magnetic pole causing the
259: UV modulation, as used by Araujo-Betancor et al. (2005) to model the
260: UV fluxes of low state Polars and by G\"ansicke et al. (2006) to model
261: the UV light curves of AM Her or 2) the UV modulation is caused by
262: phase dependent cyclotron harmonics as is known to be the cause of the
263: optical variability. We explore these two options with simple models below.
264:
265: \section{MQ Dra Spot Model}
266:
267: To determine the feasibility of a hot spot, we used the
268: BINSYN modeling code (Linnell \& Hubeny 1996) to calculate synthetic light curves and
269: spectra of MQ Dra. The model code calculates the SED of a white dwarf using
270: TLUSTY (Hubeny 1988) and SYNSPEC (Hubeny, Lanz \& Jeffery 1994).
271: The TLUSTY model uses a specified mass and radius (producing an associated
272: log g) and SYNSPEC produces continuum spectra plus
273: absorption lines of H and He.
274: To minimize the free parameters, we used the simplest approximations of a
275: circular, isothermal hot spot to produce the light variation.
276: Recognizing that satisfactory models for radiation
277: transfer in the presence of a strong magnetic field are not available
278: (Wickramasinghe \& Ferrario 2000), we used white dwarf SEDs to represent the
279: spot. The BINSYN modeling code was then used to assign spot T$_{eff}$, latitude, longitude and angular
280: radius and, for a given parameter set, to calculate a series of synthetic
281: system spectra that included contributions from the white dwarf, the spot, and
282: the
283: secondary star.
284: The code then integrated, for each synthetic system spectrum, the products
285: of the FUV, NUV and optical passbands and the synthetic system spectrum. The
286: resulting calculated synthetic light curves were then compared with the observed light curves.
287: In our model, a key element in producing the light variation with orbital phase is
288: the variable foreshortening and limb darkening of the spot as the white dwarf
289: rotates. It is important to note, in the comparison with the observations, that
290: neither the FUV nor NUV observed light curves are perfect sinusoids.
291:
292: Previous estimates of the white dwarf temperature (obtained by subtracting the
293: secondary and cyclotron harmonics and fitting white dwarf fluxes; S03,
294: S05, Ferrario et al. poster at IAU Coll. 190, unpublished) obtained
295: temperatures from 6000-8000K with an upper limit of 10,000K. Our model uses a
296: temperature of 8000K. Since the Ferrario et al. model and our cyclotron
297: modeling preferred inclinations near 60$^{\circ}$, and our $JHK$ light curve
298: modeling set a minimum inclination of 35$^{\circ}$,
299: we constructed models using these two limits.
300:
301: Our search for a simple spot model to simulate both the FUV and NUV
302: light variation of MQ Dra was unsuccessful in that
303: the observed light variation cannot be simulated by our model
304: employing a single, isothermal, circular spot which does not extend
305: beyond the pole. The challenge is the
306: approximately equal light amplitudes in FUV and NUV. Many combinations
307: of spot parameters can simulate either light curve individually. Table 3 lists
308: one pair of models for each inclination with the corresponding light
309: curves shown in Figure 3. For i= 60$^{\circ}$, the maximum spot angular radius
310: is 15$^{\circ}$ to avoid extending beyond the pole; in this case the spot center
311: is at a latitude of 75$^{\circ}$. The lower inclination (35$^{\circ}$) produces similar
312: results.
313:
314: We simulated the NUV light curve
315: first, and the final parameters required a spot radius close to the limiting
316: value. The adopted T$_{eff}$ would not produce the observed light amplitude
317: for a smaller spot. An
318: appreciably larger T$_{eff}$ would have exacerbated the difference in FUV and NUV
319: synthetic light amplitudes for a given common spot size. The FUV spot angular
320: radius was determined after the spot T$_{eff}$ had been determined, and was adjusted
321: to produce a synthetic FUV light amplitude matching the observed amplitude.
322: Note how much smaller the FUV spot radius is, as compared with the NUV spot
323: radius. While a smaller FUV spot is reasonable if there is a temperature
324: stratification, i.e. a small hot area surrounded by a larger cool area, we
325: felt we did not have enough constraints within our limited data to further
326: increase
327: the number of free parameters.
328:
329: We also extended the model fits to the optical region. As expected, the white
330: dwarf and spot model produce synthetic light amplitudes much smaller than
331: the amplitudes observed in B or V.
332: This is not surprising as the optical spectra (S03,S05) show that
333: the optical is dominated by the cyclotron harmonics, not a stellar
334: continuum.
335:
336: These spot models are purely empirical. There are several options that could lead
337: to other solutions i.e. a non-isothermal spot with a smaller
338: and hotter central region contributing the bulk of the FUV source as noted
339: above, a non-circular
340: or vertically extended spot,
341: a source other than a spot (Section 6).
342: Constraining these options will require better
343: phase-resolved UV spectra to determine the source of the light.
344:
345: \section{EF Eri Spot Model}
346:
347: We also tried to model EF Eri with our latest spot model. Our original
348: attempt to fit the GALEX data on EF Eri (S06)
349: used the Wilson-Devinney code WD2005 which involves black
350: body representations for both the white dwarf (9800K) and the spot (20,000K).
351: Applying the better modeling technique described above for MQ Dra
352: with
353: the parameters of
354: EF Eri (Table 4) resulted in a spot of 24,000K with a radius of
355: 5.5$^{\circ}$ situated at a latitude of 60$^{\circ}$. In this case, a single size spot can
356: provide a reasonable match to the amplitudes of both the FUV and NUV light curves.
357: The fits to the UV light curves are shown in Figure 4 while the
358: contribution of the spot and white dwarf at the brightest phases of the lightcurve
359: are shown in Figure 5. As found by Sw07, using actual white dwarf
360: fluxes improves the match to the spectral energy distributions at maximum and minimum
361: light,
362: but Sw07 require a larger spot than our BINSYN model fits which use
363: a hotter, smaller spot to produce the same amount of FUV flux. Table 5
364: compares the parameters of the two models.
365: An important point is that while both models show that a spot model can
366: represent the light variation, there are enough free parameters that
367: multiple solutions are possible.
368: A key difference between the Sw07 model and ours is that their spot encompasses
369: the pole and depends on the disappearance of some of the spot below the horizon
370: (self-eclipse) during part of an orbital cycle in order
371: to produce the light variation.
372: Although their model produces light variation, we note that the spot is large
373: (half opening angle of 24$^{\circ}$) and the required spot size depends on the adopted
374: orbital inclination (Sw07 adopted 60$^{\circ}$, while we used 45$^{\circ}$
375: based on the Harrison et al. (2003) analysis, and Campbell et al. (2008) found
376: 55$^{\circ}$ from their cyclotron modeling). The Sw07 model would require
377: an appreciably larger spot if i=45$^{\circ}$.
378: As in MQ Dra, it is obvious that the shape of the UV light curves is not a
379: perfect sinusoid, implying the hot spot is not a simple circular, isothermal spot.
380:
381: \section{MQ Dra Cyclotron Model}
382:
383: The cyclotron modeling was done using a Constant Lambda (CL) code first built
384: by Schwope et al. (1990), which uses four parameters: B (the magnetic field
385: strength), kT (the plasma temperature), $\Theta$ (the viewing angle to the
386: magnetic field), and
387: log$\Lambda$ (the ``size-parameter'' of the system, which is a proxy for
388: column density). The optical data used are the time-resolved spectra
389: from the Bok 2.3m that are shown in Figure 9 of S05.
390:
391: To fit the models, we tried to keep the parameters as consistent with the
392: Ferrario et al. cyclotron model
393: results as possible, using B = 60 MG, kT = 1 keV, and
394: $i$ = 60$^{\circ}$ for initial constraints. Additionally, we used the fact
395: that there is a small amount of motion ($\sim$120\AA) in the $n$ = 3 (6000 \AA) harmonic to
396: constrain the viewing angle. Interestingly, this cyclotron
397: hump appears reddest (implying
398: the highest value of $\Theta$) at the cyclotron minimum (phase 0.43). Coupled with the fact
399: that the ratio of the $n$ = 3 and $n$ = 4 harmonics show the greatest parity
400: at that phase, we conclude that the highest value for the viewing angle must occur at
401: cyclotron minimum. We initially assumed that the orbitally averaged value
402: of $\Theta$ would equal the system inclination, and that $\beta$ (the angle
403: between the rotation and magnetic axis) was the
404: absolute angular deviation of the viewing angle, while log$\Lambda$ was left as a free
405: parameter. To match the observed spectra, the models were normalized to the
406: data at 0.62 $\mu$m. The phase-resolved cyclotron models are shown in Figure 6,
407: with Table 6 listing the values of the model parameters for each spectrum.
408: Roughly, the orbitally averaged values for the salient parameters are: B = 59
409: MG, kT= 1.8 keV, and log$\Lambda$ = 3.8. The derived geometry is consistent
410: with a single spot $i$ = 68$^{\circ}$, and $\beta$ = 7$^{\circ}$.
411:
412: While this cyclotron model does an excellent job at predicting the changing
413: morphology of the visual spectrum, the observed fluxes are difficult to
414: reconcile because for the derived geometry, the spot remains continuously
415: visible. With these parameters, our models suggest that the maximum amount
416: of flux from the cyclotron emission region should occur at $\phi$ = 0.43, but
417: this is near where the observed $minimum$ occurs. Given that the n = 3 and n = 4
418: harmonics have nearly identical fluxes, and are reddest at this phase,
419: the viewing angle should be at a maximum, which in turn should produce the
420: greatest amount of observed cyclotron emission. Comparing the raw,
421: un-normalized model intensities to the photometry suggests that the fluxes
422: at photometric minimum are down by a factor of six if we assume an equal
423: emitting area at all times. We can explain this variation if we assume the
424: values for $i$ and $\beta$ derived from the cyclotron modeling are incorrect,
425: and that the accretion region is partially self-eclipsed.
426:
427: For a simple geometry (a uniform, cylindrical column), we find that we can
428: relate $i$ + $\beta$ to the column height (h) through the following equation:
429:
430: h = (1 + $\eta$)R$_{\rm WD}$(tan$^{2}$($i$ + $\beta$ - 90) - 1.0)
431:
432: where $\eta$ is the fraction of the column that remains visible, and
433: R$_{\rm WD}$ is the radius of the white dwarf. Assuming values of 0.01 $\leq$
434: h $\leq$ 0.05 R$_{\rm WD}$, we found values of 97.5$^{\circ}$ $\leq$ ($i$ +
435: $\beta$) $\leq$ 106.5$^{\circ}$, as listed in Table 7. The geometries implied by
436: these combinations of $i$ and $\beta$ cannot reproduce the observed viewing
437: angles without curvature of the field lines. Assuming a dipole field,
438: Beuermann, Stella \& Patterson (1987) show that the angle of curvature, $b$, is
439: related to the magnetic colatitude by the equation $b$ = 3/2 $\beta$. We used
440: a root-finding scheme to find values of $b$ for a range of orbital
441: inclinations that reproduce the observed phase dependence of the viewing angle
442: $\theta$ (see Fig. 7). The mean value for $b$ is 68$^{\circ}$, which implies
443: $\langle$ $\beta$ $\rangle$ $\sim$ 45$^{\circ}$. This, in turn, restricts
444: the orbital inclination to the values listed in Table 7. We believe that
445: this is the simplest interpretation, but more complex field structures
446: and/or accretion regions could also reproduce the observations.
447:
448: Finally, we attempted to fit the observed SED at both photometric minimum and maximum.
449: In Figure 8, we show the observed photometry for these two phases,
450: obtained from the GALEX, NMSU, and KPNO data for the UV, optical, and
451: near IR, respectively. Using the color-magnitude table for pure hydrogen
452: white dwarfs with log($g$) = 8 in Bergeron, Wesemael, and Beauchamp (1995)
453: at 180 pc with $\lambda F_{\lambda}\simeq$ 20\% below the NUV photometry
454: at photometric minimum implies T $\simeq$ 8400 K. In this work, we adopt
455: a slightly lower temperature of 8000 K. This WD (green line) is co-added
456: with a M4 secondary normalized to the K-band photometry. In addition, we
457: added our optical cyclotron models for both phases, finding that the SEDs were
458: well reproduced in the IR and the optical, but that the noted UV photometric
459: variability was still unexplained. Therefore, we added a second cyclotron model to
460: explain the UV photometry. As in EF Eri (Campbell et al. 2008),
461: it was difficult to constrain the UV cyclotron models without
462: proper phase-resolved UV spectroscopy. We assumed
463: that the UV cyclotron model has identical model parameters to the
464: optical model, but with the magnetic field strength increased to push the
465: harmonics into the UV. A magnetic field of B = 235 MG produced harmonics in the
466: right places; this value was the lowest field strength
467: which could be reasonably fit to the observed data. This high field strength is not
468: unknown among Polars as AR UMa shows a similar value for B (Schmidt et al. 1996). To calibrate the fluxes, we
469: assumed that the peak flux (in $F_{\lambda}$) of the $n$ = 3 harmonics for both
470: models were identical. The model spectrum was then integrated over the
471: transmission function of the GALEX NUV and FUV bandpasses using the IRAF
472: package CALCPHOT, and the synthetic photometric point compared to the GALEX
473: photometry.
474:
475: At photometric maximum, we found that while the NUV point was well
476: predicted, the model could not produce enough flux in the FUV. Thus, we
477: increased log$\Lambda$ in that model from 3.6 to 4.0, which then adequately
478: explained the fluxes in that bandpass. At photometric minimum, however, we
479: found that the predicted flux was well down (by a factor of 5) from the
480: GALEX data. While the nature of this discrepancy remains unclear, it is
481: possible that the UV cyclotron emission region is not co-located
482: with the optical accretion spot. If it were located closer to
483: the pole, the accretion spot would just barely self eclipse, leaving a much
484: larger emitting surface area and that could possibly explain this result.
485:
486:
487: \section{Discussion}
488:
489: As UV data accumulate on Polars during low states of accretion, it is
490: increasingly obvious that some areas emitting substantial UV flux remain on the
491: white dwarf even
492: during states of very low accretion when there is no evidence of a mass
493: transfer stream. Studies of Polars at random times (Ramsay et al. 2004;
494: Araujo-Betancor et al. 2005) show that many Polars are in low states, implying
495: they spend much of their lifetime in these states, but the length of these
496: low states are not usually known and the data that do exist reveal the low
497: states vary in length. For a few well-studied systems, the extra
498: UV light over that of the white dwarf continuum
499: is known as a function of time since the onset of the low state and is
500: modelled with a hot spot. The
501: study of AM Her (G\"ansicke et al. 2006) with HST and FUSE for several months
502: starting 200 days after it began a low state showed that the spot was
503: about 10,000K cooler than its high state fitted value of 47,000K, but of a
504: similar size as compared to a high state of accretion.
505: There was also a slight change in latitude of the spot between the two states.
506: These characteristics did not change during the 4 months of the low state.
507: However, AM Her is known to have frequent high states of accretion which can
508: heat up a spot and the derived accretion rate during low states is fairly high at
509: 6$\times$10$^{-12}$M$_{\odot}$ yr$^{-1}$. This value is about a factor of 10
510: below the typical mass accretion rate for a Polar in a high accretion state
511: but a factor of 100 larger than a typical LARP (S05). The white dwarf temperature at 20,000K
512: is close to the value of typical cataclysmic variables that are heated by
513: long term accretion (Sion 1999, Townsley \& Bildsten 2004).
514:
515: Our UV (S06) and the
516: X-ray (Sw07) study of EF Eri probe a longer and lower state of accretion
517: compared to AM Her. The
518: GALEX and X-ray data were obtained 7 and 5 yrs respectively into the low state
519: that began in 1997 and has now reached 10 yrs in length. SW07
520: estimate the accretion luminosity during the low state as 2.4$\times$10$^{30}$
521: erg s$^{-1}$, which corresponds to an accretion rate of about
522: 3$\times$10$^{-13}$M$_{\odot}$ yr$^{-1}$. The corresponding specific accretion
523: rate of 0.01 g cm$^{-2}$s$^{-1}$
524: puts EF Eri into the bombardment regime where there is no shock at the accretion
525: column and there is only cyclotron cooling in the atmosphere of the white dwarf
526: which result in the observed harmonics in the near-IR (Kuijpers \& Pringle 1982;
527: Woelk \& Beuermann 1996; Wickramasinghe \& Ferrario 2000). The resulting lower
528: temperature of the white dwarf in EF Eri than in AM Her is consistent with low
529: accretion values and long time intervals spent at low states.
530:
531: MQ Dra is the most extreme case studied so far. It has one of the coolest
532: white dwarfs and one of the lowest accretion rates
533: (6$\times$10$^{-14}$M$_{\odot}$ yr$^{-1}$; S05) known among close binaries
534: containing magnetic white dwarfs.
535: It has never been seen in a high accretion state and the low
536: white dwarf temperature and long orbital period indicate it may never have been
537: in an active state. Thus, it is surprising that even in this regime, there is
538: a significant UV source of light above that from a white dwarf that fits
539: the optical continuum between cyclotron harmonics, and this source is highly
540: variable during the orbit.
541: If this UV source is from a heated area around the magnetic pole, the capture of the stellar wind from the secondary by the magnetic
542: white dwarf must be able to produce a significant heating of the atmosphere for
543: extended times. As noted by S05, Li et al. (1994;1995) determined that for
544: fields near 50-100MG, the magnetic white dwarf can capture all of the
545: stellar wind from the secondary, thus increasing the amount coming to
546: the white dwarf. The figures in these papers suggest that the wind still
547: is confined to a footprint near the magnetic pole(s). While the exact shape
548: of a heated spot is not clear in this situation, it is reasonable that
549: the heated area would be in the same location as the pole and in phase
550: with the optical cyclotron emission (the low optical depth ensures the
551: lowest harmonics are optically thin and hence the cyclotron is not
552: beamed as in the optically thick case). In addition, the magnetic coupling between the two stars
553: can lead to increased stellar activity (flares and coronal mass ejections)
554: on the secondary, thus increasing the accretion rate sporadically and
555: creating more heating (Howell et al. 2006). However, a simple hot spot is not an ideal fit to
556: the observations as
557: described above.
558:
559: An alternate possibility is that the extra UV flux and modulation is
560: caused by cyclotron harmonics in the UV.
561: The greatest
562: problem with this interpretation is that to have prominent harmonics in
563: the UV requires a magnetic field $>$ 200 MG. Possible harmonics are evident
564: in the UV in AR UMa (G\"ansicke et al. 2001) which is known to have
565: a field of about 240 MG. However, recent modeling of the optical flux
566: and polarization of EF Eri (Beuermann et al. 2007) shows that this system
567: (as well as other Polars) has complex
568: fields, with multipoles that could include fields up to 110MG. Recent success
569: in modeling IR phase-resolved spectra of EF Eri (Campbell et al. 2008)
570: was achieved with a two component cyclotron model. The extension of this type of model
571: to the UV with a field of 115 MG can explain the amplitudes of
572: the {\it GALEX} light curves for EF Eri. If MQ Dra has such a complex multipole field,
573: then the optical cyclotron could be produced by fields near 60 MG and
574: the UV cyclotron by the components near 200 MG. The correct interpretation
575: could be resolved by time-resolved UV spectra which would reveal whether
576: these cyclotron harmonics actually exist in the UV.
577:
578: \section{Conclusions}
579:
580: Our GALEX NUV and FUV light curves have revealed that the lowest accretion
581: rate Polars, with no active stream accretion, still show large amplitude
582: periodic UV variability with similar phasing and amplitude to the optical
583: light curves. The extreme
584: case of MQ Dra, with a long orbital period and a white dwarf of only 8000K,
585: shows similar FUV variability to that of EF Eri, a Polar with extended low states
586: but also a known high state of accretion in the past.
587: The sequence of white dwarf temperatures appears to
588: follow the accretion rates, with the lowest temperature white dwarfs existing
589: in the systems with the lowest
590: accretion, implying some long term heating effects even in these cases of
591: wind-accretion.
592: However, attempts to fit the UV light curves with a simple isothermal hot spot
593: cannot adequately
594: describe the observations. While this simple type of spot can produce the amplitudes
595: of the NUV and FUV modulation in EF Eri,
596: it cannot reproduce the shape of
597: the light curves well, implying distorted, multi-temperature spots or additional
598: components. MQ Dra is harder to fit with a spot than EF Eri, as the similar amplitudes in NUV and
599: FUV cannot be reproduced with a similar size spot at a given temperature.
600:
601: Another explanation of the UV phasing and amplitudes might be possible
602: using cyclotron harmonics, but only if the white dwarf has multipoles with
603: fields extending above 200 MG.
604: While we could adequately model optical phase-resolved spectroscopy of MQ Dra
605: with a cyclotron model having B = 59 MG, kT $\simeq$ 2.0 keV, and log$\Lambda$ =
606: 4.0, both the observed motion of the
607: harmonics and the morphological appearance of the spectra at cyclotron minimum
608: and maximum defied our simplistic expectations. To explain these results
609: in a consistent way would require a secondary that does not fill its
610: Roche lobe, an accretion column that has a high magnetic co-latitude
611: ($\beta$ = 45$^{\circ}$), field lines that are tilted relative to the local normal,
612: and an emitting region of the white
613: dwarf that decreases by a factor of 6 from cyclotron maximum to minimum.
614: The SED from the FUV to the $K$-band can be fit with
615: an 8000 K WD, an M4 secondary normalized to the $K$-magnitude at
616: photometric minimum and two cyclotron models (our derived optical
617: cyclotron model and a second analogous cyclotron model with an equivalent peak
618: flux in $n$ = 3). This combination can reproduce the observed UV photometric
619: variability, but
620: at photometric minimum the UV cyclotron model must be scaled by an arbitrary
621: value of 5 (possibly explained by moving the UV accretion spot
622: closer to the rotation axis).
623:
624: UV phase resolved spectra are needed to determine which model to pursue
625: in more detail. The spectra would enable the temperature structure of a
626: hot spot to be determined and reveal if cyclotron features from high
627: fields are present.
628: Further theoretical models of the atmospheric heating and cooling in the
629: bombardment case of very low specific accretion would also help to
630: determine the nature of the heated areas surrounding the magnetic poles
631: at these low rates.
632:
633: \acknowledgments
634:
635: Support for this research was provided by NASA GALEX grant NNG05GG46G.
636: We thank Hugh Harris for providing standard stars for the calibration
637: of the optical data and Gary Schmidt for the use of his optical spectra
638: of MQ Dra for the cyclotron model and for his comments on our models.
639:
640: \begin{references}
641:
642: \reference{} Araujo-Betancor, S. et al. 2005, \apj, 622, 589
643: \reference{} Bergeron, P., Wesemael, F. \& Beauchamp, A. \pasp, 107, 1047
644: \reference{} Beuermann, K., Stella,L. \& Patterson, J., 1987, \apj, 316, 360
645: \reference{} Beuermann, K., Euchner, F., Reinsch, K., Jordan, S. \& G\"ansicke, B. T. 2007, \aap, 463, 647
646: \reference{} Campbell, R. K., Harrison, T. E., Schwope, A. \& Howell, S. B. 2008, \apj, 672, 531
647: \reference{} Claret, A. 2000a, \aap, 359, 289
648: \reference{} Claret, A. 2000b, \aap, 363, 1081
649: \reference{} G\"ansicke, B. T., Long, K. S., Barstow, M. A. \& Hubeny, I. 2006, \apj, 639, 1039
650: \reference{} G\"ansicke, B. T., Schmidt, G. D., Jordan, S. \& Szkody, P. 2001, \apj, 555, 380
651: \reference{} Harrison, T. E., Howell, S. B., Szkody, P. \& Cordova, F. A. 2005, \apj, 632, L123
652: \reference{} Howell, S. B., et al. 2006, \apj, 652, 709
653: \reference{} Hubeny, I. 1988, Comp. Phys. Comm., 52, 103
654: \reference{} Hubeny, I., Lanz, T., \& Jeffery, C.~S. 1994, in Newsletter on
655: Analysis of Astronomical Spectra No. 20, ed. C. S. Jeffery (CCP7;St. Andrews: St. Andrews Univ.), 30
656: \reference{} Kuijpers, J. \& Pringle, J. E. 1982, \aap, 114, L4
657: \reference{} Li, J., Wickramasinghe, D. T., \& Wu, K. W. 1995, \mnras, 276, 255
658: \reference{} Li, J., Wu, K. W. \& Wickramasinghe, D. T. 1994, \mnras, 268, 61
659: \reference{} Linnell, A. P. \& Hubeny, I. 1996, \apj, 471, 958
660: \reference{} Martin, D. C. et al. 2005, \apj, 619, L1
661: \reference{} Nordlund, A. \& Vaz, L. P. R. 1990, \aap, 228, 231
662: \reference{} Ramsay, G. et al. 2004, \mnras, 350, 1373
663: \reference{} Schmidt, G. D. et al. 1996, \apj, 473, 483
664: \reference{} Schmidt, G. D. et al. 2005, \apj, 630, 1037 (S05)
665: \reference{} Schwope, A. D. 1990, Reviews in
666: Modern Astronomy, 3, 44
667: \reference{} Schwope, A. D. et al. 2002, in ASP Conf. Vol 261, The Physics of
668: Cataclysmic Variables and Related Objects, ed. B. T. G\"ansicke, K. Beuermann, \& K. Reinsch (San Francisco:ASP), 102
669: \reference{} Schwope, A. D., Staude, A., Koester, D. \& Vogel, J. 2007, aap, 469, 1027 (Sw07)
670: \reference{} Sion, E. M. 1999, \pasp, 111, 532
671: \reference{} Szkody, P. et al. 2003, \apj, 583, 902 (S03)
672: \reference{} Szkody, P. et al. 2004, \aj, 128, 2443
673: \reference{} Szkody, P. et al. 2006, \apj, 646, L147 (S06)
674: \reference{} Townsley, D. M. \& Bildsten, L. 2004, \apj, 600, 390
675: \reference{} Wickramasinghe, D. T. \& Ferrario, L. 2000, \pasp, 112, 873
676: \reference{} Woelk, U. \& Beuermann, K. 1996, \aap, 306, 232
677: \end{references}
678:
679: \clearpage
680: \begin{deluxetable}{ccl}
681: \tablewidth{0pt}
682: \tablecaption{Summary of Observations}
683: \tablehead{
684: \colhead{Date} & \colhead{Time} & \colhead{Data}}
685: \startdata
686: 20050607 & 04:48-19:52 & GALEX NUV 10 visits\\
687: 20050723 & 11:53-12:13 & GALEX NUV 1 visit \\
688: 20060715 & 16:40-17:04 & GALEX NUV,FUV 1 visit \\
689: 20060716 & 01:08-02:55 & GALEX NUV,FUV 2 visits \\
690: 20060716 & 07:27-09:30 & GALEX NUV,FUV 2 visits \\
691: 20060716 & 17:20-21:02 & GALEX NUV,FUV 3 visits \\
692: 20060717 & 01:46-03:35 & GALEX NUV,FUV 2 visits \\
693: 20060722 & 03:13-03:33 & GALEX NUV,FUV 1 visit \\
694: \enddata
695: \end{deluxetable}
696:
697: \clearpage
698: \begin{deluxetable}{cc}
699: \tablewidth{0pt}
700: \tablecaption{Roche-filling Factor vs Inclination}
701: \tablehead{
702: \colhead{R/R$_{contact}$} & \colhead{i$^{\circ}$} }
703: \startdata
704: 1.00 & 35 \\
705: 0.94 & 40 \\
706: 0.90 & 47 \\
707: 0.86 & 54 \\
708: 0.82 & 65 \\
709: \enddata
710: \end{deluxetable}
711:
712: \clearpage
713: \begin{deluxetable}{lcc}
714: \tablewidth{0pt}
715: \tablecaption{MQ Dra Model Parameters}
716: \tablehead{
717: \colhead{Parameter} & \colhead{i=60$^{\circ}$} & \colhead{i=35$^{\circ}$}}
718: \startdata
719: WD T$_{eff}$ & 8000K & 8000K \\
720: WD mass & 0.6M$_{\odot}$ & 0.6M$_{\odot}$ \\
721: WD radius & 0.012R$_{\odot}$ & 0.012R$_{\odot}$ \\
722: WD log g & 8.0 & 8.0 \\
723: Spot T$_{eff}$ & 20,000K & 24,000K \\
724: Spot latitude & 73$^{\circ}$ & 65$^{\circ}$ \\
725: Spot ang rad & 5$^{\circ}$ (FUV) & 3.5$^{\circ}$ (FUV) \\
726: Spot ang rad & 12$^{\circ}$ (NUV) & 14$^{\circ}$ (NUV) \\
727: \enddata
728: \end{deluxetable}
729:
730: \clearpage
731: \begin{deluxetable}{lc}
732: \tablewidth{0pt}
733: \tablecaption{EF Eri Model Parameters}
734: \tablehead{
735: \colhead{Parameter} & \colhead{Value}}
736: \startdata
737: WD T$_{eff}$ & 9500K \\
738: WD mass & 0.6M$_{\odot}$ \\
739: WD radius & 0.012R$_{\odot}$ \\
740: WD log g & 8.0 \\
741: Orbit i & 45$^{\circ}$ \\
742: Spot T$_{eff}$ & 24,000K \\
743: Spot latitude & 60$^{\circ}$ \\
744: Spot ang rad & 5.5$^{\circ}$ \\
745: \enddata
746: \end{deluxetable}
747:
748: \clearpage
749: \begin{deluxetable}{lcc}
750: \tablewidth{0pt}
751: \tablecaption{Comparison of EF Eri Models}
752: \tablehead{
753: \colhead{Parameter} & \colhead{Sw07} & \colhead{This work}}
754: \startdata
755: WD T(K) & 9750 & 9500 \\
756: Spot T (K) & 18,500 & 24,000 \\
757: Inclination ($^{\circ}$) & 60 & 45 \\
758: Spot Ang radius ($^{\circ}$) & 24 & 5.5 \\
759: Spot Latitude ($^{\circ}$) & 77.5 & 60 \\
760: \enddata
761: \end{deluxetable}
762:
763: \clearpage
764: \begin{deluxetable}{cccclc}
765: %\tabletypesize{\scriptsize}
766: \tablecaption{Cyclotron Modeling Parameters for MQ Dra}
767: \tablewidth{0pt}
768: \tablehead{
769: \colhead{Spectrum} & \colhead{Phase} & \colhead{B (MG)} & \colhead{T (keV)} &
770: \colhead{$\Theta$} &\colhead {$Log(\Lambda)$} }
771: \startdata
772: 1 & 0.67 & 59.0 & 2.0 & 70.0 & 4.1\\
773: 2 & 0.73 & 59.0 & 1.6 & 68.0 & 4.1\\
774: 3 & 0.82 & 59.0 & 1.6 & 65.0 & 3.8\\
775: 4 & 0.89 & 59.0 & 1.8 & 63.0 & 3.6\\
776: 5 & 0.95 & 59.0 & 1.8 & 61.0 & 3.6\\
777: 6 & 0.02 & 59.0 & 1.8 & 63.0 & 3.6\\
778: 7 & 0.09 & 59.0 & 1.8 & 65.0 & 3.6\\
779: 8 & 0.16 & 59.0 & 1.8 & 67.0 & 3.7\\
780: 9 & 0.23 & 59.0 & 1.8 & 70.0 & 3.7\\
781: 10 & 0.30 & 59.0 & 1.9 & 72.0 & 3.7\\
782: 11 & 0.36 & 59.0 & 2.0 & 74.0 & 3.9\\
783: 12 & 0.43 & 59.0 & 2.0 & 76.0 & 3.9\\
784: 13 & 0.50 & 59.0 & 2.0 & 74.0 & 3.9\\
785: 14 & 0.57 & 59.0 & 2.0 & 72.0 & 3.9\\
786: \enddata
787: \end{deluxetable}
788:
789: \clearpage
790: \begin{deluxetable}{ccc}
791: %\tablewidth{57pc}
792: \tablewidth{0pt}
793: \tablecaption{Relation of Column Height to $i$ and $\beta$}
794: \tablehead{
795: \colhead{h$_{col}$ (R$_{WD}$)} & \colhead{$i$ + $\beta$} & \colhead{$i$} }
796:
797: \startdata
798: 0.01 & 97.5$^{\circ}$ & 52.5$^{\circ}$ \\
799: 0.02 & 100.5$^{\circ}$ & 56.5$^{\circ}$ \\
800: 0.03 & 102.9$^{\circ}$ & 57.9$^{\circ}$ \\
801: 0.04 & 104.8$^{\circ}$ & 59.8$^{\circ}$ \\
802: 0.05 & 106.5$^{\circ}$ & 61.5$^{\circ}$ \\
803: \enddata
804: \end{deluxetable}
805:
806: \clearpage
807: \begin{figure} [h]
808: \figurenum {1}
809: \plotone{f1.eps}
810: \caption{GALEX FUV and NUV and optical B,V,R,I filter light curves as a
811: function of phase (phase 0 is cyclotron max and inferior conjunction of
812: secondary). Light curves are repeated from phases 1.0 to 2.0, with error bars shown only
813: for the first cycle. Error bars for the optical data are smaller
814: than the points.}
815: \end{figure}
816:
817: \clearpage
818: \begin{figure}
819: \figurenum {2}
820: \epsscale{0.8}
821: \plotone{f2.eps}
822: \caption{$J,H,K$ light curves of MQ Dra (green is May 28 and black is May 29)
823: fit with a model of an M4 secondary filling its Roche lobe
824: and an 8000K white dwarf. Inclinations of 40, 35 and 30$^{\circ}$ are shown as
825: blue, red and yellow lines respectively; for clarity, only the best fitting
826: 35$^{\circ}$ fit is shown for J,H.}
827: \end{figure}
828:
829: \clearpage
830: \begin{figure}
831: \figurenum {3}
832: \epsscale{0.9}
833: \plotone{f3.eps}
834: \caption{Spot Model fits from Table 3 for the GALEX FUV (top) and NUV (bottom)
835: light curves of MQ Dra for
836: inclinations of 35$^{\circ}$ (left) and 60$^{\circ}$ (right).}
837: \end{figure}
838:
839: \clearpage
840: \begin{figure}
841: \figurenum {4}
842: \plotone{f4.eps}
843: \caption{Model fit for EF Eri with a 24,000K spot at 60$^{\circ}$ latitude and 5.5$^{\circ}$ radius
844: on a 9500K white dwarf.}
845: \end{figure}
846:
847: \clearpage
848: \begin{figure}
849: \figurenum {5}
850: \plotone{f5.eps}
851: \caption{Contributions of the white dwarf and spot in EF Eri at maximum light during the orbit.}
852: \end{figure}
853:
854: \clearpage
855: \begin{figure}
856: \figurenum {6}
857: \epsscale{0.80}
858: \plotone{f6.eps}
859: \caption{Phase-resolved spectra of MQ Dra, data (in black) from S05 are
860: stacked with a constant increment of 2.0 $\times$ 10$^{-12}$
861: erg s$^{-1}$ cm$^{-2}$. The best fit cyclotron models (in green) are
862: fit over the cyclotron models, with spectrum 1 at bottom and 14 at top from
863: Table 6.}
864: \end{figure}
865:
866: \clearpage
867: \begin{figure}
868: \figurenum {7}
869: \epsscale{0.60}
870: \plotone{f7.eps}
871: \caption{ (top) The change in the local field angle for three orbital
872: inclinations 35$^{\circ}$ (circles), 45$^{\circ}$ (squares), 55$^{\circ}$ (triangles), that
873: reproduced the observed viewing angle variation (bottom panel, circles).
874: The colored lines shown in the top panel to guide the eye are reproduced in
875: the bottom panel to demonstrate that this interpretation can reproduce
876: the observed variation in viewing angle.
877: }
878: \end{figure}
879:
880: \clearpage
881: \begin{figure}
882: \figurenum {8}
883: \epsscale{1.0}
884: \plottwo{f8a.eps}{f8b.eps}
885: \caption{(a) Cyclotron model SED of MQ Dra at photometric minimum ($\phi$ = 0.65). Horizontal
886: lines are GALEX bandpasses, points are UV
887: optical and IR photometric observations (red are photometric min, blue are
888: photometric max), and the green line is the spectrum of an 8000K
889: WD, normalized to $\simeq$ 20\% below the NUV photometric point. Stars are an M4
890: spectrum normalized to the $K$-band point. Our cyclotron model at photometric
891: minimum and an analogous UV model, with B = 235 MG, are added in normalized so
892: that the
893: peak-value of $F_{\lambda}$ in the $n$ = 3
894: harmonics of both the optical and UV cyclotron models are identical, but
895: the UV model is scaled up by a factor of five (see text for details).
896: The overall fit is plotted in red.
897: Orange hexagons are the NUV and FUV photometry implied by our models assuming
898: the NUV and FUV transmissions. (b)
899: Similar but for photometric max, with the final
900: composite model now shown in blue. No
901: scaling to account for the UV fluxes is used in this case.}
902: \end{figure}
903:
904: \end{document}
905: