1: \documentclass[12pt,preprint]{aastex}
2: % \documentclass[preprint2]{aastex}
3:
4: % This package is needed by pdflatex to convert eps figures into pdf files
5: \usepackage{epstopdf}
6:
7: %\setlength{\textheight}{9in}
8: %\setlength{\textwidth}{6.25in}
9: %\setlength{\topmargin}{-0.5in}
10: %\setlength{\oddsidemargin}{0.1in}
11:
12: %% Some math symbols should not be italic because they are not variables
13: \newcommand{\me}{\mathrm{e}}
14: \newcommand{\mi}{\mathrm{i}}
15: \newcommand{\dif}{\mathrm{d}}
16:
17: %% package needed to declare NEW math operators
18: %\usepackage{amsmath}
19: %\DeclareMathOperator{\sinc}{\rm sinc}
20:
21:
22: \begin{document}
23:
24: \title{\sc Constraining the Physical Parameters of the
25: Circumstellar Disk of $\chi$ Ophiuchi}
26:
27: \author{
28: C. Tycner,\altaffilmark{1}
29: C.~E.~Jones,\altaffilmark{2}
30: T.~A.~A.~Sigut,\altaffilmark{2}
31: H.~R.~Schmitt,\altaffilmark{3,4}
32: J.~A.~Benson,\altaffilmark{5}
33: D.~J.~Hutter,\altaffilmark{5}
34: R.~T.~Zavala\altaffilmark{5}}
35:
36: \altaffiltext{1}{Department of Physics, Central Michigan University,
37: Mt. Pleasant, MI 48859}
38: \altaffiltext{2}{Department of Physics and Astronomy, The University
39: of Western Ontario, London, Ontario, N6A~3K7, Canada}
40: \altaffiltext{3}{Naval Research Laboratory, Remote Sensing Division,
41: Code 7215, 4555 Overlook Ave. SW, Washington, DC 20375 }
42: \altaffiltext{4}{Interferometrics, Inc., 13454 Sunrise Valley Drive,
43: Suite 240, Herndon, VA 20171}
44: \altaffiltext{5}{US Naval Observatory, Flagstaff Station, 10391
45: W.~Naval Observatory Rd., Flagstaff, AZ 86001}
46:
47: \slugcomment{Accepted for publication in ApJ}
48:
49:
50: \begin{abstract}
51: We present a numerical model describing a circularly symmetric gaseous
52: disk around the Be star $\chi$~Ophiuchi. The model is constrained by
53: long-baseline interferometric observations that are sensitive to the
54: H$\alpha$ Balmer line emission from the disk. For the first time our
55: interferometric observations spatially resolve the inner region of the
56: circumstellar disk around $\chi$~Ophiuchi and we use these results to
57: place a constraint on the physical extent of the H$\alpha$-emitting
58: region. We demonstrate how this in turn results in very specific
59: constraints on the parameters that describe the variation of the gas
60: density as a function of radial distance from the central star.
61: \end{abstract}
62:
63: \keywords{stars: emission-line, Be --- stars: individual ($\chi$~Oph)
64: --- techniques: interferometric}
65:
66:
67: \section{Introduction}
68:
69: Models of the circumstellar gas around some B-type stars have been
70: constructed for a number of years. Because the recombination in the
71: circumstellar gas produces a series of emission lines, most commonly
72: in the hydrogen Balmer series, such systems are usually denoted
73: B-emission or Be stars for short. Models describing the geometry of
74: this circumstellar material have varied from spherical
75: shells~\citep{Gehrz74}, to slabs with constant
76: thickness~\citep{Kastner89}, to disks with constant opening
77: angles~\citep{Waters86}, to disks with an exponential density fall-off
78: in the direction perpendicular to the plane of the
79: disk~\citep{Marlborough69}. Although the first models assumed a
80: constant temperature throughout the disk, more recent models, such as
81: the ones developed by \citet{Millar98, Millar99}, \citet{Carciofi06},
82: and \citet{Sigut07}, obtain a self-consistent temperature throughout
83: the disk by enforcing radiative equilibrium (namely, by balancing the
84: microscopic heating and cooling rates operating within the disk).
85:
86: While the temperature structure of Be star disks is now the subject of
87: sophisticated modeling, the density structure largely remains {\it
88: ad-hoc\/} with simple, parameterized models. Perhaps the most popular
89: is a power-law density decrease with radius in the equatorial plane
90: coupled with the aforementioned exponential drop perpendicular to the
91: disk. This vertical density structure is assumed to result from the
92: gravitational equilibrium set by the balance between the gradient in
93: the gas pressure and the vertical component of the star's
94: gravitational acceleration. Despite the simplicity of such models,
95: they have been quite successful in matching a wide range of
96: observational diagnostics that are sensitive to the geometric
97: distribution of the circumstellar gas. These include the infrared
98: excess exhibited by Be stars over normal B stars of the same spectral
99: class \citep{dou94}, the observed continuum linear polarization of Be
100: stars~\citep{woo97}, and the shapes of the emission lines observed in
101: Be star spectra \citep[see for example][]{hum00a}. All of these
102: diagnostics rely only on spatially unresolved spectra or
103: spectropolarimetry. However, even more powerful constraints on the
104: geometry of the emitting regions in Be stars can be obtained by
105: combining these spatially {\it unresolved\/} diagnostics with new
106: optical interferometric observations that directly {\it resolve\/} the
107: structure of the emitting region on the sky~\citep{Quirrenbach97,
108: Tycner06b, Gies07}.
109:
110:
111: To demonstrate how specific constraints can be placed on a disk model
112: using observations that spatially resolve the source, we chose to
113: concentrate on the Be star $\chi\;$Oph (B2Ve; HR~6118; HD~148184).
114: This relatively nearby star with a {\it Hipparcos} distance of $150\pm
115: 17$~pc~\citep{Perryman97} has a strong H$\alpha$ emission and
116: therefore is an ideal candidate for probing the disk structure at
117: these wavelengths.
118:
119:
120: \section{Observations}
121:
122: \subsection{Interferometry}
123:
124: We acquired interferometric observations of $\chi$~Oph on four nights
125: between 2006~June~11 and 2006~June~18. All observations were made
126: with the same instrumental configuration. We used 4 different
127: telescopes, resulting in five unique baselines, with one baseline
128: signal measured twice on two independent output beams~\citep[see
129: Fig.~2 in][for a schematic of the beam combiner]{Tycner06b}. The
130: baselines used and their corresponding lengths are: AC--AE (18.9~m),
131: AC--AW (22.2~m), AW--W7 (29.5~m), AC--W7 (51.6~m), AE--W7 (64.4~m),
132: with the AC--W7 baseline measured on two output beams. The
133: observations were obtained using two spectrographs, each recording
134: simultaneously signal from 3 baselines at 16 spectral channels in the
135: wavelength range 560--870~nm. The resulting $(u,v)$-plane coverage of
136: the interferometric observations obtained from the spectral channel
137: containing the H$\alpha$ emission line is shown in
138: Figure~\ref{fig:chi-oph-uv}.
139:
140: The observations of $\chi$~Oph were interleaved with observations of a
141: nearby calibrator star $\zeta$~Oph (O9V; HR~6175; HD~149757). The
142: choice of the calibrator star was based on the relative proximity on
143: the sky to the target and the spectral type, which was similar to that
144: of the target star. Furthermore, the angular diameter of $\zeta$~Oph
145: has been directly measured by~\citet{Hanbury74} using an intensity
146: interferometer with sufficient accuracy to be competitive with other
147: methods that estimate angular diameters based on theoretical or
148: empirical models. Therefore, for the calibrator star, we adopted a
149: uniform disk angular diameter of 0.50$\pm$0.05~mas as obtained
150: by~\citet{Hanbury74}.
151:
152:
153: The data reductions followed standard NPOI procedures and will only
154: briefly be described here. The raw data are processed to produce
155: squared visibilities ($V^2$) averaged into 1~s intervals. These data
156: are then flagged to eliminate points with fringe tracking and pointing
157: problems \citep{Hummel98,Hummel03}. The flagged data set is then used
158: to obtain 30~s averages~(we refer to these averages as scans), which
159: are bias corrected using data obtained off the fringe (these are also
160: known as incoherent scans). Table~\ref{tab:obs} lists the individual
161: nights and the number of scans acquired on each night. Lastly, the
162: squared visibilities from the spectral channel containing the
163: H$\alpha$ emission line are calibrated with respect to the continuum
164: channels~\citep[see][for a detailed description of the
165: procedures]{Tycner03, Tycner06a}. Our final data set~(shown in
166: Fig.~\ref{fig:chi-oph-best-fit}) consists of the ensemble of all the
167: $V^2$ data from the H$\alpha$ channel obtained on all four nights
168: listed in Table~\ref{tab:obs}. The individual calibrated H$\alpha$
169: $V^2$ values are also listed in Table~\ref{tab:v2data}. Because
170: squared visibilities represent the normalized Fourier power of the
171: source structure on the sky, the interferometric observations can be
172: compared directly to the Fourier transform of the synthetic image
173: produced by the model.
174:
175:
176:
177: The observations of $\chi$~Oph used in this study were not optimized
178: for the purpose of calibrating with respect to an external calibrator.
179: However, such a calibration allows one to inspect the continuum
180: channels for the signature of a resolved stellar disk or a binary
181: companion. Using the scans obtained on $\zeta$~Oph to obtain the
182: instrumental system response, we found that large systematic residuals
183: from scan to scan remained after the calibration, which we attribute
184: to atmospheric variations on a timescale of minutes that are poorly
185: sampled by the scans on the calibrator. For this reason, we can only
186: conclude that the central star at the continuum channels is marginally
187: resolved with a uniform disk diameter of $\lesssim$~1~mas. An
188: interferometric signature of a resolved binary is less susceptible to
189: systematic variations from scan to scan because of its distinct
190: sinusoidal functional form that can be seen across different spectral
191: channels. In fact, the NPOI has been used successfully to detect
192: binaries over a wide range of separations~\citep[see for
193: example][]{Hutter04}. The search for a binary signature in
194: observations of $\chi$~Oph, however, did not yield any convincing
195: evidence of such a signature in our interferometric observations that
196: would correspond to a magnitude difference~($\Delta m$) of $\sim 3$ or
197: less at separations of few hundred milli-arcseconds~(mas) or less.
198: Although we cannot rule out a fainter companion, such a companion
199: would not contribute significantly to the total emission from the
200: system and therefore for the purpose of the analysis presented in this
201: study, $\chi$~Oph can be treated as a single star. It should be
202: noted, however, that $\chi$~Oph has been classified as a single-lined
203: spectroscopic binary by more than one study~\citep{Abt78, Harmanec87,
204: Levato87}, although there is no agreement on the period of the binary,
205: which was estimated to be 138.8~d by~\citet{Abt78} and 34.1~d
206: by~\citet{Harmanec87}.
207:
208:
209: \subsection{Spectroscopy}
210:
211: To estimate the strength of the H$\alpha$-emission in $\chi$~Oph at
212: the time of interferometric observations we obtained a high resolution
213: spectrum in the H$\alpha$ region using a fiber-fed Echelle
214: spectrograph at the Lowell Observatory's John S. Hall telescope. The
215: spectrum was acquired on 2006~June~10, just one day before our one
216: week long interferometric run. The spectroscopic data were processed
217: using standard routines developed specifically for the instrument used
218: to acquire the observations~\citep{Hall94}. The final reduced
219: spectrum in the H$\alpha$ region reaches a resolving power of 10,000
220: and a signal-to-noise ratio of few hundred (see
221: Fig.~\ref{fig:Halpha}).
222:
223: In addition to the spectrum shown in Figure~\ref{fig:Halpha} we have
224: also acquired spectroscopic observations of $\chi$~Oph in 2006 May and
225: 2006 September. The peak intensities and the equivalent widths (EWs)
226: of the H$\alpha$ emission line all agree to within 1\% with the line
227: profile shown in Figure~\ref{fig:Halpha}. Therefore, we assume that
228: during our interferometric run the overall emission in the H$\alpha$
229: line was stable and combining all of the interferometric observations
230: into one data set is justified.
231:
232:
233:
234: \section{Determining the Size of the H$\alpha$-emitting Region}
235: \label{sec:interferometry}
236:
237: We followed the procedure described by \cite{Tycner06b} to determine
238: the angular extent of the H$\alpha$ disk. We fitted a circularly
239: symmetric Gaussian model to the interferometric data and obtained a
240: best-fit diameter (defined as full-width at half-maximum; FWHM) for
241: the disk of 3.46~$\pm$~0.07~mas, which at a distance of 150~pc
242: corresponds to a diameter of 0.52~AU~(112~$R_\sun$). The fit results
243: in a reduced $\chi^2$ value of 0.95, and therefore, there is no
244: indication that the observational signature deviates from circular
245: symmetry or a Gaussian shape for that matter~(shown as dash-dotted
246: line in Fig.~\ref{fig:chi-oph-best-fit}). In fact, this is the best
247: observational evidence to date that a Gaussian model is not only the
248: simplest mathematical description of the radial distribution of the
249: H$\alpha$ emission in a circumstellar disk, but it is also fully
250: consistent with the observed distribution. This is in addition to
251: similar conclusions obtained previously by \citet{Tycner06b} for two
252: other Be stars, $\gamma$~Cas and $\phi$~Per.
253:
254:
255: It is interesting to compare our results based on the Gaussian model
256: fit with the simple estimator based on H$\alpha$ EW obtained by
257: \citet{Grundstrom06}. With the stellar parameters for $\chi$~Oph
258: listed in Table~\ref{tab:parameters} and assuming that the disk is
259: viewed at 20$^{\circ}$~(see \S~\ref{sec:grid}), for an EW of $-7.1$~nm
260: based on our spectroscopic observations \citet{Grundstrom06}
261: predict\footnote{We have used the programs provided by the authors at
262: http://www.chara.gsu.edu/$\sim$gies/Idlpro/BeDisk.tar to extrapolate
263: for equivalent widths beyond the values considered in Fig.~1 of
264: \citet{Grundstrom06}.} a disk radius of 9.2~$R_{\star}$. This agrees
265: well with the radius of $\approx 9.8 R_{\star}$ we obtain based on the
266: Gaussian fit~(using half of our FWHM measure).
267:
268:
269: Although characterizing the H$\alpha$ emitting region with a Gaussian
270: model can be a useful tool, especially when only a limited
271: interferometric data set is available to constrain the disk
272: characteristics, it is only a simple observational parameterization.
273: However, in cases where the interferometric data set covers wide range
274: of spatial frequencies it is desirable to extract more information
275: about the physical characteristics of the disk, such as density and
276: temperature distribution. For this purpose a much more sophisticated
277: model is needed.
278:
279:
280:
281: \section{Disk Model}
282: \label{sec:disk}
283:
284: The radiative equilibrium model used to represent the thermal
285: structure of the gaseous circumstellar disk surrounding $\chi\,$Oph
286: has been computed using the {\sc bedisk} code, which is described in
287: detail by~\citet{Sigut07}. This code incorporates many improvements
288: over previous treatments, most notably the use of a solar chemical
289: composition for the circumstellar gas, which is an important
290: ingredient for calculating heating and cooling rates. These models are
291: ideally suited to compare with interferometric observations because
292: model monochromatic, 2-dimensional images of the circumstellar disk at
293: specific wavelengths can be computed. The {\sc bedisk} code can also
294: be used to compute the hydrogen line spectra and the overall spectral
295: energy distribution. For the calculation of the hydrogen line
296: profiles, the disk is assumed to be in pure Keplerian rotation. There
297: is considerable evidence that Be star disks are indeed rotationally
298: supported based on the detailed analysis of line profiles
299: \citep{hum00b} and on the interpretation of V/R variations as
300: one-armed density waves in the disk, which are predicted for Keplerian
301: disks \citep{oka07}. Recent work based on long-baseline
302: interferometry has also provided observational evidence in support of
303: Keplerian rotation~\citep{Meilland07}.
304:
305:
306: The disk density model of~\citet{Sigut07} requires only a few input
307: stellar and disk parameters. It is assumed that the density in the
308: equatorial plane of the disk is given by an $R^{-n}$ power-law in
309: radial distance (where $R$ is the distance from the star's rotation
310: axis), and that the density distribution vertical to the equatorial
311: plane (in the $Z$ direction) is set by the hydrostatic equilibrium
312: established by the gas pressure gradient and the vertical component of
313: the star's gravitational acceleration. Such a model generally
314: produces a thin disk in which $Z/R\ll 1$, and in this case the form of
315: the density distribution is particularly simple:
316: \begin{equation}
317: \label{eq:rhoTo}
318: \rho(R,Z) = \rho_0 \left(\frac{R_{\star}}{R}\right)^{n}
319: e^{-\left(\frac{Z}{H}\right)^2} ,
320: \end{equation}
321: where $\rho_0$ is the density at the inner edge of the disk in the
322: equatorial plane, $n$ is the index in the radial power-law, and $H$ is
323: the scale height in the $Z$ direction and is given by
324: \begin{equation}
325: H = \sqrt{\frac{2R^3}{\alpha_0}},
326: \end{equation}
327: with the parameter $\alpha_0$ of the form:
328: \begin{equation}
329: \alpha_0 = GM_{\star} \frac{\mu_0 m_{\rm H}}{kT_0} .
330: \end{equation}
331: In these expressions, $M_{\star}$ and $R_{\star}$ are the stellar mass
332: and radius, respectively; $T_0$ and $\mu_0$ are the assumed
333: (vertically isothermal) temperature and mean-molecular weight at the
334: radial distance $R$. This simple, analytical form for the density is
335: possible because it is assumed that the vertical pressure scale height
336: at each $R$ can be represented by a single temperature $T_0$.
337: Although this assumption might not be directly applicable to dense
338: disks where strong vertical temperature gradients in the disk might be
339: present, \citet{SMJ07} examined the accuracy of this assumption in
340: detail (in pure hydrogen disks) and found that in most cases, the
341: differences in typical predicted diagnostics, such as the H$\alpha$
342: profiles and IR excesses, are not large for models in consistent
343: radiative and (vertical) hydrostatic equilibrium if the
344: density-averaged disk temperature is used for the parameter $T_0$. The
345: largest variations are found for the densest models (with largest
346: $\rho_0$) as these models develop a cool, equatorial zone close to the
347: star. Hence for this current work, we have adopted the simpler density
348: structure given by equation~\ref{eq:rhoTo} with the parameter $T_0$
349: being constant for all values of $R$ and $Z$. In the case of models
350: constructed for $\chi$~Oph we set $T_0$ at 10,500~K based on typical
351: density-weighted disk temperatures obtained for models in reasonable
352: agreement with the observations of $\chi$~Oph.
353:
354: With this model for the disk gas density the parameters $\rho_0$ and
355: $n$ are free to be varied to match the interferometric visibilities
356: and H$\alpha$ line profile of $\chi\;$Oph. Before we can obtain a
357: self-consistent solution for a disk model with specific $\rho_0$ and
358: $n$ values, we need to adopt the parameters for the central star. The
359: reported $T_{\rm eff}$ values for $\chi$~Oph in the literature range
360: from the low of 18,000~K to a high of
361: 29,600~K~\citep{Goraya84,Waters86,Fremat05, Zorec05}. Furthermore,
362: \citet{Goraya84} not only detected variations in visual magnitude of
363: $\chi$~Oph but also reported that such variations were accompanied by
364: changes in the slope of the continuum, resulting in variations of
365: derived $T_{\rm eff}$ values by more than 4,000~K. Because we do not
366: attempt to model the intrinsic variability of the source in this
367: study, we adopt $T_{\rm eff}$ based on the spectral type of
368: $\chi$~Oph, a main sequence B2 star, which based on tabulations
369: of~\citet{deJager87} gives a $T_{\rm eff}$ of 20,900~K. The stellar
370: mass and radius are less critical parameters for the disk models and
371: we adopted the average values based on the spectral type tabulated by
372: \citet{AQ} of $10.9 M_{\sun}$ and $5.7 R_{\sun}$. The model input
373: parameters for the stellar component are listed in
374: Table~\ref{tab:parameters}.
375:
376:
377: Figure~\ref{fig:tempplot} shows the thermal structure of a disk model
378: with $n=2.5$ and $\rho_0 = 2 \times 10^{-11}$ g cm$^{-3}$ as a
379: function of radial distance from the central star and distance from
380: the equatorial plane. Although the computational grid extends out to
381: 285~$R_{\sun}$~(50 $R_{\star}$) from the central star, the figure
382: shows only the region out to 200~$R_{\sun}$. There is a conspicuous
383: cooler region in the equatorial plane of the disk within the first
384: $\sim 50 R_{\sun}$ from the central star. This cooler volume
385: represents the region of the disk where the density and the
386: corresponding optical depths are larger. The disk temperature reaches
387: a minimum of $\approx 6,000$~K in this volume. The density
388: perpendicular to the equatorial plane falls off approximately
389: exponentially, so regions surrounding this cool volume have
390: significantly lower densities and are optically thin resulting in
391: greater temperatures. These features in the disk thermal structure
392: are typical of other models presented in the literature for Be stars
393: with moderate to high disk densities \citep[for example see][]{jon04,
394: Carciofi06, Sigut07}. The overall average temperature of the disk
395: over the computation grid is $\approx 10,400$~K, which is consistent
396: with our adopted temperature used in the vertical hydrostatic
397: equilibrium calculations (recall $T_0$). It should also be noted that
398: the temperature structure shown in Figure~\ref{fig:tempplot} is
399: influenced by the incoming radiation from the central star and
400: therefore ultimately the thermal structure will be affected by the
401: choice of stellar parameters, such as $T_{\rm eff}$ and $R_{\star}$.
402:
403:
404:
405: \section{Constraining the Model}
406:
407: \subsection{Synthetic H$\alpha$ Image}
408: \label{sec:image}
409:
410: A self-consistent temperature solution for the disk density model
411: (like that shown in Fig.~\ref{fig:tempplot}) can be used to directly
412: compute the specific intensity of radiation emitted perpendicular to
413: the plane of the disk. As the hydrogen level populations are found
414: naturally as part of the thermal solution, the H$\alpha$ line
415: emissivity and opacity,
416: \begin{equation}
417: \label{eq:emission}
418: \eta^{32}_{\nu} = \frac{h\nu_{23}}{4\pi} N_3\,A_{32}\,\phi_{\nu} \,,
419: \end{equation}
420: and
421: \begin{equation}
422: \label{eq:absorption}
423: \chi^{23}_{\nu} = \frac{h\nu_{23}}{4\pi} (N_2\,B_{23}-N_3\,B_{32})\,\phi_{\nu} \,,
424: \end{equation}
425: are known at each $(R,Z)$ location in the disk. Here $N_2$ and $N_3$
426: are the hydrogen number densities in levels 2 and 3 respectively,
427: $B_{23}$, $B_{32}$ and $A_{32}$ are the Einstein probability
428: coefficients for H$\alpha$, and $\phi_{\nu}$ is the H$\alpha$ line
429: profile. For the latter, we have adopted the routines of
430: \citet{Barklem03} which include the contributions of (thermal) Doppler
431: broadening and collisional broadening (including the linear Stark
432: effect). Turbulent (or microturbulent) broadening was not considered.
433:
434: To form the total opacity and emissivity at the frequency of
435: H$\alpha$, we have added the continuous free-free and bound-free
436: emissivity and opacity of hydrogen to equations~\ref{eq:emission}
437: and~\ref{eq:absorption}. We have also included electron scattering as
438: a coherent source of opacity but not as a source of emissivity. In a
439: radiative equilibrium (or thermal-balance) calculation it is not usual
440: to include continuum scattering as it leads to no net energy gain or
441: loss by the gas. However, non-coherent electron scattering is capable
442: of broadening spectral lines and blurring continuum
443: edges~\citep{Rybicki94}, and this process could potentially effect the
444: H$\alpha$ EW measure. Although these effects are ignored in the
445: current study, non-coherent electron scattering will be included in
446: future work. Given the opacity and emissivity, we solve the equation
447: of radiative transfer at each radial distance $R$ in the disk along a
448: ray in the $Z$-direction, i.e.,
449: \begin{equation}
450: \frac{dI_{\nu}(R,Z)}{dZ} = \eta^{32}_{\nu} - \chi^{23}_{\nu}\,I_{\nu}(R,Z) \,.
451: \end{equation}
452: The formal solution was performed with the short-characteristics
453: method of \citet{Olson87}. This gives the specific intensity emerging
454: perpendicular to the disk and is what an external observer would see
455: if the disk was viewed at $i=0^{\circ}$.
456:
457: To obtain a profile that can be compared with observations, the
458: kinematic broadening of the disk's (assumed) Keplerian rotation must
459: be included. As we estimate that the inclination of the rotation axis
460: of $\chi\,$Oph is quite small ($i\approx\,20^{\circ}$, see below), we
461: have assumed that the specific intensity at this small inclination is
462: well approximated by the $i=0^{\circ}$ intensities computed
463: above. Given this, we simply shift the $i=0^{\circ}$ spectrum of each
464: projected area of the disk and star on the sky by its radial velocity.
465: For areas on the stellar surface, we have adopted the photospheric
466: H$\alpha$ profiles computed in LTE by the code of
467: \citet{Barklem03}. The radial velocity of each patch on the stellar
468: surface followed from the assumption of a spherical star in solid body
469: rotation with an equatorial velocity of 375~km~s$^{-1}$ (giving a
470: $v\sin i$ of 144~km~s$^{-1}$; see below). Clearly this procedure is
471: not appropriate for large inclination angles as the perpendicular rays
472: through the disk would no longer be representative of the rays in the
473: direction towards the external observer. However, direct comparison of
474: the above procedure with a complete routine that solves the transfer
475: equation along the inclined rays pointing at the observer shows that
476: the simpler procedure gives good results for small inclinations and
477: the approximate procedure is computationally much faster.
478:
479:
480: We specify more than a hundred wavelength steps to cover the 15~nm
481: wide spectral region centered at the H$\alpha$ line, which corresponds
482: to the spectral channel used to obtain the interferometric
483: observations. The integrated net intensity over that 15~nm spectral
484: window is shown in Figure~\ref{fig:radial-intensity} where clearly the
485: total intensity is dominated by the bright central star. This is
486: expected because the 15~nm wide spectral region is much wider than the
487: H$\alpha$ emission line~(recall Fig.~\ref{fig:Halpha}) and therefore
488: the H$\alpha$ channel receives significant fraction of the light from
489: the central star. In turn, this radial intensity distribution can be
490: used to construct a circularly symmetric image that corresponds to the
491: total intensity in the H$\alpha$ channel.
492:
493:
494: The relatively small $v \sin i$ value reported for $\chi$~Oph of
495: 144~km~s$^{-1}$ by \citet{Zorec05} and the lack of interferometric
496: signature that would suggest large deviations from circular
497: symmetry~(recall \S\ref{sec:interferometry}), both suggest that the
498: inclination angle of the disk to the plane of the sky is small.
499: Assuming that the star is rotating near its critical velocity of
500: 477~km~s$^{-1}$~\citep[estimated based on its spectral type; see][and
501: references therein]{Porter96}, the inclination angle $i$ must be $\sim
502: 18^{\circ}$ or more for the star not to rotate above its critical
503: velocity. As rapid rotation is a known characteristic of Be stars, we
504: adopt 20$^{\circ}$ for the inclination angle of $\chi$~Oph.
505: Furthermore, because the axial ratio produced by the projection effect
506: on geometrically thin and circularly symmetric disks scales as a
507: cosine of the inclination angle, an inclination of 20$^{\circ}$
508: produces only a $\sim$6\% departure from circular symmetry (i.e., an
509: axial ratio of 0.94). For circumstellar disks with non-negligible
510: opening angles the effect would be even smaller. However, an axial
511: ratio of 0.94 would be undetectable in our data because it affects the
512: $V^2$ values at a level much smaller than the precision of our
513: interferometric observations~(compare the dash-dotted and dotted
514: curves in Fig.~\ref{fig:chi-oph-best-fit}). Therefore, for the
515: purpose of this study we will treat the synthetic images as circularly
516: symmetric and will ignore the small projection effect on the images.
517:
518:
519: The squared visibilities from the H$\alpha$ channel measure the
520: normalized Fourier power of the source structure on the sky, and
521: therefore the Fourier transform of the model image and the
522: interferometric observations can be directly compared. To accomplish
523: this we first construct a circularly symmetric image corresponding to
524: the radial distribution of the integrated intensity over the 15~nm
525: wide spectral bandpass covered by the NPOI spectral channel containing
526: the H$\alpha$ emission line. Because the model intensities are
527: calculated for radial region extending up to 50~$R_{\star}$ from the
528: central star, we require minimum dimensions of the model image of
529: 100~$R_{\star}$ $\times$ 100~$R_{\star}$ to fully describe the output.
530: However, we extend our synthetic image to 1000~$R_{\star}$ $\times$
531: 1000~$R_{\star}$ dimensions (the outer regions are filled with zeros)
532: to increase the sampling frequency of the model in the Fourier space.
533: The image is sampled every 0.2~$R_{\star}$ and this is sufficient to
534: avoid any aliasing problems at the high spatial frequencies covered by
535: the observations. The region of interest of the synthetic image is
536: shown in Figure~\ref{fig:image}. The 2-D Fourier transform of this
537: circularly symmetric image results in a function that also has
538: circular symmetry. The normalized Fourier power of this transform can
539: then be plotted as a function of the radial spatial frequency~(shown
540: as a solid line in Fig.~\ref{fig:chi-oph-best-fit}). We use the
541: standard $\chi^2$ statistic to assess the goodness-of-fit of the model
542: to the actual data obtained on the source.
543:
544:
545: \subsection{The Parameter Space}
546: \label{sec:grid}
547:
548: We explore the parametric region of of the disk density
549: model~(eq.~\ref{eq:rhoTo}) by computing over 500 parameter pairs of
550: $n$ and $\rho_0$. We covered the range of 1.8 -- 5.3 in $n$ and
551: adjusted $\rho_0$ from a low of $1.0 \times 10^{-12}$~g~cm$^{-3}$ to a
552: high of $8.0 \times 10^{-8}$~g~cm$^{-3}$. The majority of the
553: solutions were completely inconsistent with the interferometric
554: observations, and those solutions that yielded reduced $\chi^2$ value
555: of 7 or less were concentrated in a very well defined 'valley' in the
556: $\chi^2$ space~(see Fig.~\ref{fig:chi2}).
557:
558:
559: The model that results in the lowest reduced $\chi^2$ value of 1.17
560: corresponds to $n$ of 2.5 and $\rho_0$ of $2.0 \times
561: 10^{-11}$~g~cm$^{-3}$. The thermal
562: structure~(Fig.~\ref{fig:tempplot}), H$\alpha$ model
563: image~(Fig.~\ref{fig:image}), and the model curve shown in
564: Figure~\ref{fig:chi-oph-best-fit} all correspond to this best-fit
565: model. Furthermore, assuming that the disk is viewed at an
566: inclination angle of 20$^{\circ}$~(as discussed in
567: \S~\ref{sec:image}), we obtain an excellent agreement with the
568: observed H$\alpha$ line profile obtained through spectroscopy~(see
569: Fig.~\ref{fig:Halpha}). In fact, for the disk model defined by our
570: best-fit to interferometric data, the H$\alpha$ profile can only be
571: reproduced assuming $i$ of 20$^{\circ}$. For any smaller inclination
572: the model profile is simply too sharp~(due to the decreased rotational
573: broadening) and for larger values of $i$ the line becomes too broad
574: and weaker than the actual observed profile.
575:
576:
577: In addition to the best-fit solution found at $n$ of 2.5 and $\rho_0$
578: of $2.0 \times 10^{-11}$~g~cm$^{-3}$ it is clearly evident from
579: Figure~\ref{fig:chi2} that there exists a range of parameter values in
580: the $\chi^2$ space that produce disk models that fit the observational
581: data acceptably well. However, these solutions appear to be confined
582: to a very narrow range in the $\chi^2$ space. We attribute this range
583: of solutions to the changing total H$\alpha$ flux generated by a
584: model, where most of the solutions in the upper-left part of the
585: figure produce too much H$\alpha$ flux with respect to the central
586: star and those in the lower-right part of the figure do not produce
587: enough H$\alpha$ flux. These effects can be related to different disk
588: thermal structures and different densities, which in turn will affect
589: the apparent sizes of the H$\alpha$-emitting regions. A linear
590: least-squares fit to solutions with reduced $\chi^2 < 7$ in the
591: log~$\rho_0$ versus $n$ plane yields a slope of $1.18\pm 0.03$ (and an
592: intercept of $-13.6\pm 0.1$), where the uncertainty does not account
593: for the effects of gridding or the choice of the reduced $\chi^2$
594: cutoff value. Nevertheless, this indicates that at least for a range
595: of $n$ values between 2.5 and 4.0 the disk solutions that produce
596: interferometric signatures closest to the observational data fall
597: along a relation where log~$\rho_0$ is directly proportional to $n$.
598: This is in agreement with similar conclusions made by \citet{Gies07}
599: who approximate the size of the disk on the sky using a boundary
600: between the optically thick and thin disk regions, which then leads to
601: the expectation that good fits fall along a relation where
602: log~$\rho_0$ is proportional to $n$.
603:
604:
605:
606: \section{Discussion}
607:
608: It is useful to compare our best-fit disk model for $\chi$~Oph to
609: previous results of \citet{Waters86} who modeled the IRAS IR excesses
610: of a large sample of Be stars with disk models with fixed opening
611: angles of $15^{\circ}$ and where the density varied only with the
612: distance from the stellar center, $r$, in the form of
613: \begin{equation}
614: \label{eq:rhorens}
615: \rho(r) = \rho_0' (r/R_{\star})^{-n} .
616: \end{equation}
617: Assuming an isothermal disk with a temperature of 18,000~K, a
618: temperature that is considerably larger than the $\sim$10,400~K
619: density-weighted average temperature of our best-fit model,
620: \citet{Waters86} found a radial power-law index $n$ of 2.4 and a
621: density at the base of the disk of $\log \rho_0=-11.4$. These results
622: compare well to our results of $n=2.5$ and $\log\rho_0=-10.7$, even
623: though the disk-density models used in the two studies are not the
624: same. The smaller base disk density found by \citet{Waters86} is
625: likely the result of the fact that his density distribution is
626: constant along radial arcs over the entire opening angle of the disk
627: which places more material in the disk as compared to our models.
628:
629:
630: \citet{Porter99} modeled the circumstellar regions of Be stars as
631: isothermal viscous disks and predicted $n$ of $7/2$. Viscous models
632: that drop the assumption of an isothermal disk can find a range of $n$
633: values as the disk fills or empties \citep{JSP08}. \citet{Porter99}
634: also analyzed the IR excess of $\chi$~Oph using the data of
635: \citet{Waters86}, but with a disk model density essentially equivalent
636: to ours. In the case of an isothermal disk (also set to 18,000~K),
637: \citet{Porter99} found an index $n$ of 2.2 with $\log \rho_0 =-11.2$;
638: in the case of a disk in which the density-averaged temperature was
639: allowed to vary with radial distance as a power-law, \citet{Porter99}
640: obtained $n=1.9$ and $\log(\rho_0)=-11.2$. Porter indicates that the
641: non-isothermal disk formally fits the IR data best, but that the
642: improvement is not large.
643:
644: In addition to the comparison between model and observed H$\alpha$
645: interferometric visibilities and the H$\alpha$ line profile, it is
646: also instructive to compare the predicted IR excess of our best-fit
647: model to the IR observations reported in the literature.
648: Figure~\ref{fig:IR-fit} illustrates the comparison between the SED of
649: our best-fit model and the visual and IR observations from
650: \citet{Waters86}. The overall agreement is quite reasonable,
651: especially considering that the model was not fit to the SED data,
652: although the observations in the $1-10\,\mu$m region fall below the
653: model. We should also point out that variability between the
654: observations used by \citet{Waters86} and our 2006 H$\alpha$
655: observations (interferometric and spectroscopic) cannot be ruled out.
656: For example, \citet{ban00} obtained the H$\alpha$ profile for
657: $\chi$~Oph in 1998 and reported an equivalent width of approximately
658: $-3.6$~nm, as compared to the $-7.1\pm0.2$~nm found in the present
659: work for the 2006 epoch~(recall Fig.~\ref{fig:Halpha}). Hence, the
660: H$\alpha$ emission of $\chi\,$Oph has approximately doubled in 8
661: years, and assuming that the H$\alpha$ emission was similarly weaker
662: at the time of observations of \citet{Waters86} we can then expect our
663: best-fit model to over-predict IR excess as compared to Waters'
664: results. For this reason, we feel that our model cannot be improved
665: using the IR observations that were acquired using the IRAS satellite
666: in the early 1980's.
667:
668:
669:
670: \section{Summary and Future Work}
671:
672: In this paper we presented the study of the circumstellar disk of the
673: Be star $\chi$~Oph, based on the approach of combining high-spatial
674: resolution interferometric observations with numerical disk models
675: requiring only a few input parameters. Using the interferometric data
676: we were also able to determine that the H$\alpha$-emitting disk can be
677: fit by a circular symmetric Gaussian with FWHM diameter of
678: 3.46$\pm$0.07~mas. Using the code {\sc bedisk}~\citep{Sigut07} we
679: created a grid of models with a range of inner edge densities
680: ($\rho_0$) and equatorial plane density distribution fall-off rates
681: ($n$), while keeping the mass, radius and effective temperature of the
682: central star fixed. These models were used to create synthetic
683: H$\alpha$ images by adding the emission line and continuum flux,
684: including the contribution from the central star, in a 15~nm spectral
685: region around H$\alpha$, similar to that of the H$\alpha$
686: interferometric channel. These images were then Fourier transformed
687: and compared directly to the interferometric observations, resulting
688: in the best-fit model with $n = 2.5$ and $\rho_0 = 2.0 \times
689: 10^{-11}$~g~cm$^{-3}$. The best-fit disk model was also used to
690: calculate a synthetic H$\alpha$ line profile and spectral energy
691: distribution of this system over 0.1--100~$\mu$m, which show good
692: agreement with the observed line profile and the photometric
693: observations published in the literature.
694:
695:
696: The technique demonstrated here, of combining models with
697: interferometric observations, demonstrates a new and independent
698: method for obtaining properties of circumstellar disks of Be stars.
699: This is possible because the density distribution in the disk is
700: strongly affected by the value of the disk density at the stellar
701: surface, $\rho_0$, and to a lesser extent by $n$. Variations in
702: $\rho_0$ and $n$ result in dramatic changes in the thermal structure
703: of the disk. Ultimately, the density distribution and the resulting
704: thermal structure directly affect the H$\alpha$ flux predicted by the
705: theoretical models, which in turn can be constrained by observations
706: that spatially resolve the region. In the future, we plan to
707: investigate more sophisticated forms of the model density distribution
708: within Be star disks, such as rotating density perturbation models
709: \citep[see for example][]{wis07}, and density distributions that are
710: consistent with hydrodynamic simulations.
711:
712:
713: \acknowledgements
714:
715: The Navy Prototype Optical Interferometer is a joint project of the
716: Naval Research Laboratory and the U.S. Naval Observatory, in
717: cooperation with Lowell Observatory, and is funded by the Office of
718: Naval Research and the Oceanographer of the Navy. We thank Doug Gies
719: for the very helpful suggestions on how to improve this manuscript.
720: C.~T. thanks Lowell Observatory for the generous telescope time
721: allocation on the John S. Hall Telescope. C.~T. would also like to
722: thank Erika Grundstrom for useful discussions and thanks Nick Melena
723: who contributed to the reductions of interferometric observations.
724: C.~E.~J. and T.~A.~A.~S. would like to acknowledge support from the
725: Natural Sciences and Engineering Research Council of Canada.
726:
727:
728: {\it Facilities:} \facility{NPOI}, \facility{LO:42in}
729:
730:
731:
732: %% The BIBLIOGRAPHY
733:
734:
735: %\normalsize
736:
737: \begin{thebibliography}{}
738: \bibitem[Abt \& Levy(1978)]{Abt78} Abt, H.~A., \& Levy, S.~G.\ 1978,
739: \apjs, 36, 241
740: \bibitem[Banerjee et al.(2000)]{ban00} Banerjee, D.\ P.\ K., Rawat, S.\ D., \&
741: Janardhan, P.\ 2000, A\&AS 147, 229
742: \bibitem[Barklem \& Piskunov(2003)]{Barklem03} Barklem, P.~S., \&
743: Piskunov, N.\ 2003, Modeling of Stellar Atmospheres, 210, 28P
744: \bibitem[Carciofi \& Bjorkman(2006)]{Carciofi06} Carciofi, A.~C., \&
745: Bjorkman, J.~E.\ 2006, \apj, 639, 1081
746: \bibitem[Cot\'{e} \& Waters(1987)]{Cote87} Cot\'{e}, J., \& Waters,
747: L. B. F. M. 1987, \aap, 176, 93
748: \bibitem[Cox(2000)]{AQ} Cox, A.~N.\ 2000, Allen's astrophysical
749: quantities, 4th ed.~Publisher: New York: AIP Press; Springer, 2000
750: \bibitem[de Jager \& Nieuwenhuijzen(1987)]{deJager87} de Jager, C., \&
751: Nieuwenhuijzen, H.\ 1987, \aap, 177, 217
752: \bibitem[Dougherty(1994)]{dou94} Dougherty, S.M.\ et al., 1994, A\&A
753: 290, 609
754: \bibitem[Fr{\'e}mat et al.(2005)]{Fremat05} Fr{\'e}mat, Y., Zorec, J.,
755: Hubert, A.-M., \& Floquet, M.\ 2005, \aap, 440, 305
756: \bibitem[Gehrz et al.(1974)]{Gehrz74} Gehrz, R.~D., Hackwell, J.~A.,
757: \& Jones, T.~W.\ 1974, \apj, 191, 675
758: \bibitem[Gies et al.(2007)]{Gies07} Gies, D.~R., et al.\ 2007, \apj,
759: 654, 527
760: \bibitem[Goraya(1984)]{Goraya84} Goraya, P.~S.\ 1984, \aap, 138, 19
761: \bibitem[Grundstrom \& Gies(2006)]{Grundstrom06} Grundstrom, E.~D., \&
762: Gies, D.~R.\ 2006, \apjl, 651, L53
763: \bibitem[Hall et al.(1994)]{Hall94} Hall, J. C., Fulton, E. E.,
764: Huenemoerder, D. P., Welty, A. D., \& Neff, J. E. 1994, \pasp, 106,
765: 315
766: \bibitem[Hanbury Brown et al.(1974)]{Hanbury74} Hanbury Brown, R.,
767: Davis, J., \& Allen, L.~R.\ 1974, \mnras, 167, 121
768: \bibitem[Harmanec(1987)]{Harmanec87} Harmanec, P.\ 1987, Bulletin of
769: the Astronomical Institutes of Czechoslovakia, 38, 283
770: \bibitem[Hummel et al.(1998)]{Hummel98} Hummel, C. A., Mozurkewich, D., Armstrong,
771: J. T., Hajian, a. R., Elias II, N. M., \& Hutter, D. J. 1998, \aj, 116, 2536
772: \bibitem[Hummel(2000)]{hum00a} Hummel, W., in {\it The Be Phenomena in
773: Early-Type Stars}, M.\ A.\ Smith, H.\ F.\ Henrichs, \& J.\ Fabregat
774: (eds), 2000, ASP Conf.\ Ser.\ 214, 396
775: \bibitem[Hummel \& Vrancken(2000)]{hum00b} Hummel, W., \& Vrancken, M
776: 2000, \aap, 359, 1075
777: \bibitem[Hummel et al.(2003)]{Hummel03} Hummel, C. A., et al. 2003a,
778: \aj, 125, 2630
779: \bibitem[Hutter et al.(2004)]{Hutter04} Hutter, D.~J., Benson, J.~A.,
780: Zavala, R.~T., Johnston, K.~J., Pauls, T.~A., Hummel, C.~A., \&
781: Armstrong, J.~T.\ 2004, \procspie, 5491, 73
782: \bibitem[Jones, Sigut \& Marlborough(2004)]{jon04} Jones, C. E.,
783: Sigut, T. A. A., \& Marlborough, J. M. 2004, \mnras, 352, 841
784: \bibitem[Jones et al.(2008)]{JSP08} Jones, C.\ E., Sigut, T.\ A.\ A.,
785: \& Porter, J.\ M.\ 2008, \mnras, 386, 1922
786: \bibitem[Kastner \& Mazzali(1989)]{Kastner89} Kastner, J.~H., \&
787: Mazzali, P.~A.\ 1989, \aap, 210, 295
788: \bibitem[Levato et al.(1987)]{Levato87} Levato, H., Malaroda, S.,
789: Morrell, N., \& Solivella, G.\ 1987, \apjs, 64, 487
790: \bibitem[Marlborough(1969)]{Marlborough69} Marlborough, J.~M.\ 1969,
791: \apj, 156, 135
792: \bibitem[Meilland et al.(2007)]{Meilland07} Meilland, A., et al.\
793: 2007, \aap, 464, 59
794: \bibitem[Millar \& Marlborough(1998)]{Millar98} Millar, C. E., \&
795: Marlborough, J. M. 1998, \apj, 494, 715, MM
796: \bibitem[Millar \& Marlborough(1999)]{Millar99} Millar, C.~E., \&
797: Marlborough, J.~M.\ 1999, \apj, 526, 400
798: \bibitem[Okazaki(2007)]{oka07} Okazaki, A.~T.\ 2007, Active OB-Stars:
799: Laboratories for Stellare and Circumstellar Physics, Edited by
800: S. Stefl, S. P. Owocki, and A. T. Okazaki., 361, 230
801: \bibitem[Olson \& Kunasz(1987)]{Olson87} Olson, G.~L., \& Kunasz,
802: P.~B.\ 1987, Journal of Quantitative Spectroscopy and Radiative
803: Transfer, 38, 325
804: \bibitem[Perryman et al.(1997)]{Perryman97} Perryman, M.~A.~C., et
805: al.\ 1997, \aap, 323, L49
806: \bibitem[Porter(1996)]{Porter96} Porter, J.~M.\ 1996, \mnras, 280, L31
807: \bibitem[Porter(1999)]{Porter99} Porter, J.~M.\ 1999, \aap, 348, 512
808: \bibitem[Quirrenbach et al.(1997)]{Quirrenbach97} Quirrenbach, A., et
809: al.\ 1997, \apj, 479, 477
810: \bibitem[Rybicki \& Hummer(1994)]{Rybicki94} Rybicki, G.~B.,
811: \& Hummer, D.~G.\ 1994, \aap, 290, 553
812: \bibitem[Sigut \& Jones(2007)]{Sigut07} Sigut, T. A. A. \& Jones,
813: C. E. 2007, \apj, 668, 481
814: \bibitem[Sigut et al.(2007)]{SMJ07} Sigut, T.\ A.\ A., McGill, M.\ A.,
815: \& Jones, C.\ E.\ 2007, ApJ, submitted
816: \bibitem[Tycner et al.(2003)]{Tycner03} Tycner, C., Hajian, A. R.,
817: Mozurkewich, D., Armstrong, J. T., Benson, J. A., Gilbreath, G. C.,
818: Hutter, D. J., Pauls, T. A., \& Lester, J. B. 2003, \aj, 125, 3378
819: \bibitem[Tycner et al.(2005)]{Tycner05}Tycner, C. et al. 2005, \apj,
820: 624, 359
821: \bibitem[Tycner et al.(2006a)]{Tycner06a} Tycner, C., Benson, J.~A.,
822: Hutter, D.~J., Schmitt, H.~R., \& Zavala, R.~T.\ 2006a, \procspie,
823: 6268, 49
824: \bibitem[Tycner et al.(2006b)]{Tycner06b} Tycner, C., et al.\ 2006b,
825: \aj, 131, 2710
826: \bibitem[Waters(1986)]{Waters86} Waters, L. B. F. M. 1986 \aap, 162, 121
827: \bibitem[Waters et al.(1987)]{wat87} Waters, L. B. F. M., Cote, J., \&
828: Lamers, H. J. G. L. M. 1987, \aap, 185, 206
829: \bibitem[Wisniewski et al.(2007)]{wis07} Wisniewski, J. P., Kowalski,
830: A. F., Bjorkman, K. S., Bjorkman, J. E., \& Carciofi, A. C. 2007,
831: \apj, 656, L21
832: \bibitem[Wood et al.(1997)]{woo97} Wood K., Bjorkman K. S., \&
833: Bjorkman J. E. 1997, \apj, 477, 926
834: \bibitem[Zorec et al.(2005)]{Zorec05} Zorec, J., Fr{\'e}mat, Y., \&
835: Cidale, L.\ 2005, \aap, 441, 235
836: \end{thebibliography}
837:
838:
839: %% TABLES %%
840:
841: \clearpage
842:
843:
844:
845: \begin{table}[htp]
846: \caption[]{\sc \small NPOI Observations of $\chi$~Oph}
847: \label{tab:obs}
848: \begin{tabular}{lcc} \hline\hline
849: \hspace{2cm} UT Date \hspace{2cm} & \# of Scans & \# of Baselines \\ \hline
850: 2006 June 11 \dotfill & 5 & 5 \\
851: 2006 June 13 \dotfill & 4 & 5 \\
852: 2006 June 17 \dotfill & 6 & 5 \\
853: 2006 June 18 \dotfill & 7 & 5 \\
854: \hline
855: \end{tabular}
856: \end{table}
857:
858:
859: \begin{table}[htp]
860: \caption[]{\sc \small Calibrated H$\alpha$ Squared Visibilities of $\chi$~Oph}
861: \label{tab:v2data}
862: \begin{tabular}{crrcc} \hline\hline
863: Julian Date & Spatial Frequency $u$ & Spatial Frequency $v$ & & \\
864: (JD $-$ 2,450,000) & ($10^6$ cycles/radian) & ($10^6$ cycles/radian) & $V_{\rm{H}\alpha}^2$ & Baseline$^{\dag}$ \\ \hline
865: 3897.785 & $ 27.003$ & $ -7.122$ & 0.749 $\pm$ 0.048 & AC-AE \\
866: 3897.785 & $ 64.880$ & $ 23.478$ & 0.534 $\pm$ 0.032 & W7-AC \\
867: 3897.785 & $ 91.883$ & $ 16.357$ & 0.502 $\pm$ 0.023 & W7-AE \\
868: 3897.785 & $ -29.723$ & $ -8.345$ & 0.688 $\pm$ 0.061 & AC-AW \\
869: 3897.785 & $ 64.989$ & $ 23.518$ & 0.519 $\pm$ 0.027 & W7-AC \\
870: 3897.785 & $ 35.266$ & $ 15.173$ & 0.680 $\pm$ 0.030 & W7-AW \\
871: 3897.801 & $ 27.324$ & $ -7.970$ & 0.696 $\pm$ 0.042 & AC-AE \\
872: 3897.801 & $ 61.798$ & $ 21.501$ & 0.548 $\pm$ 0.045 & W7-AC \\
873: 3897.801 & $ 89.123$ & $ 13.531$ & 0.520 $\pm$ 0.026 & W7-AE \\
874: 3897.801 & $ -28.537$ & $ -7.436$ & 0.724 $\pm$ 0.043 & AC-AW \\
875: 3897.801 & $ 61.902$ & $ 21.537$ & 0.511 $\pm$ 0.027 & W7-AC \\
876: 3897.801 & $ 33.365$ & $ 14.101$ & 0.681 $\pm$ 0.037 & W7-AW \\
877: \hline
878: \end{tabular}\\[0.9ex]
879: \parbox{6.0in}{\footnotesize \quad {\sc Note.} ---
880: Table~\ref{tab:v2data} is published in its entirety in the electronic
881: edition of the Astrophysical Journal. $^{\dag}$ There are 6 baselines
882: per scan (same JD) with the W7--AC measured at two output beams with
883: slightly different wavelength scales and thus resulting in slightly
884: different $u$ and $v$ values for the same baseline length. }
885: \end{table}
886:
887:
888:
889: \begin{table}[htp]
890: \caption[]{\sc \small Model Stellar Parameters}
891: \label{tab:parameters}
892: \begin{tabular}{lccl} \hline\hline
893: \hspace{1.7cm} Parameter \hspace{1.7cm} & Symbol & Value & Reference \\ \hline
894: Mass ($M_\odot$) \dotfill & $M_{\star}$ & 10.9 & \citet{AQ} \\
895: Radius ($R_\odot$) \dotfill & $R_{\star}$ & 5.7 & \citet{AQ} \\
896: Effective temperature (K) \dotfill & $T_{\rm eff}$ & 20,900 & \citet{deJager87} \\
897: Luminosity ($L_\odot$) \dotfill & $L_{\star}$ & $5.6\times 10^3$ & calculated$^{\dag}$ \\
898: Surface gravity \dotfill & $\log g$ & $4.0$ & calculated$^{\ddag}$ \\
899: \hline
900: \end{tabular}\\[0.9ex]
901: \parbox{6.0in}{\footnotesize \quad {\sc Note.} --- $^{\dag}$ $L_\star
902: = 4\pi R_{\star}^2 \sigma T_{\rm eff}^4$ where $\sigma$ is the
903: Stefan-Boltzmann constant. $^{\ddag}$ $\log g = \log (GM_\star
904: R_\star^{-2})$ where $G$ is the gravitational constant. }
905: \end{table}
906:
907: \clearpage
908:
909:
910: %% FIGURES
911:
912: \begin{figure}
913: \plotone{f1.eps}
914: \caption{The ($u$,$v$)-plane coverage of the interferometric
915: observations of $\chi$~Oph obtained in the H$\alpha$ channel with five
916: unique baselines from four nights of observations~({\it circles}).
917: The dotted-lines indicate possible coverage from meridian to 6 hr east
918: and the solid-lines indicate possible coverage from the meridian to 6
919: hr west.}
920: \label{fig:chi-oph-uv}
921: \end{figure}
922:
923:
924: \begin{figure}
925: \plotone{f2.eps}
926: \caption{ Interferometric data obtained from the spectral channel
927: containing the H$\alpha$ emission line from the circumstellar disk of
928: $\chi$~Oph. The 0.52~AU diameter (FWHM) circumstellar disk is fully
929: resolved, whereas the central star is assumed to have an uniform disk
930: diameter of 0.35~mas ({\it dashed line}). The best-fit model~({\it
931: solid line}) obtained by taking the Fourier transform of the synthetic
932: image from Figure~\ref{fig:image} is shown along with a circularly
933: symmetric Gaussian model with FWHM diameter of 3.46~mas ({\it
934: dash-dotted line}). The effect of an axial ratio of 0.94 on the
935: Gaussian model is also shown ({\it dotted line}). }
936: \label{fig:chi-oph-best-fit}
937: \end{figure}
938:
939: \begin{figure}
940: \plotone{f3.eps}
941: \caption{The H$\alpha$ emission line of $\chi$~Oph observed on
942: 2006~June~10 ({\it solid-line}) and the synthetic H$\alpha$ profile
943: calculated from the best-fit disk model ({\it dashed-line}). The
944: model profile has been obtained for an inclination angle $i$ of
945: 20$^{\circ}$ and has been broadened by a Gaussian kernel to match the
946: resolving power~($R$ = 10,000) of the spectroscopic observations. The
947: observed H$\alpha$ profile has an EW of $-7.1\pm 0.2$~nm and the
948: theoretical profile has EW of $-6.6$~nm.}
949: \label{fig:Halpha}
950: \end{figure}
951:
952:
953: \begin{figure}
954: \plotone{f4.eps}
955: \caption{The thermal structure of the disk model of $\chi$~Oph
956: obtained using $n = 2.5$, $\rho_0 = 2.0 \times 10^{-11}$ g cm$^{-3}$
957: and the stellar parameters listed in Table~\ref{tab:parameters}. The
958: thermal structure is shown as a function of radial distance~($R$) from
959: the center of the star and the vertical distance~($Z$) above and below
960: the equatorial plane. }
961: \label{fig:tempplot}
962: \end{figure}
963:
964:
965: \begin{figure}
966: \plotone{f5.eps}
967: \caption{Integrated model intensity for the direction normal to the
968: plane of the disk in a 15~nm wide spectral region centered at
969: H$\alpha$ ({\it solid line}) for the best-fit model of $n = 2.5$ and
970: $\rho_0 = 2.0 \times 10^{-11}$ g cm$^{-3}$. The Gaussian radial
971: intensity distribution that reproduces the interferometric signature
972: in Fig.~\ref{fig:chi-oph-best-fit} is also shown~({\it dashed line}).}
973: \label{fig:radial-intensity}
974: \end{figure}
975:
976: \begin{figure}
977: \plotone{f6.eps}
978: \caption{Synthetic image of the best-fit disk model viewed pole on.
979: Only intensities in a 15~nm spectral window centered at H$\alpha$
980: emission line were used to construct this synthetic image. The
981: surface brightness of the central star is derived using the stellar
982: atmosphere model integrated over the same 15~nm spectral window as
983: used to calculate the net intensity from the disk.}
984: \label{fig:image}
985: \end{figure}
986:
987:
988: \begin{figure}
989: \plotone{f7.eps}
990: \caption{Contour plot showing the reduced $\chi^2$ value as a function
991: of $\rho_0$~(g cm$^{-3}$) and $n$. Only a region of reduced $\chi^2 <
992: 7$ is shown and the best-fit model with reduced $\chi^2$ of 1.17 is
993: marked with a plus sign. Some of the gaps in the linear trend are
994: caused by the grid spacing and the limitations of the contour plotting
995: routine dealing with unequal grid spacing. The approximate boundary
996: of the region of parameter space sampled by our models is also
997: indicated~({\it dotted lines}), along with a linear fit~({\it dashed
998: line}) to the models with reduced $\chi^2 < 7$ in the log~$\rho_0$ and
999: $n$ plane~(see \S~\ref{sec:grid}). }
1000: \label{fig:chi2}
1001: \end{figure}
1002:
1003:
1004: \begin{figure}
1005: \plotone{f8.eps}
1006: \caption{ The spectral energy distribution (SED) of our best-fit model
1007: corresponding to $n = 2.5$ and $\rho_0 = 2.0 \times 10^{-11}$ g
1008: cm$^{-3}$ (solid-line), which consists of the sum of the stellar
1009: contribution~(dotted-line) and IR excess produced by the circumstellar
1010: disk. The model SED is compared to visual, near-IR and far-IR
1011: observations reported in literature \citep[see][and references
1012: therein]{Waters86}. }
1013: \label{fig:IR-fit}
1014: \end{figure}
1015:
1016:
1017: \end{document}
1018: