astro-ph0009502/ms.tex
1: %%
2: %% Beginning of file 'sample.tex'
3: %%
4: %% This is a sample manuscript marked up using the
5: %% AASTeX v5.0 LaTeX 2e macros.
6: 
7: %% The first piece of markup in an AASTeX v5.0 document
8: %% is the \documentclass command. LaTeX will ignore
9: %% any data that comes before this command.
10: 
11: %% The command below calls the default manuscript style,
12: %% which will produce a double-spaced document on one column.
13: %% Examples of commands for other substyles follow. Use
14: %% whichever is most appropriate for your purposes.
15: 
16: %%\documentclass{aastex}
17: 
18: %% preprint produces a one-column, single-spaced document:
19: 
20: \documentclass[preprint]{aastex}
21: %%\usepackage{psfig}
22: 
23: %% preprint2 produces a double-column, single-spaced document:
24: 
25: %% \documentclass[preprint2]{aastex}
26: 
27: %% If you want to create your own macros, you can do so
28: %% using \newcommand. Your macros should appear before
29: %% the \begin{document} command.
30: %%
31: %% If you are submitting to a journal that translates manuscripts
32: %% into SGML, you need to follow certain guidelines when preparing
33: %% your macros. See the AASTeX v5.0 Author Guide
34: %% for information.
35: 
36: %%\newcommand{\vdag}{(v)^\dagger}
37: %%\newcommand{\myemail}{skywalker@galaxy.far.far.away}
38: 
39: %% You can insert a short comment on the title page using the command below.
40: 
41: \slugcomment{To appear in ApJ}
42: 
43: %% If you wish, you may supply running head information, although
44: %% this information may be modified by the editorial offices.
45: %% The left head contains a list of authors,
46: %% usually a maximum of three (otherwise use et al.).  The right
47: %% head is a modified title of up to roughly 44 characters.  Running heads
48: %% will not print in the manuscript style.
49: 
50: \shorttitle{Modeling Nova Cygni 1992}
51: \shortauthors{Short, {\it et al.}}
52: 
53: 
54: %% This is the end of the preamble.  Indicate the beginning of the
55: %% paper itself with \begin{document}.
56: 
57: \begin{document}
58: 
59: %% LaTeX will automatically break titles if they run longer than
60: %% one line. However, you may use \\ to force a line break if
61: %% you desire.
62: 
63: \title{Non-LTE Modeling of Nova Cygni 1992}
64: 
65: 
66: %% Use \author, \affil, and the \and command to format
67: %% author and affiliation information.
68: %% Note that \email has replaced the old \authoremail command
69: %% from AASTeX v4.0. You can use \email to mark an email address
70: %% anywhere in the paper, not just in the front matter.
71: %% As in the title, you can use \\ to force line breaks.
72: 
73: \author{C. Ian Short}
74: \affil{Department of Physics, Florida Atlantic University,
75:     Boca Raton, FL 33431-0991}
76: 
77: \and
78: 
79: \author{Peter H. Hauschildt}
80: \affil{Department of Physics and Astronomy and Center for
81: Simulational Physics, University of Georgia,
82:     Athens, GA 30602-2451}
83: 
84: \and
85: 
86: \author{S. Starrfield}
87: \affil{Department of Physics and Astronomy, Arizona State University, P.O. Box 871504,
88: Tempe, AZ 85287-1504 } 
89: 
90: \and
91: 
92: \author{E. Baron}
93: \affil{Dept. of Physics and Astronomy, University of Oklahoma,
94:     440 W. Brooks, Rm 131, Norman, OK 73091-0225}
95: 
96: %% Notice that each of these authors has alternate affiliations, which
97: %% are identified by the \altaffilmark after each name.  Specify alternate
98: %% affiliation information with \altaffiltext, with one command per each
99: %% affiliation.
100: 
101: %% Mark off your abstract in the ``abstract'' environment. In the manuscript
102: %% style, abstract will output a Received/Accepted line after the
103: %% title and affiliation information. No date will appear since the author
104: %% does not have this information. The dates will be filled in by the
105: %% editorial office after submission.
106: 
107: \begin{abstract}
108: 
109:    We present a grid of nova models that have an extremely large number
110: of species treated in NLTE, and apply it to the analysis of an extensive time
111: series of ultraviolet spectroscopic data for Nova Cygni 1992.  We use ultraviolet 
112: colors to derive the time development of
113: the effective temperature of the expanding atmosphere during the fireball phase
114: and the first ten days of the optically thick wind phase.  We find that the
115: nova has a pure optically thick wind spectrum until about 10 days after the explosion.
116: During this interval, we find that synthetic
117: spectra based on our derived temperature sequence agree very well with the observed spectra.
118: We find that a sequence
119: of hydrogen deficient models provides an equally good fit providing the
120: model effective temperature is shifted upwards by $\sim 1000$ K.  We find that 
121: high resolution UV spectra of the optically thick wind phase are fit moderately
122: well by the models.   We find that a high 
123: resolution spectrum of the {\it fireball} phase is better fit by a model with a steep density
124: gradient, similar to that of a supernova, than by a nova model. 
125:     
126: \end{abstract}
127: 
128: %% Keywords should appear after the \end{abstract} command. The uncommented
129: %% example has been keyed in ApJ style. See the instructions to authors
130: %% for the journal to which you are submitting your paper to determine
131: %% what keyword punctuation is appropriate.
132: 
133: \keywords{novae, cataclysmic variables --- stars: atmospheres --- stars: individual (Nova Cygni 1992, V1974 Cygni) --- ultraviolet: stars}
134: 
135: %% From the front matter, we move on to the body of the paper.
136: %% In the first two sections, notice the use of the natbib \citep
137: %% and \citet commands to identify citations.  The citations are
138: %% tied to the reference list via symbolic KEYs. The KEY corresponds
139: %% to the KEY in the \bibitem in the reference list below. We have
140: %% chosen the first three characters of the first author's name plus
141: %% the last two numeral of the year of publication as our KEY for
142: %% each reference.
143: 
144: \section{Introduction}
145: 
146:     Energetic stellar explosions such as novae and supernovae provide
147: us with the opportunity to study the physics of rapidly expanding gas shells.  Due to
148: the rapid response of observers and the density of observational coverage in time,   
149: the classical \ion{ONeMg}{0} nova Cygni 92 (V1974 Cygni) \citep{collins} is the best observed nova in 
150: history.  The existence
151: of a frequent, well spaced set of low and high resolution International Ultraviolet Explorer (IUE)
152: spectra allow
153: us to model the time development of the expanding gas shell in a spectral region
154: where the object emits most of its flux during the fireball and optically thick wind 
155: phases of the explosion.  The gradual emergence of flux in this region, caused by the lifting 
156: of heavy line
157: blanketing (``the iron curtain'' (\citet{ncyg1}, \citet{ncyg2}),
158: determines the shape of the early UV light curve, and is a particularly dramatic
159: example of how these explosions serve as laboratories for the study of 
160: a radiating plasma under changing conditions.
161:  
162: \paragraph{}
163:   
164:    Among stellar atmospheric modeling problems, that of novae is especially complicated due
165: to the steepness of the $T$ and $\rho$ gradients, the large depth of translucency, the sphericity, 
166: and the significance of first order special relativistic effects in the transfer of radiation.
167: The first two of these conditions cause the line forming region (LFR) to span a wide range
168: of physical conditions so that many overlapping line and continuum transitions of many ionization 
169: stages of many elements are simultaneously present in the emergent spectrum.  The second condition
170: ensures that many species in the LFR will deviate significantly from Local Thermodynamic 
171: Equilibrium (LTE) \citep{short1}.  As a result, the accurate modeling of data such as 
172: that obtained for Cygni 92 requires a very general treatment that incorporates non-LTE (NLTE) effects 
173: as accurately and completely as possible.  The multi-purpose atmospheric modeling code {\tt PHOENIX}
174: \citep{hausphx} was developed to model, among other things, nova and supernova explosions.
175: {\tt PHOENIX} makes use of a fast and accurate Operator Splitting/Accelerated Lambda Iteration
176: (OS/ALI) scheme to solve self-consistently the first order Special Relativistic radiative
177: transfer equation and the NLTE rate equations for many species and overlapping transitions
178: \citep{hausphx}.  Cygni 92 was first modeled with {\tt PHOENIX} by \citet{ncyg2}.  However,
179: recently \citet{short1} have greatly
180: increased the number of species and ionization stages treated in 
181: detailed NLTE by {\tt PHOENIX}.  The lowest six ionization stages of the 19 most
182: important elements are now treated in NLTE.  The purpose of this study is to calculate 
183: a set of improved nova models and synthetic spectra with the expanded version of {\tt PHOENIX}, 
184: to compare them to the IUE time series spectra of Cygni 92, and to extract stellar 
185: parameters at each observed time. 
186:  
187: \section{Observations}\label{obs}
188: 
189:   Table 1 is a log of the observations that we have extracted from the IUE archive for
190: both the low and high resolution spectrographs and for both the long and short wavelength cameras 
191: (LWP and SWP).  Superscripts in Table 1 indicate approximately coincident pairs
192: of SWP and LWP spectra.  The spectra span a range of 144.75 days, from the initial
193: fireball stage of the explosion, through the optically thick wind phase, to the nebular 
194: stage.   Fig. \ref{surf_lo} shows the fourteen pairs of approximately coincident SWP and LWP
195: spectra as a function of time.  We have discarded the portion of the SWP spectrum that lies below 
196: $\lambda 1250$ \AA~ due to the noisiness of the signal and contamination by geocoronal Lyman $\alpha$, and the
197: portion of the LWP spectrum that lies below $\lambda 2400$ due to severe degradation of the signal.  The gap in the
198: $\lambda$ dimension arises because the reduced SWP and LWP spectra are disjoint in $\lambda$.
199: The spectra have
200: been recalibrated in wavelength and flux using the procedure of \citet{massa}, and
201: de-reddened with a value of $E(B-V)=0.26$ \citep{chochol} using the galactic extinction curve of
202: \citet{fitz}.
203: 
204: \paragraph{}
205: 
206:    We have calculated the light curves, $\bar{F}_{\rm SWP}(t)$ and $\bar{F}_{\rm LWP}(t)$, for the SWP and LWP 
207: bands by calculating the wavelength integrated 
208: mean flux, $\bar{F}_\lambda$, for each flux-corrected, de-reddened spectrum.  The light curves are shown 
209: projected onto the sides of the data cube in Fig. \ref{surf_lo}.
210:     From the shape of the light curves and from the 
211: global shape of the spectra, we can see that the first two spectra were taken during a 
212: stage of rapid decline in UV flux and rapid softening of the UV spectral energy 
213: distribution, which is characteristic of the fireball stage of a nova.  Later, 
214: the UV flux gradually increases and then becomes approximately constant and the spectral distribution gradually 
215: hardens as the nova progresses through the optically thick wind phase.  Finally, the
216: latest LWP spectrum in Fig. \ref{surf_lo} shows prominent \ion{Mg}{2} $hk$ emission,
217: which indicates the onset of the optically thin nebular phase of the nova.  
218: 
219: \paragraph{}
220: 
221:    Fig. \ref{lcurv} 
222: shows the SWP and LWP light curves with the fourteen pairs of coincident spectra shown in 
223: Fig. \ref{surf_lo} marked with vertical lines.  The SWP light peaks about 20 days after the LWP light, which 
224: is consistent with the expected photometric time development of novae (see discussion is
225: Section \ref{modeling}). 
226:    We have formed the observed IUE color, $(SWP-LWP)_{\rm obs}$, at these times by computing the value of 
227: $\log (\bar{F}_{\rm LWP}/\bar{F}_{\rm SWP})$.  
228: The center panel of Fig. \ref{time_teff} shows the time development of $(SWP-LWP)_{\rm obs}$.  
229: As expected, the color becomes more positive (redder) during the 
230: fireball phase where $T_{\rm eff}$ is decreasing, and then gradually becomes more negative
231: (bluer) during the optically thick wind phase where $T_{\rm eff}$ is increasing.  
232: 
233: \section{Modeling}\label{modeling}
234: 
235:    We have calculated a set of atmospheric models that span a $T_{\rm eff}$ range
236: of 12 to 30 kK in increments of 1 kK.  The most important model parameters
237: are the
238: maximum expansion velocity of the linear velocity law, $v_{\rm o}$, which is 
239: $2000$ km s$^{-1}$ \citep{ncyg1}, the exponent of the exponential $\rho$ law, $N$, which is 3,
240: the micro-turbulent velocity broadening width, $\xi$, which is $50$ km s$^{-1}$, 
241: and the abundances, $[\frac{A}{H}]$, which are solar.  A linear velocity law and 
242: a value of $N$ equal to $3$ are consistent with a constant mass loss rate ($\dot{M}=$constant).  The 
243: radius $R=r(\tau_{5000}=1)$, where
244: $\tau_{5000}$ is the continuum optical depth at 5000\AA, is adjusted for each
245: value of $T_{\rm eff}$ to keep the bolometric luminosity, $L_{\rm bol}$, equal to 
246: $50\, 000 {\rm L}_\odot$.  \citet{pistin} found that the calculated spectrum is relatively
247: insensitive to the value of $L_{\rm bol}$, therefore, we do not vary this parameter. 
248: Abundance analyses of nova ejecta find them to have abundances of CNO and Fe that are
249: enhanced with respect to the solar values (see, for example, \citet{nebul}, or \citet{ncyg2}).
250: The enhanced metal abundance is thought to be caused by mixing of the ejected material
251: with the metal rich White Dwarf (WD) material prior to the outburst, and by  
252: H burning during the thermonuclear runaway (TNR) that drives a nova explosion.  
253: \citet{ncyg2} found $[\frac{CNO}{H}]\approx 1$ and 
254: $[\frac{Fe}{H}]\approx 0.3$, by number density with respect to the Sun.  Therefore, we have also 
255: calculated a sub-set of the grid with the hydrogen abundance reduced to half of its solar
256: value, by number, which makes $n_{\rm H}$ equal to $n_{\rm He}$ and $[\frac{A}{H}]$ equal to 
257: 0.3~ (${{A/H}\over{{A/H_\odot}}}=2$ by number) for all metals.
258: 
259: \paragraph{}
260: 
261: Table 2 shows all the species that are treatable in NLTE with {\tt PHOENIX} and
262: indicates which ones have been included in the NLTE rate equations for each value of 
263: $T_{\rm eff}$ in the model grid.  Because ionization stages that are important for some values
264: of $T_{\rm eff}$ in the grid are unimportant for others, the species that are included in NLTE 
265: varies throughout the grid.  In particular, the lowest ionization stages of some species
266: that are included in the coolest models are excluded in hotter models, and higher
267: ionization stages that are included in the hottest models are excluded in the cooler 
268: models.  In practice, test calculations with the full suite of NLTE species were performed 
269: at a sparse sub-grid of models, and ionization stages that contribute less than $1\%$ of the total
270: population of the species at all depths were treated in LTE for nearby models in the complete grid.
271: Fig. \ref{surf_syn} shows the synthetic spectra in the IUE wavelength range that were 
272: computed with the grid of models.  The spectra were computed with a sampling, $\Delta\lambda$,
273: of 0.03 \AA~ so that all spectral lines are well sampled, and then were degraded to the resolution of 
274: the IUE low resolution grating.   
275: 
276: \section{Analysis}\label{analysis}
277: 
278: The physical interpretation of the early photometric and spectroscopic time development of
279: a nova in the UV is well summarized by \citet{ncyg2}. 
280: From Fig. \ref{surf_syn} we see that as the $T_{\rm eff}$ of the model increases, the synthetic
281: spectral energy distribution hardens.  This is similar to the spectral hardening with time
282: during the optically thick wind phase, and is opposite
283: to the spectral softening with time during the fireball phase shown by
284: the observed spectra in Fig. \ref{surf_lo}.  During the brief 
285: fireball phase, an initial, thin shell of gas is ejected by the blast wave that is produced by the
286: detonation.  This initial fireball rapidly expands and cools adiabatically
287: and becomes transparent after the first few days.  This accounts for the initial rapid 
288: decline in UV brightness.  Subsequently, we see the thicker, more extended secondary ejection
289: that forms an expanding photosphere around the WD.  During this optically thick wind phase, $T_{\rm eff}$
290: gradually increases as $\rho$ declines and the expanding atmosphere becomes 
291: increasingly less self-shielded from the central WD.  
292: At the same time,
293: the declining $\rho$ also causes the photospheric radius ($r(\tau=1)$) to decrease so that deeper, 
294: hotter layers are progressively exposed.  
295: As a result, the peak of the emitted flux moves to progressively shorter $\lambda$, as
296: can be seen from Fig. \ref{lcurv}.  
297: During this stage the nova evolves at constant $L_{\rm bol}$ because the expanding atmosphere 
298: re-processes the energy
299: emitted by the underlying WD, which is still burning H on its surface, and does not contain any 
300: energy sources of its own. 
301: The evolution of the UV spectrum during the optically thick wind phase can also be understood by considering
302: the line blanketing.  Near the beginning of the optically thick wind phase, the temperature
303: of the atmosphere is such that Fe is mostly in the form of \ion{Fe}{1} and \ion{}{2}, both of which have a 
304: very rich absorption spectrum in the UV.  As
305: a result, the UV flux is blocked by massive line absorption (the ``iron curtain'').  As the gas 
306: expands it thins and becomes increasingly radiatively heated by the central engine until it 
307: gradually takes on a more
308: nebular character and \ion{Fe}{0} becomes multiply ionized again.  As a result, the UV
309: opacity decreases and the UV flux increases and reaches a plateau about 50 days after the explosion.  
310: Some time after 150 days, the atmosphere expands to the point where the total 
311: optical depth through the atmosphere is less than unity and the continuum flux disappears.
312: 
313: \subsection{Low resolution spectra}
314: 
315: We have computed the synthetic IUE color, $(SWP-LWP)_{\rm syn}$, from the synthetic spectra using the same 
316: procedure that we used for $(SWP-LWP)_{\rm obs}$.  Fig. \ref{color} shows the relationship between 
317: $(SWP-LWP)_{\rm syn}$ and $T_{\rm eff}$.  As expected, the color becomes bluer as 
318: $T_{\rm eff}$ increases.  We note that throughout the $T_{\rm eff}$ range of
319: our model grid, the slope of the $T_{\rm eff}(SWP-LWP)$ relation allows for an unambiguous
320: $T_{\rm eff}$ determination from the $(SWP-LWP)$ color, given a sufficiently small error
321: margin for the color.
322: 
323: \paragraph{}
324: 
325:    We have assigned a Julian Date
326: to each of the synthetic colors, $(SWP-LWP)_{\rm syn}$, shown in Fig. \ref{color} by 
327: interpolation
328: within the $(SWP-LWP)_{\rm obs}(t)$ relation.  This allows us to arrange our 
329: models in a chronological 
330: sequence that reflects the time development of the nova.  The center panel of
331: Fig. \ref{time_teff} shows a comparison of the observed and synthetic 
332: color curves, $(SWP-LWP)_{\rm obs}(t)$ and $(SWP-LWP)_{\rm syn}(t)$.
333: We have also assigned a time sequence of UV color temperatures, $T_{\rm UV}(t)$, 
334: to the sequence of observed colors, $(SWP-LWP)_{\rm obs}(t)$, by interpolation within 
335: the $T_{\rm eff}((SWP-LWP)_{\rm syn})$ relation.
336: The lower panel of Fig. \ref{time_teff} shows the 
337: derived values of $T_{\rm UV}$ as a function of time.  The
338: UV radiation temperature in the IUE range initially cools from $\sim17$ kK to 
339: $\sim12$ kK during the fireball phase,
340: then gradually heats up during the optically thick wind phase, till it reaches
341: $\sim24$ kK about 30 days after the explosion.  
342: %The sudden dip in $T_{\rm UV}$ 
343: %at $\log t=1.2$ (16 days) is due to just one data point, and by comparing all three
344: %panels in Fig. \ref{time_teff} we can see that it is entirely due to a corresponding
345: %upward fluctuation in the LWP light level.  We know of no physical explanation
346: %for such a fluctuation, and surmise that it is a spurious feature of the data.   
347: 
348: \paragraph{}
349: 
350:   Because the IUE $\lambda$ range is close to the peak of the spectral
351: energy distribution, $F_\lambda(\lambda)$, 
352: for this $T_{\rm eff}$ range, we have approximated the time development 
353: of $T_{\rm eff}$, $T_{\rm eff}(t)$,
354: by setting it equal to $T_{\rm UV}(t)$.  On this basis, we assign a synthetic
355: spectrum from the model grid to each of the times marked in Fig. \ref{lcurv}.
356: 
357: \paragraph{}
358: 
359: Fig. \ref{timedev} shows the time development of four model quantities,
360: according to the chronological ordering established above: $R(\tau_{5000}=1)$, 
361: $\log\rho(\tau_{5000}=1)$, the {\it absolute value} of the outward acceleration due to radiation pressure
362: at a continuum optical depth of unity, 
363: $|a_{\rm rad}(\tau_{5000}=1)|$, and the total mass loss rate per year, $dM/dt$.
364: As discussed above, the value of $R(\tau_{5000}=1)$ decreases throughout
365: the optically thick wind phase because, although the atmosphere is expanding,
366: the physical radius at which is becomes optically thick (the photosphere) is contracting as
367: the gas becomes thinner.  The recession of the photosphere to deeper 
368: atmospheric layers also causes
369: the value of $\rho(\tau_{5000}=1)$  to increases
370: during this time.  The value of $|a_{\rm rad}(\tau_{5000}=1)|$ increases
371: during this time because the radiation pressure increases as the atmosphere 
372: becomes hotter.  For reference we have also plotted $g$, the inward
373: acceleration due to gravity at $R(\tau_{5000}=1)$, assuming a $1.25 M_\odot$
374: WD \citep{wdmass}.  The radiative acceleration of the atmosphere is discussed
375: further below.  
376: The value of $g$ increases with time as $R(\tau_{5000}=1)$ decreases,
377: but is always {\it less} that the value of $|a_{\rm rad}(\tau_{5000}=1)|$. 
378: At each time, the value of $dM/dt$ is independent of 
379: depth in the atmosphere, which is consistent with a linear velocity law, but 
380: {\it not} with a 
381: radiatively driven velocity law.  The value of $dM/dt$ increases with
382: time as the value of $|a_{\rm rad}|$ increases.
383: 
384: \paragraph{}
385: 
386:   Fig. \ref{arad} shows a comparison of $\log |a_{\rm rad}|$ and
387: $\log g$ as a function of depth throughout two of the models: 1) a model
388: of $T_{\rm eff}=12$ kK, which corresponds to the time of the first spectrum 
389: of the optically thick wind phase (JD 2448674.40320), and 2) a model of $T_{\rm eff}=17$ 
390: kK, which corresponds to a time about two weeks later later (JD 2448688.85069).  
391: The figure
392: also shows the difference of these quantities (ie $\log |a_{\rm rad}|/g$).
393: For the cooler model, $|a_{\rm rad}|$ is greater than $g$ by as much
394: as a factor of three at depths where $\log\tau_{\rm 5000}$ ranges from
395: $2$ to $-2$. 
396: It is slightly less than $g$ in the $\log\tau_{\rm 5000}$ range between 
397: $-2.5$ and $-6$,
398: and exceeds $g$ again in the outermost part of the atmosphere.
399: The hotter model shows similar behavior in the $\log\tau_{\rm 5000}$ range 
400: between $2$ and $-2$, but $|a_{\rm rad}|$ remains greater than $g$ everywhere.
401: These results suggest that radiation pressure could possibly play a role
402: in determining the velocity structure of the wind.  We are currently
403: implementing the treatment of dynamics in {\tt PHOENIX} with the goal
404: of investigating this question further.  
405: 
406: \paragraph{}
407: 
408: Figs. \ref{temp_plot_0} through \ref{temp_plot_8} show for each of these times
409: the comparison between the observed spectrum and the synthetic spectrum for the 
410: model whose $T_{\rm eff}$ value is closest to the $T_{\rm UV}$ value of the 
411: observed spectrum.  Because we do not know the angular diameter of Cygni 92, we
412: arbitrarily adjust the flux level of the observed spectra to approximately match the
413: absolute flux level of the synthetic spectra.  However, note that in keeping with the
414: condition of constant $L_{\rm bol}$, the {\it same} scale 
415: factor has been used in all plots ({\it ie.} individual pairs of observed and
416: synthetic spectra have {\it not} had their flux calibration ``tuned'').  The observed 
417: spectrum in each
418: panel has a gap in the middle because the reduced SWP and LWP spectra are 
419: disjoint in $\lambda$.  For times later than JD 2448683, the \ion{Mg}{2} $hk$
420: resonance doublet at $\lambda 2800$ becomes increasingly strong in emission with
421: respect to the pseudo-continuum, which indicates that the nova is becoming increasingly
422: nebular at that time.  
423: 
424: \paragraph{}
425:   
426:    For the fireball phase, and for all times throughout the optically thick wind phase
427: up to JD 2448682.8368, 
428: a synthetic spectrum of $T_{\rm eff}\approx T_{\rm UV}$ provides 
429: a good fit to the global shape of the pseudo-continuum throughout the entire 
430: IUE $\lambda$ range.  This is significant because we 
431: see from both Figs. \ref{surf_lo} and \ref{surf_syn} that the SWP range is 
432: sensitive to $T_{\rm eff}$ in this $T_{\rm eff}$ range.  
433: 
434: \paragraph{}
435: 
436:    To illustrate the similarity of the fireball phase to the 
437: optically thick wind phase, we have identified two times, JD2448673.4167
438: (fireball phase), and JD2448686.3241 (optically thick wind 
439: phase) at which the $(SWP-LWP)_{\rm obs}$ color, and, therefore, the value of $T_{\rm UV}$,
440: is approximately the same.  Fig. \ref{fireball2} shows a comparison of the IUE
441: spectra at these two times.  Both spectra have been multiplied by the {\it same} factor to 
442: approximately remove the dilution factor, but they have {\it not} been shifted with 
443: respect to one another.  The absolute flux level and overall shape of the
444: pseudo-continuum is approximately the same.  However, the optically thick wind phase spectrum 
445: exhibits significantly stronger emission lines than the fireball phase spectrum.  The difference 
446: in the strength of the emission lines is due to the greater amount of
447: Doppler broadening from a steeper velocity gradient through the LFR in the thin fireball
448: \citep{ncyg2}.
449: %The later spectrum (JD2448686.3241) occurs at 1.14 $\log$ days on the abscissa axis of
450: %Fig. \ref{lcurv}, where the SWP integrated flux is still rising.  Therefore, the
451: %stronger emission lines in this spectrum are not likely to be due to the
452: %onset of nebulosity, which does not occur till after UV light maximum.
453: 
454: \paragraph{}
455: 
456: For the last two times shown, JD2448688.8507 and JD2448695.6914, none of the
457: synthetic spectra provide as close a fit as that for earlier times.  Furthermore,
458: in contrast to what we find for earlier times, the closest fit synthetic spectrum 
459: is that of a model for which $T_{\rm eff}$ is significantly lower that the
460: observed value of $T_{\rm UV}$.  Fig. \ref{nebular} shows the observed spectrum
461: at these two times compared with an IUE low resolution spectrum that was taken
462: over 100 days later (SWP45135).  At the later time, the nova has become nebular and exhibits a
463: pure emission spectrum with permitted, semi-forbidden, and forbidden lines in the UV.  
464: The velocity of the outflowing gas during the nebular phase is $4500$ km s$^{-1}$
465: (\citep{nebul}).  Therefore, we have shifted the nebular spectrum red-ward by
466: $2000$ km s$^{-1}$.  The nebular emission lines labeled in Fig. \ref{nebular} were
467: identified using Table 1 in \citet{nebul}.  The comparison of the nebular phase
468: spectrum with the two earlier spectra shows that by JD2448688.8507 many of the nebular 
469: lines were already present in the spectrum.  The expanding atmosphere was
470: already partially nebular by JD2448688.8507 in that many of the nebular lines had already
471: begun to ``contaminate'' the optically thick wind phase spectrum.  As a result,
472: we expect that {\tt PHOENIX} models will provide an increasingly worse fit beyond
473: this time because {\tt PHOENIX} is capable of modeling only the component
474: of the outburst that constitutes an optically thick atmosphere.
475: 
476: %\paragraph{} 
477: %    
478: %For both of these times, the \ion{Mg}{2} $hk$ resonance doublet at $\lambda 2800$
479: %is becoming increasingly strong in emission with respect to the pseudo-continuum.
480: %This indicates the onset of the nebular phase of the nova atmosphere.  The fit
481: %may be affected because the derived value of $T_{\rm UV}$ and, hence, $T_{\rm eff}$,
482: %is based on an $(SWP-LWP)$ color for which the LWP light level is increasingly contaminated
483: %by the strong \ion{Mg}{2} $hk$ emission and is no longer entirely representative
484: %of the background pseudo-continuum flux level.  This will cause $(SWP-LWP)$ to be
485: %more positive (redder) than it would be without the strong \ion{Mg}{2} emission line
486: %and will lead to a depression of the inferred value of $T_{\rm UV}$.  To investigate this possibility,
487: %we have removed the $hk$ lines from the observed spectra and replaced them with the
488: %mean of the flux in 50\AA~ windows on either side of the $hk$ doublet.  We then re-computed 
489: %$(SWP-LWP)_{\rm obs}$ and found that it was negligibly altered.  We conclude that the 
490: %$hk$ emission lines, though strong, are too narrow to contribute significantly to the
491: %total flux in the LWP band pass at these phases.  
492: %%Furthermore, for JD2448686.3241, we note that this time corresponds to 
493: %%the anomalous dip in the LWP
494: %%light level, and the consequent spike in the inferred value of $T_{\rm UV}$, that was
495: %%discussed in connection with Fig. \ref{time_teff}.
496: 
497: \subsection{High resolution spectra}
498: 
499:     Figs. \ref{hi_plot0} and \ref{hi_plot4} show a time sequence of high resolution IUE spectra in the
500: LWP pass band that overlaps in time the sequence shown in Figs. \ref{temp_plot_0} through \ref{temp_plot_8}.
501: Synthetic spectra have been assigned to each time by interpolation in the time vs $T_{\rm UV}$ 
502: relation that was derived from the low resolution spectra (lower panel of 
503: \ref{time_teff}).   The prominent emission feature at $\lambda 2800$ is the $hk$ resonance doublet
504: of \ion{Mg}{2}.  The \ion{Mg}{2} $hk$ lines become stronger with time because the expanding
505: atmosphere gradually becomes more nebular.  As a result, the models, which only account for 
506: the photospheric part of the expanding gas shell, increasingly underpredict the strength of these 
507: lines as time increases.  The prominent emission
508: features at $\lambda 2630$ and $\lambda 2900$, which are approximately reproduced by the models at many 
509: times, are not emission lines, but
510: are regions of relatively lower line opacity through which more flux escapes than through the neighboring 
511: regions.  
512: 
513: \paragraph{}
514: 
515: Generally, the models approximately reproduce the broad features of the high resolution spectra
516: during the optically thick wind phase.  However, there are discrepancies in the detailed fit.
517: One possible source of discrepancy is the distribution of metal abundances.  The nova ejecta
518: may be depleted in H and enriched in C, N, and O as a result of WD material having been 
519: mixed into the accreted layers, or as a result of nuclear processed elements from the surface H
520: burning that drives the nova explosion having been ejected.  \citet{ncyg2} found that $[\frac{CNO}{H}]$
521: is approximately one while $[\frac{Fe}{H}]$ is only about 0.3.  Some discrepancy may also
522: be due to the form of the velocity law.  If the ejecta are being driven outward by radiation pressure
523: at this stage, then $v(r)$ may have the power law form of a wind rather than a linear form \citep{CAK}.  Indeed, since the time of \citet{mclaugh} it has been 
524: observed that relatively blue shifted spectral absorption components arise
525: in relatively deep layers of the expanding atmosphere, which suggests a
526: more complex velocity law than the linear one employed here.  
527: We are currently in the process of converging power law wind models. 
528:  Another possible source of
529: discrepancy is inhomogeneities in the expanding gas shell, which have been observed (see, for example,
530: \citet{payne} or \citet{nebul}), and which are expected to arise if the velocity
531: field is complex.  These will serve to make the line
532: profiles more complex than those predicted with a homogeneous model.
533: 
534: \paragraph{}
535: 
536: The upper left panel in Fig. \ref{hi_plot0} shows the fit to the only high resolution spectrum
537: taken during the fireball phase.  At this resolution it is clear that the nova models systematically
538: over-predict the strength of all emission features by a factor of approximately two.  The dashed
539: line is the synthetic spectrum of a model with $T_{\rm eff}$ equal to $12000$ K, $v_{\rm max}$
540: equal to $4000$ km s$^{-1}$, and $N=7$ rather than $3$ ($\rho(r)~\alpha ~r^{-N}$).  
541: This model is much thinner in the radial direction and has much steeper $\rho(r)$ and $v(r)$ gradients
542: than the models in our nova grid.  Moreover, because $N>3$, the motion of the atmosphere does
543: not correspond to a constant mass loss rate, but, rather to homologous expansion.  The
544: characteristics of this model are more 
545: similar to those of supernova models rather than typical nova models.  Such a model yields a synthetic
546: spectrum with much weaker emission features than that of a more typical nova model because the steeper 
547: $v(r)$ gradient leads to a greater Doppler smearing of the emergent spectrum.  As a
548: result, this model provides a significantly better fit to the observed spectrum.  This is
549: consistent with the idea that the fireball phase is due to a thin shell of gas that is ejected
550: by the blast wave before the expanding atmosphere that causes the optically thick wind phase
551: develops.  We note
552: that the best fit value of $T_{\rm eff}$ for the fireball spectrum is $4000$ K lower when it
553: is fit with the thinner, steeper model rather than a typical nova model.
554: \citet{ncyg2} found that a thin shell model with a steep ($N=15$) $\rho$ law
555: and enhanced \ion{CNO}{0} and \ion{Fe}{0} abundances was required to fit
556: the first two observed spectra.  A comparison of the 2600 to 3000 \AA~ region in their Fig. 5b with 
557: Fig. \ref{hi_plot0} shows
558: that we achieve a better fit to the fireball phase with a solar abundance model than they did. 
559: The improvement may be due to the greater number of species treated in NLTE.  
560: 
561: \paragraph{}
562:  
563:     Fig. \ref{hires_id} shows a representative high resolution observed spectrum from the early optically thick
564: wind phase (LWP22457, JD2448677.67626, $T_{\rm UV}\approx 15$ kK).  Also shown is the best fit
565: synthetic spectrum.  The plot has been labeled with the identifications of the strongest lines,
566: as determined by the spectrum synthesis calculation.  During the line identification procedure the Fe group
567: elements were ignored to prevent the plot from being completely saturated with the labels of \ion{Fe}{2}
568: and, to a lesser extent,
569: of \ion{Fe}{1}, \ion{Fe}{3}, \ion{Ti}{2}, \ion{Cr}{2}, \ion{Co}{2}, and \ion{Ni}{2}, which are the most 
570: important source of opacity at most wavelength points.
571: 
572: \subsection{Abundance effects}
573: 
574:   Fig. \ref{2H} shows a comparison between the synthetic spectra of models with solar abundance and 
575: those that are hydrogen deficient in a representative region of the IUE spectral range.  Reducing 
576: $n_{\rm H}$ to half of its solar value by number makes 
577: $\frac{n_{\rm He}}{n_{\rm H}}=1$, and $[\frac{A}{H}]=0.3~ ({{A/H}\over{{A/H}_\odot}}=2$ 
578: by number) for all metals.  
579: From Fig. \ref{2H} we see that the hydrogen deficiency suppresses the
580: overall flux throughout the near UV by approximately a factor of two while preserving
581: the relative strength of the spectral features.  We also show for comparison the synthetic
582: spectrum of a hydrogen deficient model with a value of $T_{\rm eff}$ that is 1kK greater than that
583: of the solar abundance model.  The synthetic spectrum from the hydrogen deficient model with the
584: larger $T_{\rm eff}$ approximately matches both the overall flux level and the detailed structure of
585: the synthetic spectrum of the solar abundance model throughout the IUE wavelength range.  To a first 
586: approximation, a solar abundance model is interchangeable with a hydrogen deficient model that is
587: 1 kK hotter when fitting the near UV spectrum.  For the accuracy of the fitting done here, it is 
588: necessary to fit models to a region of
589: the spectrum other than the UV to disentangle $T_{\rm eff}$ from $[\frac{A}{H}]$.
590: 
591: \paragraph{}
592: 
593:     Fig. \ref{2H_plot} shows a comparison of the observed spectrum and synthetic spectra from
594: the best fit solar abundance model ($T_{\rm eff}=T_{\rm UV}$) and a hydrogen deficient model
595: of $T_{\rm eff}=T_{\rm UV}+1$ kK at two times during the optically thick wind phase.  From a 
596: visual inspection, both synthetic spectra provide approximately the same goodness of fit at both times.
597: We conclude that the result of fitting hydrogen deficient models to the observed spectral
598: sequence is to shift the derived $T_{\rm eff}$ evolution upward by approximately 1 kK.  There
599: are no spectral features in this range that distinguish between two such models.  Therefore,
600: there is a degeneracy in $T_{\rm eff}$ and $n_{\rm H}$, at least within the range of parameters
601: explored here.  Superficially, the degeneracy arises because a larger value of $T_{\rm eff}$ 
602: enhances the UV flux (see Fig. \ref{surf_syn}), whereas the increased value of $[{Fe\over H}]$
603: suppresses the UV flux.  The net result is that increasing both in a particular proportion will lead
604: to an approximately similar UV flux spectrum.  A more fundamental discussion of spectral
605: similarity and non-uniqueness problems in atmospheric models of novae can be found in 
606: \citet{p&s}.        
607: 
608: \section{Conclusion}\label{conclusions}
609: 
610: We conclude that the UV radiation temperature ($T_{\rm UV}$) is a good measure of $T_{\rm eff}$ during
611: the early stage of the nova when the spectrum is purely that of an optically thick
612: wind.   We find that this stage lasts until JD2448682.8368, which is about 10 days after the
613: first fireball phase spectrum was taken.  
614: During this interval, {\tt PHOENIX} models are able to reproduce the overall shape of the low resolution
615: UV spectra in the IUE range.  This is significant because the nova 
616: photosphere increases in $T_{\rm eff}$ by $\approx 5000$ K during this
617: time, and the hardness of the near UV radiation field is sensitive to 
618: $T_{\rm eff}$ in the range from ten to twenty kK.  The models 
619: provide a moderately good fit to the high resolution spectra during 
620: the optically thick wind phase, but there are significant discrepancies
621: which are due to, among other things, a non-solar abundance distribution
622: due to nuclear processing, an inaccurate velocity law, and inhomogeneities 
623: in the expanding atmosphere.  
624: 
625: \paragraph{}
626: 
627:    Models in which H is depleted to a tenth of its 
628: solar value and in which $\frac{A}{H}$ is twice the solar value provide a similarly good fit
629: to the observed spectra in the UV if the value of $T_{\rm eff}$ of the model
630: is increased by $\approx 1$ kK.  We conclude that $[\frac{A}{H}]$ and $T_{\rm eff}$
631: are degenerate parameters when fitting models to the near UV spectrum.  It is necessary
632: to fit models to another $\lambda$ region to uniquely determine $T_{\rm eff}$ and
633: $[\frac{A}{H}]$.   
634: 
635: \paragraph{}
636:  
637: The high resolution spectrum of the fireball
638: phase of the nova has weaker emission features than an optically thick
639: wind phase spectrum of the same best fit $T_{\rm eff}$.  The fireball phase
640: spectrum is better fit by a supernova model, in which the atmosphere
641: is thinner and has a steeper $\rho(r)$ and $v(r)$ gradient.       
642: 
643: %% If you wish to include an acknowledgments section in your paper,
644: %% separate it off from the body of the text using the \acknowledgments
645: %% command.
646: 
647: %% Included in this acknowledgments section are examples of the
648: %% AASTeX hypertext markup commands. Use \url without the optional [HREF]
649: %% argument when you want to print the url directly in the text. Otherwise,
650: %% use either \url or \anchor, with the HREF as the first argument and the
651: %% text to be printed in the second.
652: 
653: \acknowledgments
654: 
655: This work was supported in part by NSF grant
656: AST-9720704, NASA ATP grant NAG 5-8425 and LTSA grant NAG 5-3619, as
657: well as NASA/JPL grant 961582 to the University of Georgia, by NSF
658:  grant
659: AST-9731450, NASA grant NAG5-3505 and an IBM SUR grant to the
660: University of Oklahoma, and by NSF and NASA grants to 
661: Arizona State University.  PHH was supported in part by the P\^ole
662: Scientifique de Mod\'elisation Num\'erique at ENS-Lyon.  Some of the
663: calculations presented in this paper were performed on the 
664: the IBM SP2 and SGI Origin 2000 of the UGA UCNS, on the IBM BlueHorizon of
665: the San Diego Supercomputer Center (SDSC), with support from the
666: National Science Foundation, and on the IBM SP and Cray T3E of the NERSC with
667: support from the DoE.  We thank all these institutions for a generous
668: allocation of computer time.
669: 
670: %% Appendix material should be preceded with a single \appendix command.
671: %% There should be a \section command for each appendix. Mark appendix
672: %% subsections with the same markup you use in the main body of the paper.
673: 
674: %% Each Appendix (indicated with \section) will be lettered A, B, C, etc.
675: %% The equation counter will reset when it encounters the \appendix
676: %% command and will number appendix equations (A1), (A2), etc.
677: 
678: %% The reference list follows the main body and any appendices.
679: %% Use LaTeX's thebibliography environment to mark up your reference list.
680: %% Note \begin{thebibliography} is followed by an empty set of
681: %% curly braces.  If you forget this, LaTeX will generate the error
682: %% "Perhaps a missing \item?".
683: %%
684: %% thebibliography produces citations in the text using \bibitem-\cite
685: %% cross-referencing. Each reference is preceded by a
686: %% \bibitem command that defines in curly braces the KEY that corresponds
687: %% to the KEY in the \cite commands (see the first section above).
688: %% Make sure that you provide a unique KEY for every \bibitem or else the
689: %% paper will not LaTeX. The square brackets should contain
690: %% the citation text that LaTeX will insert in
691: %% place of the \cite commands.
692: 
693: %% We have used macros to produce journal name abbreviations.
694: %% AASTeX provides a number of these for the more frequently-cited journals.
695: %% See the Author Guide for a list of them.
696: 
697: %% Note that the style of the \bibitem labels (in []) is slightly
698: %% different from previous examples.  The natbib system solves a host
699: %% of citation expression problems, but it is necessary to clearly
700: %% delimit the year from the author name used in the citation.
701: %% See the natbib documentation for more details and options.
702: 
703: \begin{thebibliography}{}
704: \bibitem[Castor {\it et al.}(1975)]{CAK} Castor, J.D., Abbott, D., \& Klein, R., 1975, \apj, 195, 157
705: \bibitem[Chochol {\it et al.}(1997)]{chochol} Chochol, D., Grygar, J., Pribullai, T., Komzik, R., 
706: Hric, L., \& Elkin, V. 1997, \aap, 318, 908
707: \bibitem[Collins(1992)]{collins} Collins, M., 1992, IAU Circ. No. 5454
708: \bibitem[Fitzpatrick(1999)]{fitz} Fitzpatrick, E. 1999, \pasp, 111, 63
709: \bibitem[Hauschildt \& Baron(1999)]{hausphx} Hauschildt, P. H. \& Baron, E., 1999, J. Comput. App. Math. 102, 41 
710: \bibitem[Hauschildt {\it et al.}(1994)]{ncyg2} Hauschildt, P. H., Starrfield, S., Austin, S., 
711: Wagner, R. M., Shore, S. N., \& Sonneborn, G., 1994, \apj, 422, 831
712: \bibitem[Massa and Fitzpatrick(1998)]{massa} Massa, D. \& Fitzpatrick, E. 1998, \aap, 193, 1122
713: \bibitem[McLaughlin (1947)]{mclaugh} McLaughlin, D., 1947, \pasp, 59, 244 
714: \bibitem[Payne-Gaposchkin(1957)]{payne} Payne-Gaposchkin, C., 1957, The Galactic Novae (Amsterdam:
715: North-Holland)
716: \bibitem[Pistinner and Shaviv(1996)]{p&s} Pistinner, S. \& Shaviv, G., 1996, \apj, 461, L45 
717: \bibitem[Pistinner {\it et al.}(1995)]{pistin} Pistinner, S., Shaviv, G., Hauschildt, P. \& Starrfield, S., 1995, \apj 451, 724
718: \bibitem[Shore {\it et al.}(1994)]{ncyg1} Shore, S. N., Sonneborn, G., Starrfield, S., Gonzalez-Riestra \&
719: R., Polidan, R. S., 1994, \apj, 421, 344 
720: \bibitem[Shore {\it et al.}(1997)]{nebul} Shore, S. N., Starrfield, S., Ake, T. B. \& Hauschildt, P. H.,
721: 1997, \apj, 490, 393  
722: \bibitem[Short {\it et al.}(1999)]{short1} Short, C. I., Hauschildt, P. H., \& Baron, E., 1999, \apj, 525, 375
723: \bibitem[Starrfield {\it et al.}(1997)]{wdmass} Starrfield, S., Truran, J. W., Wiescher, M. C., \& Sparks, W. M., 1997, \mnras, 296, 502 
724: \end{thebibliography}
725: 
726: %% No more than seven \figcaption commands are allowed per page,
727: %% so if you have more than seven captions, insert a \clearpage
728: %% after every seventh one.
729: 
730: \clearpage
731: 
732: %\voffset=2.0truein
733: \begin{figure}
734: \plotone{f1.eps}
735: \figcaption[f1.eps]
736: {Time series of IUE SWP and LWP spectra at the fourteen phases 
737: of approximate simultaneity of the bands.  The gap along the $\lambda$ axis
738: occurs because the reduced SWP and LWP spectra are disjoint in wavelength.  The mean wavelength 
739: integrated flux, $\bar{F}_\lambda$, for the SWP and LWP bands as a function of time
740: (light curve) is plotted on the short wavelength and long wavelength 
741: walls of the plot cube, respectively.  
742: \label{surf_lo} }
743: \end{figure}
744: 
745: \clearpage
746: 
747: \hoffset=-0.5truein
748: \begin{figure}
749: \plotone{f2.eps}
750: \figcaption[f2.eps]
751: {UV light curves.  Mean $\lambda$ integrated flux, $\bar{F}_\lambda$, as a function of
752: time for the LWP (squares, darker line) and SWP (triangles, lighter line).
753: Vertical lines connecting the two light curves denote times of 
754: approximate simultaneity of the SWP and LWP observations where
755: IUE colors were calculated (see text). 
756: \label{lcurv} }
757: \end{figure}
758: 
759: \clearpage
760: 
761: \begin{figure}
762: \plotone{f3.eps}
763: \figcaption[f3.eps]
764: {Deriving the time development of $T_{\rm UV}$ from the IUE data.  Upper panel: Mean $\lambda$ 
765: integrated flux, $\bar{F}_\lambda$, for the LWP (squares, darker line) and SWP (triangles, lighter line).  
766: Center panel: IUE color, $\log\bar{F}_{\rm LWP}/\bar{F}_{\rm SWP}$.
767: Observed (plus symbols, dotted line), synthetic (asterisks, dashed line).  Lower panel: derived
768: UV color temperature, $T_{\rm UV}$ (see text). 
769: \label{time_teff} }
770: \end{figure}
771: 
772: \clearpage
773: 
774: %\voffset=1.0truein
775: \begin{figure}
776: \plotone{f4bm.ps}
777: \figcaption[f4.eps]
778: {Model flux surface for nova grid consists of closely spaced synthetic spectra.  Surface
779: gives general overview of how the global flux level and spectral energy distribution changes
780: with $T_{\rm eff}$.
781: \label{surf_syn}} 
782: \end{figure}
783: 
784: \clearpage
785: 
786: \begin{figure}
787: \plotone{f5.eps}
788: \figcaption[f5.eps]
789: {Model $T_{\rm eff}$ as a function of synthetic color $(SWP-LWP)$.  
790: \label{color} }
791: \end{figure}
792: 
793: \clearpage
794: 
795: %\voffset=3.0truein
796: \begin{figure}
797: \plotone{f5a.eps}
798: \figcaption[f5a.eps]
799: {The time development of four model quantities according to the 
800: chronological ordering based on the IUE color (see text).  Upper 
801: left panel: $R(\tau_{\rm 5000}=1)$; upper right panel: 
802: $\log\rho(\tau_{\rm 5000}=1)$; lower left panel:
803: $|a_{\rm rad}(\tau_{\rm 5000}=1)|$, and, for reference,
804: the value of $g$ assuming a 1.25 $M_\odot$ WD; lower right panel: $dM/dt$. 
805: \label{timedev} }
806: \end{figure}
807: 
808: \clearpage
809: 
810: %\voffset=3.0truein
811: \begin{figure}
812: \plotone{f5b.eps}
813: \figcaption[f5b.eps]
814: {Comparison of $\log |a_{\rm rad}|$ and $\log g$ throughout two models 
815: that represent
816: two different times during the optically thick wind phase.  Darker line:
817: JD 2448674.40320; lighter line: JD 2448688.85069.  Upper panel: 
818: $\log |a_{\rm rad}|$ (solid line); $\log g$ (dashed line).  Lower panel:
819: $\log |a_{\rm rad}|/g$.
820: \label{arad}} 
821: \end{figure}
822: 
823: \clearpage
824: 
825: %\voffset=2.0truein
826: \hoffset=-0.0truein
827: \begin{figure}
828: \plotone{f6bm.ps}
829: \figcaption[f6.eps]
830: {Comparison of observed (lighter line) and synthetic (darker line) spectra
831: at the first twelve phases of approximate band simultaneity.  The gap
832: in $\lambda$ for the observed spectra occurs because the reduced SWP and LWP 
833: spectra are disjoint in $\lambda$.  Each panel is
834: labeled with the JD of the observed spectrum, the derived value of $T_{\rm UV}$
835: from the $(SWP-LWP)$ color, and the value of $T_{\rm eff}$ of the model used
836: to synthesize the spectrum. 
837: \label{temp_plot_0} }
838: \end{figure}
839: 
840: \clearpage
841: 
842: %\voffset=2.0truein
843: \hoffset=-0.5truein
844: \begin{figure}
845: \plotone{f7bm.ps}
846: \figcaption[f7.eps]
847: {See Fig. \ref{temp_plot_0}
848: \label{temp_plot_4} } 
849: \end{figure}
850: 
851: \clearpage
852: 
853: 
854: %\voffset=2.0truein
855: \hoffset=-0.5truein
856: \begin{figure}
857: \plotone{f8bm.ps}
858: \figcaption[f8.eps]
859: {See Fig. \ref{temp_plot_0}
860: \label{temp_plot_8} }
861: \end{figure}
862: 
863: \clearpage
864: 
865: %\voffset=3.0truein
866: \begin{figure}
867: \plotone{f9.eps}
868: \figcaption[f9.eps]
869: {The last two observed spectra that were fit in Fig. \ref{temp_plot_8} (JD2448688.85069
870: and JD2448695.69135) (lighter line).  A spectrum from the nebular phase of the outburst taken
871: over 100 days later (darker line).  The line annotations refer to nebular lines. 
872: \label{nebular}}
873: \end{figure}
874: 
875: \clearpage
876: 
877: %\voffset=3.0truein
878: \begin{figure}
879: \plotone{f10.eps}
880: \figcaption[f10.eps]
881: {Comparison of fireball and optically thick wind phase at equal color.  Upper panel: observed
882: IUE color $(SWP-LWP)$ as a function of time.  The squares connected by the horizontal line indicate
883: the two phases being compared in the lower panel.  Lower panel: Comparison of fireball spectrum 
884: (darker line) and optically thick wind phase spectrum (lighter line) at the two phases indicated
885: in the upper panel.
886: \label{fireball2} }
887: \end{figure}
888: 
889: \clearpage
890: 
891: %\voffset=3.0truein
892: \begin{figure}
893: \plotone{f11bm.ps}
894: \figcaption[f11.eps]
895: {Comparison of high resolution observed (lighter line) and synthetic (darker line) spectra
896: at eight phases.  Each panel is
897: labeled with the JD of the observed spectrum, the interpolated value of $T_{\rm UV}$
898: from the time vs $(SWP-LWP)$ relation (see text), and the value of $T_{\rm eff}$ of the model used
899: to synthesize the spectrum.  Upper left panel: the dashed line is the synthetic 
900: spectrum of a supernova model with $T_{\rm eff}$ equal to 12 kK.
901: \label{hi_plot0} } 
902: \end{figure}
903: 
904: \clearpage
905: 
906: %\voffset=3.0truein
907: \begin{figure}
908: \plotone{f12bm.ps}
909: \figcaption[f12.eps]
910: {See Fig. \ref{hi_plot0}
911: \label{hi_plot4} } 
912: \end{figure}
913: 
914: \clearpage
915: 
916: %\voffset=3.0truein
917: \begin{figure}
918: \plotone{f13.eps}
919: \figcaption[f13.eps]
920: {Comparison of high resolution observed (lighter line) and synthetic (darker line) spectra
921: at a representative time during the optically thick wind phase.  
922: The strongest lines in the spectrum synthesis, other than Fe group lines, have been labeled with
923: their identifications.
924: \label{hires_id} }
925: \end{figure}
926: 
927: \clearpage
928: 
929: %\voffset=3.0truein
930: \begin{figure}
931: \plotone{f14bm.ps}
932: \figcaption[f14.eps]
933: {Comparison of synthetic spectra with various values of $T_{\rm eff}$ and $[\frac{A}{H}]$.
934: Lighter line: Solar abundances, $T_{\rm eff}=14$ kK (upper panel) and $16$ kK (lower panel).
935: Darker solid line: $[\frac{A}{H}]=0.3$, $T_{\rm eff}=14$ kK (upper panel) and $16$ kK (lower panel). 
936: Darker dashed line: $[\frac{A}{H}]=0.3$, $T_{\rm eff}=15$ kK (upper panel) and $17$ kK (lower panel). 
937: \label{2H} }
938: \end{figure}
939: 
940: \clearpage
941: 
942: %\voffset=3.0truein
943: \begin{figure}
944: \plotone{f15bm.ps}
945: \figcaption[f15.eps]
946: {Comparison of observed (darker line) and synthetic (lighter lines) spectra
947: at two representative times.  Lighter solid line: $[{A\over H}]=0$ (solar abundance) 
948: models; lighter dashed line: $[{A\over H}]=0.3$ models. 
949: \label{2H_plot} } 
950: \end{figure}
951: 
952: \clearpage
953: 
954: %\voffset=-0.5truein
955: \hoffset=+0.0truein
956: \begin{deluxetable}{lrlr}
957: \scriptsize
958: \tablecaption{Log of low resolution IUE spectra}
959: \tablecomments{Matching superscripts indicate approximately simultaneous LWP/SWP pairs.} 
960: \label{t2}
961: \tablecolumns{4}
962: \tablewidth{0pt}
963: \tablehead{
964: \multicolumn{2}{c}{LWP} & \multicolumn{2}{c}{SWP} \\
965: \colhead{Exp. No.} & \colhead{JD-2448000} & \colhead{Exp. No.} & \colhead{JD-2448000} }
966: \startdata 
967:  ${\rm lwp}22425^1$    &  673.40  & ${\rm swp}44030^1$    &  673.41 \\
968:  ${\rm lwp}22426^2$    &  673.45  & ${\rm swp}44031^2$    &  673.46 \\
969:  ${\rm lwp}22430^3$    &  674.40  & ${\rm swp}44039^3$    &  674.40 \\
970:  ${\rm lwp}22434^4$    &  675.19  & ${\rm swp}44040$      &  674.45 \\
971:  ${\rm lwp}22435^5$    &  675.25  & ${\rm swp}44043^4$    &  675.20 \\
972:  ${\rm lwp}22448$      &  676.41  & ${\rm swp}44044^5$    &  675.27 \\
973:  ${\rm lwp}22449^6$    &  676.45  & ${\rm swp}44050^6$    &  676.44 \\
974:  ${\rm lwp}22456^7$    &  677.62  & ${\rm swp}44051$      &  676.69 \\
975:  ${\rm lwp}22459^8$    &  678.57  & ${\rm swp}44055^7$    &  677.61 \\
976:  ${\rm lwp}22462^9$    &  678.86  & ${\rm swp}44056$      &  677.66 \\
977:  ${\rm lwp}22470$      &  680.50  & ${\rm swp}44060^8$    &  678.58 \\
978:  ${\rm lwp}22483^{10}$ &  682.86  & ${\rm swp}44062$      &  678.69 \\
979:  ${\rm lwp}22485$      &  682.94  & ${\rm swp}44064^9$    &  678.86 \\
980:  ${\rm lwp}22513^{11}$ &  686.38  & ${\rm swp}44073$      &  680.43 \\
981:  ${\rm lwp}22533^{12}$ &  688.85  & ${\rm swp}44086^{10}$ &  682.83 \\
982:  ${\rm lwp}22573$      &  693.43  & ${\rm swp}44102$      &  684.48 \\
983:  ${\rm lwp}22592^{13}$ &  695.68  & ${\rm swp}44115^{11}$ &  686.32 \\
984:  ${\rm lwp}22635^{14}$ &  700.49  & ${\rm swp}44130^{12}$ &  688.85 \\
985:  ${\rm lwp}23077$      &  752.34  & ${\rm swp}44174^{13}$ &  695.69 \\
986:  ${\rm lwp}23134$      &  761.76  & ${\rm swp}44193^{14}$ &  700.49 \\
987:  ${\rm lwp}23170$      &  767.66  & ${\rm swp}44209$      &  703.85 \\
988:  ${\rm lwp}23188$      &  770.31  & ${\rm swp}44233$      &  707.50 \\
989:  ${\rm lwp}23210$      &  772.64  & ${\rm swp}44268$      &  711.89 \\
990:  ${\rm lwp}23276$      &  783.24  & ${\rm swp}44305$      &  715.32 \\
991:  ${\rm lwp}23312$      &  790.26  & ${\rm swp}44338$      &  717.77 \\
992:  ${\rm lwp}23348$      &  794.42  & ${\rm swp}44377$      &  723.43 \\
993:  ${\rm lwp}23351$      &  795.67  & ${\rm swp}44389$      &  725.76 \\
994:  ${\rm lwp}23399$      &  802.58  & ${\rm swp}44439$      &  732.75 \\
995:  ${\rm lwp}23424$      &  807.15  & ${\rm swp}44632$      &  752.31 \\
996:  ${\rm lwp}23425$      &  807.20  & ${\rm swp}44634$      &  752.42 \\
997:  ${\rm lwp}23426$      &  807.31  & ${\rm swp}44717$      &  761.70 \\
998:  ${\rm lwp}23501$      &  818.17  & ${\rm swp}44761$      &  767.67 \\
999:  ${\rm lwp}23599$      &  833.04  & ${\rm swp}44762$      &  767.70 \\
1000:  ${\rm lwp}23670$      &  845.01  & ${\rm swp}44790$      &  770.31 \\
1001:  ${\rm lwp}23706$      &  851.43  & ${\rm swp}44808$      &  772.64 \\
1002:  ${\rm lwp}23802$      &  864.30  & ${\rm swp}44901$      &  783.26 \\
1003:  ${\rm lwp}23869$      &  873.00  & ${\rm swp}44937$      &  790.24 \\
1004:  ${\rm lwp}24153$      &  921.60  & ${\rm swp}44970$      &  794.43 \\
1005: \enddata
1006: \end{deluxetable}
1007: 
1008: \clearpage
1009: 
1010: \begin{deluxetable}{lrrrrrrr}
1011: \scriptsize
1012: \tablecaption{Ranges of $T_{\rm eff}$ (in kK) in the model grid for which each ionization stage of each
1013: element was treated in NLTE.}
1014: \tablecomments{Elements in bold face have been added since the last NLTE {\tt PHOENIX} analysis
1015: of Cygni 1992.  Those in italics have had their treatment improved since the last analysis.} 
1016: \label{t3}
1017: \tablecolumns{8}
1018: \tablewidth{0pt}
1019: \tablehead{
1020: \colhead{Element} & \multicolumn{7}{c}{Ionization Stage} \\
1021: \colhead{} & \colhead{\ion{}{1}} & \colhead{\ion{}{2}} & \colhead{\ion{}{3}} & \colhead{\ion{}{4}} & \colhead{\ion{}{5}} & \colhead{\ion{}{6}} & \colhead{\ion{}{7}}   } 
1022: \startdata
1023: {\it H}  & All &\nodata &\nodata &\nodata &\nodata &\nodata &\nodata \\
1024: He       & All & All &\nodata &\nodata &\nodata &\nodata &\nodata \\
1025: Li       & None & None &\nodata &\nodata &\nodata &\nodata &\nodata \\
1026: C        & 12-24 & All & All & All &\nodata &\nodata &\nodata \\
1027: N        & All & All & All & All & None & None  &\nodata\\
1028: O        & All & All & All & All & None & None &\nodata\\
1029: Ne       & All &\nodata &\nodata &\nodata &\nodata &\nodata &\nodata\\
1030: {\it Na} & 12-24 & All & All & All & None & None  &\nodata\\
1031: {\it Mg} & 12-24 & All & All & All & None & None  &\nodata\\
1032: {\bf Al} & 12-24 & All & All & All & None & None  &\nodata\\
1033: Si       & 12-24 & All & All & All & 30 & None &\nodata \\
1034: {\bf P}  & 12-22 & 12-22 & 12-22 & 12-22 & None & None  &\nodata\\
1035: S        & 12-24 & All & All & All & 30 & None  &\nodata\\
1036: {\bf K}  & None  & None & None & None & None  &\nodata &\nodata\\
1037: Ca       & 12-24 & 12-24 & All & All & All & None & None \\
1038: Ti       & None & None  &\nodata &\nodata &\nodata &\nodata &\nodata\\
1039: Fe       & 12-24 & All & All & All & All & None  &\nodata\\
1040: Co       & None & None & None  &\nodata &\nodata &\nodata &\nodata\\
1041: {\bf Ni} & None & 25-30 & 25-30 & 25-30 & 30 & None &\nodata\\
1042: \enddata
1043: \end{deluxetable}
1044: 
1045: %% Tables should be submitted one per page, so put a \clearpage before
1046: %% each one.
1047: 
1048: 
1049: \end{document}
1050: 
1051: %%
1052: %% End of file `sample.tex'.
1053: 
1054: