1: %% tar -cvf bruntt2.tar f*.eps ms.tex ms.bib ms.bbl tab2.tex response
2: %% This is a manuscript on Polaris for ApJ by Bruntt et al. (2008)
3: %% AASTeX v5.x LaTeX 2e macros.
4:
5: %% The command below calls the preprint style
6: %% which will produce a one-column, single-spaced document.
7: %% Examples of commands for other substyles follow. Use
8: %% whichever is most appropriate for your purposes.
9: %%
10: %%\documentclass[12pt,preprint]{aastex}
11:
12: %% ApJ migration by HB 19 Dec 2007
13: %% Granulation interpertation started Dec 17.
14: %% This is revision 1 after incl. comments by the referee.
15:
16: % \documentclass[usegraphicx,usenatbib,manuscript]{aastex}
17: % \documentclass[preprint2,usenatbib]{aastex}
18: \documentclass[preprint2,usenatbib]{emulateapj}
19:
20: % \documentclass[12pt,preprint]{aastex} %% for the referee
21: \bibliographystyle{apj}
22:
23: \newcommand{\vdag}{(v)^\dagger}
24: \newcommand{\myemail}{bruntt@physics.usyd.edu.au}
25:
26: %%%%% HB's commands:
27: \newcommand{\micromag}{{$\mu$mag}}
28: \newcommand{\rms}{{\emph{rms}}}
29: \newcommand{\msun}{\ensuremath{M_\odot}} % Maths solar mass symbol
30: \newcommand{\rsun}{\ensuremath{R_\odot}} % Maths solar radius symbol
31: \newcommand{\cms}{\,cm\,s$^{-2}$} % cm/s symbol for cgs \logg
32: \newcommand{\vsini}{\ensuremath{\,v \sin i}} % v sin i
33: \newcommand{\Msun}{\ensuremath{\,{\rm M}_\odot}} % Solar mass symbol
34: \newcommand{\Rsun}{\ensuremath{\,{\rm R}_\odot}} % Solar radius symbol
35: \newcommand{\Lsun}{\ensuremath{\,{\rm L}_\odot}} % Solar luminosity symbol
36: \newcommand{\lsun}{\ensuremath{L_\odot}} % Maths solar luminosity symbol
37: \newcommand{\Teff}{\ensuremath{T_{\rm eff}}} % Effective temperature symbol
38: \newcommand{\Vsync}{\ensuremath{V_{\rm synch}}} % Synchronous rotational velocity symbol
39: \newcommand{\kms}{km\,s$^{-1}$} % km/s symbol
40: \newcommand{\msec}{m\,s$^{-1}$} % km/s symbol
41: \newcommand{\kmsmag}{\,km\,s$^{-1}$\,mag$^{-1}$} % km/s/mag symbol
42: \newcommand{\mc}[1]{\multicolumn{2}{c}{#1}} % common multicolumn command
43: \newcommand{\Mbolsun}{\ensuremath{M_{\rm bol \odot}}} % Solar bolometric magnitude symbol
44: \newcommand{\Mbol}{\ensuremath{M_{\rm bol}}} % Bolometric magnitude symbol
45: \newcommand{\sci}[2]{{:1}\!\times10^{:2}} % This creates maths scientific notation
46: \newcommand{\Vsys}{\ensuremath{V_\gamma}} % Systemic recession velocity symbol
47: \newcommand{\mumag}{$\mu$mag}
48: \newcommand{\etal}{et~al.}
49: \newcommand{\ea}{et~al.}
50: \newcommand{\ie}{i.e.}
51: \newcommand{\eg}{e.g.}
52: \newcommand{\etc}{{\em etc.}}
53: \newcommand{\cf}{{cf.}}
54: \newcommand{\teff}{$T_{\rm eff}$}
55: \newcommand{\logg}{$\log g$}
56: \newcommand{\topp}{\emph{top}}
57: \newcommand{\midd}{\emph{middle}}
58: \newcommand{\bott}{\emph{bottom}}
59: \newcommand{\upp}{\emph{upper}}
60: \newcommand{\loo}{\emph{lower}}
61: \newcommand{\rii}{\emph{right}}
62: \newcommand{\lee}{\emph{left}}
63: \newcommand{\panel}{\emph{panel}}
64: \newcommand{\panels}{\emph{panels}}
65: \newcommand{\inset}{\emph{inset}}
66: \newcommand{\insets}{\emph{insets}}
67: \newcommand{\smeisymb}{$\times$}
68: \newcommand{\wiresymb}{small dots}
69: \newcommand{\specsymb}{$\bullet$}
70: \newcommand{\cday}{c\,day$^{-1}$}
71: \newcommand{\mhz}{$\mu$Hz}
72: \newcommand{\micmag}{$\mu$mag}
73: \newcommand{\aumi}{$\alpha$~UMi}
74: \newcommand{\aumii}{$\alpha$~Ursae~Minoris}
75: \newcommand{\pol}{Polaris} % $\alpha$~UMi}
76: \newcommand{\polaris}{Polaris}
77: %% \newcommand{\kms}{\ifmmode{\rm km\thinspace s^{-1}}\else km\thinspace s$^{-1}$\fi}
78: \newcommand{\dss}{$\delta$~Scuti}
79: \newcommand{\betacep}{$\beta$~Cep}
80: \newcommand{\str}{Str\"omgren}
81: \newcommand{\ebop}{{\sc ebop}}
82: \newcommand{\wink}{{\sc wink}}
83: \newcommand{\mad}{{\sc mad}}
84: \newcommand{\cles}{{\sc cl\'es}}
85: \newcommand{\dinshaw}{D89}
86: \newcommand{\hatzes}{H00}
87: \newcommand{\spec}{{\sc ast}}
88: \newcommand{\virgo}{{\sc virgo}}
89: \newcommand{\wire}{{\sc wire}}
90: \newcommand{\uves}{{\sc uves}}
91: \newcommand{\smei}{{\sc smei}}
92: \newcommand{\vlt}{{\sc vlt}}
93: \newcommand{\nasa}{{\sc nasa}}
94: \newcommand{\eso}{{\sc eso}}
95: \newcommand{\saao}{{\sc saao}}
96: \newcommand{\coriolis}{{\sc coriolis}}
97: \newcommand{\hipp}{{\sc hipparcos}}
98: \newcommand{\vizier}{{\sc vizier}}
99: \newcommand{\simbad}{{\sc simbad}}
100: \newcommand{\kepler}{{\sc kepler}}
101: \newcommand{\corot}{{\sc corot}}
102: \newcommand{\period}{{\sc period04}}
103:
104: % Joel Eaton:
105: \newcommand{\Fei}{Fe\,{\sc i}}
106: \newcommand{\kmps}{km~s$^{-1}$}
107:
108: %% You can insert a short comment on the title page using the command below.
109: \slugcomment{Accepted by ApJ - 2008 April 21}
110: \shorttitle{Polaris the Cepheid returns}
111: \shortauthors{Bruntt et al.}
112:
113: \begin{document}
114:
115: %% LaTeX will automatically break titles if they run longer than
116: %% one line. However, you may use \\ to force a line break if
117: %% you desire.
118:
119: \title{Polaris the Cepheid returns:\\$4.5$ years of monitoring from ground and space}
120:
121: \author{H.~Bruntt\altaffilmark{1} and N.~R.~Evans\altaffilmark{2} and D.~Stello\altaffilmark{1} and A.~J.~Penny\altaffilmark{3} and J.~A.~Eaton\altaffilmark{4} and D.~L.~Buzasi\altaffilmark{5} and D.~D.~Sasselov\altaffilmark{6} and H.~L.~Preston\altaffilmark{5} and E.~Miller-Ricci\altaffilmark{6}}
122:
123: \altaffiltext{1}{School of Physics, University of Sydney, NSW 2006, Australia}
124: \altaffiltext{2}{Smithsonian Astrophysical Observatory, 60 Garden St., Cambridge, MA 02138, USA}
125: \altaffiltext{3}{University of St Andrews, School of Physics and Astronomy, North Haugh, St Andrews, KY16 9SS, United Kingdom}
126: \altaffiltext{4}{Center of Excellence in Information Systems, Tennessee State University, Nashville, TN 37203, USA}
127: \altaffiltext{5}{Department of Physics, US Air Force Academy, Colorado Springs, CO 80840, USA}
128: \altaffiltext{6}{Astronomy Department, Harvard University, 60 Garden St., Cambridge, MA 02138, USA}
129:
130:
131: \begin{abstract}
132: We present the analysis of $4.5$ years of nearly continuous
133: observations of the classical Cepheid \polaris,
134: which comprise the most precise data available for this star.
135: We have made spectroscopic measurements from ground and photometric measurements from
136: the \wire\ star tracker and the \smei\ instrument on the \coriolis\ satellite. Measurements
137: of the amplitude of the dominant oscillation ($P=4$\,d), that go back more than a century,
138: show a {\it decrease} from $A_V=120$ mmag to $30$ mmag around the turn of the millennium.
139: It has been speculated that the reason for the decrease in
140: amplitude is the evolution of \polaris\ towards the edge of the instability strip.
141: However, our new data reveal an {\it increase} in the amplitude by $\sim30\%$ from 2003--2006.
142: It now appears that the amplitude change is cyclic rather than monotonic,
143: and most likely the result of a pulsation phenomenon.
144: In addition, previous radial velocity campaigns have claimed the detection
145: of long-period variation in \polaris\ ($P>40$\,d).
146: Our radial velocity data are more precise than previous datasets,
147: and we find no evidence for additional variation for periods
148: in the range 3--50\,d with an upper limit of 100\,\msec.
149: However, in the \wire\ data we find evidence of variation
150: on time-scales of 2--6 days, which we interpret as being due to granulation.
151: \end{abstract}
152:
153: %% Keywords should appear after the \end{abstract} command. The uncommented
154: %% example has been keyed in ApJ style. See the instructions to authors
155: %% for the journal to which you are submitting your paper to determine
156: %% what keyword punctuation is appropriate.
157:
158: \keywords{
159: stars: individual: \aumi\ (HD~8890; Polaris) --
160: stars: variables: Cepheids
161: }
162:
163: %% Change in amplitude from SMEI fit:
164: %% 1. april 2003 = 52730
165: %% 31. dec 2006 = 54100
166: %% print, (10.94+[1,0,-1]*0.08 + (2.56e-3+[-1,0,1.]*0.13e-3) * (52730.50-53000D ) ) * 2.
167: %% print, (10.94+[1,0,-1]*0.08 + (2.56e-3+[-1,0,1.]*0.13e-3) * (54100.50-53000D ) ) * 2.
168:
169:
170: %% Authors who wish to have the most important objects in their paper
171: %% linked in the electronic edition to a data center may do so by tagging
172: %% their objects with \objectname{} or \object{}. Each macro takes the
173: %% object name as its required argument. The optional, square-bracket
174: %% argument should be used in cases where the data center identification
175: %% differs from what is to be printed in the paper. The text appearing
176: %% in curly braces is what will appear in print in the published paper.
177: %% If the object name is recognized by the data centers, it will be linked
178: %% in the electronic edition to the object data available at the data centers
179: %%
180: %% Note that for sources with brackets in their names, e.g. [WEG2004] 14h-090,
181: %% the brackets must be escaped with backslashes when used in the first
182: %% square-bracket argument, for instance, \object[\[WEG2004\] 14h-090]{90}).
183: %% Otherwise, LaTeX will issue an error.
184:
185: % --------------------------------------------------------------
186: \section{Introduction\label{sec:intro}}
187: % --------------------------------------------------------------
188:
189: In addition to being arguably the most famous and in practice useful star
190: other than the Sun, \polaris\ has a number of properties
191: that may provide insights that are important to stellar astrophysics.
192: It is the nearest and brightest classical Cepheid,
193: oscillating in a single mode of pulsation with a period around four days.
194: The \hipp\ parallax constrains its luminosity and
195: allowed \cite{feast97} to argue that the mode of pulsation
196: must be the first overtone, which is upheld by the recent
197: reevaluation of the \hipp\ data \citep{leeuwencepheid07}.
198: As discussed by \cite{evans02}, \polaris\ has a number of
199: unusual pulsation properties, including a very
200: small pulsation amplitude, and, as for other overtone pulsators,
201: it has a rapid period change. It has been clear from studies as far
202: back as \cite{par56} and \cite{sza77} that overtone
203: pulsators (``s Cepheids" as they were then called) had
204: rapid period changes, more rapid than can be
205: explained by evolution during a second or third crossing of the instability strip.
206:
207: It was found by \cite{ferro83} that the main period
208: of \polaris\ is increasing (316\,s per century) and that its peak-to-peak
209: amplitude has decreased significantly, from about 140 to 70 mmag (Johnson $B$ filter),
210: based on photoelectric measurements collected in the period 1930--1980.
211: \cite{kamper98} analysed radial velocity measurements from 1900 to 1998
212: and could confirm the decrease in amplitude.
213: However, their radial velocity data from the end of the period
214: showed that the decrease had apparently stopped.
215: \cite{turner05} analysed both radial velocity and photometry data and
216: found evidence for a sudden change in oscillation period around 1963--1966,
217: and that the amplitude change became steeper at the same time.
218:
219: Our aim is to shed light on the properties of the oscillation of \polaris,
220: based on the analysis of a nearly continuous dataset spanning $4.5$ years.
221: We have photometry from two satellite missions and simultaneous
222: spectroscopic monitoring from the ground, each dataset
223: being superior to previous data available for \polaris.
224:
225:
226:
227:
228: % ___________________________________________________________________________________________
229: %
230: % \clearpage
231: \begin{table}
232: \caption{Observing log. The point-to-point
233: scatter, $\sigma$, is given in mmag (\wire\ $+$ \smei\ photometry) and \kms\ (spectroscopy).}
234: \label{tab:data} \centering
235: \begin{tabular}{l l l r r | l} \hline \hline
236: & & & \multicolumn{1}{c}{} & \multicolumn{1}{c}{Data} & \\
237: Source & Date start & Date end & \multicolumn{1}{c}{ $T_{\rm obs}$\,[d]} & \multicolumn{1}{c}{points} & $\sigma$ \\ \hline
238: \wire & 2004 Jan 17 & 2004 Feb 15 & 29.0 & 41\,162 & 0.88 \\ %% 7 cycles
239: \wire & 2004 Jul 10 & 2004 Jul 30 & 20.2 & 33\,937 & 0.30 \\ %%
240: \wire & 2005 Jan 31 & 2005 Feb 12 & 12.6 & 14\,240 & 0.29 \\ %% 3 cycles
241: \smei & 2003 Apr 6 & 2006 Dec 31 & 1365.4 & 13\,543 & 6.4 \\ \hline %%
242: \spec & 2003 Dec 25 & 2007 Oct 26 & 1401.0 & 517 & 0.15 \\ %%
243: \dinshaw&1987 May 6 & 1988 Jan 3 & 241.0 & 175 & 0.44 \\ %% Dinshaw et al. 1989 -- 0.42 telescope
244: \hatzes &1991 Nov 21 & 1993 Aug 2 & 640.4 & 42 & 0.07 \\ %% Hatzes \& Cocran 2000
245:
246: \hline
247: %% \multicolumn{6}{l}{$^*$Uncertainties given in mmag (photometry) and km/s (spectroscopy).}
248:
249: \end{tabular}
250: \end{table}
251: % \clearpage
252: % ___________________________________________________________________________________________
253: %
254:
255:
256: %--------------------------------------------------------------
257: % Program: .r wire_polaris_lc
258: % Phased plot included on 12 Nov 2007
259: \begin{figure*}
260: \centering
261: \includegraphics[width=14.5cm]{f1x.eps}
262: \caption{
263: The photometry datasets from \smei\ and \wire\ and the radial velocities from \spec.
264: The \bott\ \panel\ is an expanded view of 40 days of observations
265: surrounding the last run with \wire.
266: The thick gray curve is the fit of a single sinusoid to the \smei\ data.
267: \label{fig:lc}}
268: \end{figure*}
269: % \clearpage
270: %
271: %______________________________________________________________
272:
273:
274:
275:
276:
277: % --------------------------------------------------------------
278: \section{Observations and data reduction}
279: \label{sec:obs}
280: % --------------------------------------------------------------
281:
282: \polaris\ (\aumii) was observed with the star tracker on the Wide-field InfraRed Explorer (\wire; \citealt{bruntt07wire}) satellite
283: in January and February 2004, July 2004, and February 2005.
284: These runs lasted about 4, 3 and 2 weeks, respectively.
285: In addition, \polaris\ was monitored using 3.8 years of
286: nearly continuous photometry from the \smei\ instrument on the \coriolis\ satellite.
287: These observations were obtained between April 2003 and the end of 2006.
288: We further used the 2\,m Tennessee State University Automatic Spectroscopic Telescope (AST; \citealt{eaton04, eaton07})
289: to collect 517 high dispersion spectra of \polaris\ over a period of $3.8$ years,
290: from late 2003 to late 2007.
291: A log of the photometric and radial velocity observations is given in Table~\ref{tab:data}.
292: We also list datasets from two previous radial velocity campaigns that
293: we have used for comparison in our analysis in Sect.~\ref{sec:rv}
294: (D89: \citealt{dinshaw89} and H00: \citealt{hatzes00}).
295:
296: The complete photometric light curve from \smei\ and \wire\
297: and the radial velocity data from \spec\ are shown
298: in the \topp\ \panel\ in Fig.~\ref{fig:lc}.
299: Note that different units for the photometry (mmag) and velocities (\kms)
300: are given on the left and right abscissa, respectively.
301: The right abscissa is adjusted
302: by the ratio of the measured amplitudes in the \spec\ spectroscopy
303: and the \smei\ photometry.
304: The \bott\ \panel\ in Fig.~\ref{fig:lc} shows the details
305: of the variation during 40 days.
306: It shows the last run done with \wire\ and illustrates
307: the typical coverage with \smei\ and the \spec.
308: The thick gray curve is the fit of a single sinusoid to the complete \smei\ dataset.
309:
310: % Scaling SMEI to AST:
311: % print, 0.998 / 12.63e-3, '+-', sqrt( (0.014 / 12.63e-3)^2 + (0.998*0.13e-3/(12.63e-3)^2. )^2. ),$
312: % format='(F5.1,A3,F5.1)'
313:
314: % ___________________________________________________________________________________________
315: %
316: %% .r wire_polaris_oc.pro ; updated 6 Nov 2007
317: %% readcol,'/export/brixx1/bruntt/papers/bruntt/polaris/ocdata.polaris.nov07.dat',tt,ett,ee,ss,format='d,d,i,i'
318: %% s = sort(tt) & tt=tt(s) & ett=ett(s) & ss=ss(s) & ee=ee(s)
319: %% ww = where(tt gt 53380. and tt lt 53440,cc)
320: %% for i=0,cc-1 do print,tt(ww(i)),ett(ww(i)),ee(ww(i)),ss(ww(i)),format='(D12.5,D12.5,I6,I3)'
321: % \clearpage
322: \begin{table}
323: \caption{Times of maximum light in the \smei\ and \wire\ datasets.}
324: \label{tab:oc} \centering
325: \begin{tabular}{ccc}
326: \hline \hline
327: \multicolumn{1}{c}{$t_E$} & \multicolumn{1}{c}{Epoch} & Source \\
328: \multicolumn{1}{c}{[HJD\,$-2,453,000$]} & & \\ \hline
329:
330: %% Times updated 27/12/2007 by HB: now from sine-fit.
331:
332: $ 391.327 $ & 6329 & \smei \\
333: $ 395.339 $ & 6330 & \smei \\
334: $ 399.387 $ & 6331 & \smei \\
335: $ 403.323 $ & 6332 & \wire \\
336: $ 407.274 $ & 6333 & \wire \\
337: $ 411.262 $ & 6334 & \wire \\
338: $ 415.418 $ & 6335 & \smei \\
339: $ 419.160 $ & 6336 & \smei \\
340: $ 423.215 $ & 6337 & \smei \\
341: $ 427.152 $ & 6338 & \smei \\
342:
343: \hline
344: \end{tabular}
345: \end{table}
346: % \clearpage
347: % ___________________________________________________________________________________________
348: %
349:
350:
351:
352: The \wire\ dataset consists of about three million CCD stamps
353: extracted from the 512$\times$512 CCD SITe star tracker camera.
354: Each window is 8$\times$8 pixels and we carried out aperture photometry
355: using the pipeline described by \cite{bruntt05}.
356: The resulting point-to-point scatter range from $0.3$ to $0.9$ mmag (see Table~\ref{tab:data}).
357: The noise is somewhat higher than the Poisson noise,
358: \eg\ the high noise in the first \wire\ dataset is due to a high background sky level.
359:
360: Data from the \spec\ consist of \'echelle spectra covering
361: the wavelength range $5\,000$--$7\,100$ \AA\ at a resolution of about $30\,000$.
362: The velocities are derived from
363: the correlation between the observed spectrum and a list of 74 mostly \Fei\ lines.
364: More details about the pipeline used to extract the radial velocities are given by \cite{eaton07}.
365: We find that the velocities deviate slightly from
366: the IAU velocity system ($\Delta v=-0.35\pm0.09$\,\kmps).
367: The drift of the velocities due to the motion of \polaris\ in its binary orbit \citep{kamper96}
368: was subtracted before the time-series analysis was carried out.
369:
370:
371: The Solar Mass Ejection Imager (\smei) \citep{eyles03, jackson04} is a
372: set of three cameras mounted on the side of the \coriolis\ satellite.
373: Each camera covers a $3\times60$ degree strip of sky and this is
374: projected onto $1260\times40$ pixels of a CCD. As the nadir-pointing
375: satellite orbits every 101 minutes in its Sun-Synchronous polar orbit
376: the cameras take continuous 4-second exposures, which result in roughly
377: 4,500 images per orbit covering most of the sky. After bias and dark removal and
378: flat-fielding processing, the pixels about the target star are selected
379: from about ten images that contain the star for each orbit. These pixels are
380: then aligned and a standard PSF is fitted by least-squares.
381: Thus, the result is one brightness measure every 101 minutes.
382: There are problems with particle hits and a complex PSF, which shows
383: temporal variations that are not fully understood.
384: The short-term ($t<1.0$\,d) accuracies
385: are of the order of a few mmag, but there are longer term systematic
386: errors of the order of 10 mmag on the timescales of days and months,
387: which are not yet well understood. Allowing for time allocated for
388: calibrations and satellite operation, a time coverage of about 85\% was
389: maintained.
390:
391: %--------------------------------------------------------------
392: % Program: .r wire_polaris_phase
393: % .r wire_polaris_omc.pro
394: %--------------------------------------------------------------
395: % \clearpage
396: \begin{figure}
397: \centering
398: \includegraphics[width=8.8cm]{f2x.eps}
399: \caption{$O-C$ diagram. The solid line is a fit to the data taken after 1965.
400: The \inset\ shows the details of the new $O-C$ data from \smei\ and \wire.
401: \label{fig:oc}}
402: \end{figure}
403: % \clearpage
404: %
405: %______________________________________________________________
406:
407:
408:
409: % ___________________________________________________________________________________________
410: %
411: \section{Period change of the main mode\label{sec:oc}}
412: % ___________________________________________________________________________________________
413: %
414:
415: Observed epochs of maximum light in \polaris, spanning more than a century,
416: have indicated a significant change in period \citep{ferro83, fernie93, turner05}.
417: These studies have considered $O-C$ diagrams\footnote{$O-C$ diagrams
418: display observed times of maximum light minus calculated times,
419: assuming a constant period, plotted vs.\ time.}
420: and fitted a parabola, which is equivalent
421: to assuming the period changes linearly with time.
422: We have combined new measured epochs with those from \cite{turner05}.
423: A sudden change in the period has been noted in the 1960's and
424: therefore we have only used epochs observed since 1965.
425:
426:
427: % The results indicate a departure from a parabola in the 1960's,
428: % and therefore we have combined our new data
429: % with epochs from \cite{turner05} collected since 1965.
430:
431: We measured 223 and 16 epochs of maximum light in the \smei\ and \wire\ datasets, respectively.
432: From the $O-C$ analysis by \cite{ferro83} we expected the period to
433: decrease by 14 seconds over the $4.5$ years time span of our dataset.
434: Thus, when predicting the epochs of maximum light we could safely assume a constant period.
435: The epochs were predicted by fitting a single sinusoid to the \smei\ data.
436: The fit is of the form $S(t)=A \sin [2 \pi (f\,(t-t_0) + \phi) ]$, where
437: $A$ is the amplitude\footnote{Previous studies on \polaris\ have used peak-to-peak amplitudes.
438: We use amplitudes unless otherwise specified.}, $t$ is the time (Heliocentric Julian Date),
439: $t_0$ is the zero point, $P=1/f$ is the period, and $\phi$ is the phase.
440: The result is $P = 3.972111\pm0.000054$~days and
441: phase $\phi=0.2158\pm0.0022$ for the zero point $t_0=2,530,000$.
442: This fit is shown with a thick gray line in the \bott\ \panel\ in Fig.~\ref{fig:lc}.
443: The uncertainties were determined from simulations as discussed in Sect.~\ref{sec:uncert}.
444:
445: %% freq = 0.2517553D & efreq = 34D * 1e-7 & print,1D / freq, efreq/freq^2.,format='(D16.6,D16.6)'
446:
447: At each epoch, $E$, we selected all data points within half a period ($|t-E*P|<P/2$).
448: For the \smei\ data we required that at least 30 data points
449: were available with at least 10 data points before and after the maximum.
450: To estimate the time of maximum light, we fitted these data with a sinusoid
451: with fixed frequency and amplitude, while the phase was a free parameter.
452: In Table~\ref{tab:oc} we list these times, including the epoch number
453: following the definition by \cite{turner05}, and the source of the data.
454: The complete table is available in the on-line version of the paper, while
455: the times listed here correspond to the time interval covered in the \bott\ \panel\ in Fig.~\ref{fig:lc}.
456:
457: Since the quality of the three datasets are very different,
458: we computed weights based on the uncertainty of the epoch times, $t_E$.
459: This was determined by calculating the point-to-point scatter
460: after subtracting a parabola fitted to the $O-C$ data of each dataset.
461: The uncertainty on $t_E$ from \cite{turner05} is $\sigma(t_E) = 0.18$\,d while
462: for \smei\ and \wire\ data the uncertainties are 0.12\,d and 0.014\,d.
463: Finally, we calculate the $O-C$ values using weights, $1 / \sigma^2(t_E)$,
464: and the most accurate value of the period, which is the one from \smei:
465: $O-C = (t_E-t_0) - (E-E_0) \cdot P_{\rm SMEI}$.
466: We chose the reference epoch to be $E_0=6333$, and using the fit to the \smei\ data
467: found above, this corresponds to the time $t_0 = 2,453,407.277\pm0.009$.
468:
469: Our final $O-C$ diagram is shown in Fig.~\ref{fig:oc}.
470: The recent data continue to match the smooth period change as indicated by the weighted parabolic fit,
471: corresponding to a period increase of $dP/dt=365\pm27$~sec/century.
472: We note that the most recent \smei\ data lie systematically below the fit, which is
473: tightly constrained by the extremely well-defined times from \wire.
474: Note that if we only use data more recent than 1985,
475: we conclude that the period has not changed, \ie\ $dP/dt = -19\pm95$~sec/century.
476:
477: %%% Time of epoch 6333 found from the global fit to all SMEI data:
478: %%% phase = 0.2158D
479: %%% period = 3.972111D
480: %%% ephase = 0.0022D
481: %%% eperiod = 0.000054D
482: %%% t0 = 53000D ; + period * 102D ; 102 epochs since t0
483: %%% n = 103D ; 103 cycles completed to get to epoch 6333 since time T = 53000
484: %%% ; Uncertainty on the epoch time:
485: %%% varr = (eperiod/4.)^2. + (eperiod^2. * n)^2. + (ephase*period)^2. + (phase * eperiod)^2.
486: %%% print,'HJD(Epoch 6333)=',(- 1D / 4D + n - phase) * period + t0,'+-', sqrt(varr),'d',format='(A16,D9.3,A2,D5.3,A1)'
487:
488: %
489: %--------------------------------------------------------------
490: % Program: .r wire_lc_residuals ; new 2 nov 2007
491: % \clearpage
492: \begin{figure*}
493: \centering
494: \includegraphics[width=14cm]{f3ax.eps}
495: \includegraphics[width=14cm]{f3b.eps}
496: \caption{Amplitude spectra of the \spec\ radial velocity data
497: and the photometric data from \smei\ and \wire.
498: The \topp\ \panel\ is for the raw data and the \bott\ \panel\ is after
499: subtracting the main mode at $\simeq0.25$\,\cday.
500: The \insets\ show details of the amplitude spectra.
501: \label{fig:amp}}
502: \end{figure*}
503: % \clearpage
504: %
505: %______________________________________________________________
506:
507:
508: % \clearpage
509: \begin{table}
510: \caption{Results of the frequency analysis for the five datasets.
511: \label{tab:per}} \centering
512: \begin{tabular}{llcc} \hline \hline
513: Source & \multicolumn{1}{c}{$f$ [\cday]} & \multicolumn{1}{c}{$a$ [mmag/\kms]} & \multicolumn{1}{c}{$\phi$ [0..1]} \\ \hline
514:
515: \wire & $0.251732(14) $ & $12.26(18) $ & $0.2195(34)$ \\
516: \smei & $0.2517553(34)$ & $12.63(13)$ & $0.2158(22)$ \\
517: \spec & $0.2517506(43)$ & $ 0.998(14)$ & $0.1550(34)$ \\ \hline
518: \dinshaw & $0.25174(14)$ & $ 0.808(58)$ & \\
519: \hatzes & $0.251722(22)$ & $ 0.781(22)$ & \\
520: \hline
521:
522: % \wire~Feb'04 & $0.251468(99) $ & $12.510(81)$ & $0.2317(39)$ \\ %% Prelimn. uncertainty for Feb'04
523: % \wire~Jul'04 & $0.25103(32) $ & $11.92(12)$ & $0.360(66)$ \\
524: % \wire~Feb'05 & $0.25166(27) $ & $12.599(86)$ & $0.25(11)$ \\
525:
526: \end{tabular}
527: \end{table}
528: % \clearpage
529: %% -----------------------------------------------------------------------
530: %% -----------------------------------------------------------------------
531: \section{Amplitude change of the main mode}
532: \label{sec:fou}
533: %% -----------------------------------------------------------------------
534: %% -----------------------------------------------------------------------
535:
536: To analyse the time series in the frequency domain,
537: we calculated the Fourier amplitude spectra for each dataset.
538: In the \topp\ \panels\ in Fig.~\ref{fig:amp} we show the
539: amplitude spectra for the photometry from the \wire\ (red)
540: and \smei\ (green) and the radial velocities from \spec\ (black).
541: The \inset\ shows the prominent peak around 0.25\,\cday, corresponding to the known 4-day period.
542: Due to the long gaps between the three \wire\ datasets the spectral
543: window shows a more complex pattern than the nearly continuous datasets
544: from \smei\ and \spec.
545:
546: We fitted a single sinusoid to the dominant peak
547: and the results for each dataset are given in Table~\ref{tab:per}.
548: We list the phase of the \wire, \smei\ and \spec\ datasets,
549: relative to the zero point $t_0 = 2,453,000$.
550: It is seen that the frequencies are in very good agreement.
551: Although the point-to-point precision of the \wire\ data is superior,
552: the long span of the \smei\ and \spec\ datasets means
553: the frequency is determined more accurately by a factor of 3--4.
554: We also note that the phases from the fit to
555: the \smei\ and \wire\ photometry are in good agreement,
556: but the phase of the fit to the \spec\ data is quite different.
557: This corresponds to a shift in the times of maximum
558: in the flux and radial velocity:
559: ${\rm HJD}(F_{\rm max})={\rm HJD}(RV_{\rm max})+(1.744\pm0.016\,{\rm d})$.
560: In comparison, \cite{moska00} found the offset to be $1.682\pm0.017\,{\rm d}$
561: using combined photometry and radial velocity data from \cite{kamper98}.
562: The shifts appear to be marginally different ($2.7\,\sigma$),
563: but we note that the combined \cite{kamper98} data is calibrated to Johnson $V$,
564: while \smei\ has a filter response roughly centered at Johnson $R$ but being wider \citep{tarrant07}.
565:
566: % \cite{kamper98} found the value $1.652$\,d, but since no uncertainty
567: % was given by them, we cannot determine if there has been a shift in the delay.
568:
569: We have subtracted the main oscillation mode
570: and the residual amplitude spectra are shown in the \bott\ \panel\ in Fig.~\ref{fig:amp}.
571: All spectra show a higher amplitude towards lower frequencies.
572: This could be either due to instrumental drift or long-period variation intrinsic to \polaris,
573: which we discuss in Sect.~\ref{sec:rv}.
574: The \lee\ \inset\ shows that the peaks in the \smei\ data
575: do not coincide with peaks in the \spec\ data,
576: so we cannot claim they are due to coherent pulsations.
577: The \rii\ \inset\ shows the details around $0.25$\,\cday, where
578: two significant residual peaks are seen in the \smei\ and \spec\ datasets.
579: They are an indication that the frequency, amplitude or phase of
580: the oscillation has changed during the $\sim4$ years of observation.
581:
582: %
583: %--------------------------------------------------------------
584: % Program: .r wire_polaris_ampvar -- run program once --> polaris_ampvar_fixed_phase_freq.eps
585: % \clearpage
586: \begin{figure}
587: \centering
588: \includegraphics[width=8.3cm]{f4.eps}
589: \caption{Increase in the amplitude of the 4\,d main mode measured over $4.5$ years
590: with \wire\ (\wiresymb), \smei\ (\smeisymb), and \spec\ (\specsymb).
591: \label{fig:ampvar}}
592: \end{figure}
593: % \clearpage
594: %
595: %______________________________________________________________
596:
597:
598:
599: To test the possible change in either frequency, amplitude or phase of the
600: main mode, we divided the datasets into subsets.
601: Each of the \smei\ and \spec\ subsets contain 30 pulsation cycles,
602: and the three \wire\ datasets were analysed independently.
603: Each subset is fitted by a single sinusoid and we analysed the output parameters.
604: We made the ana\-lysis for two different assumptions:
605: \begin{itemize}
606: \item We assumed that the frequency, amplitude and phase change with time, so each is a free parameter.
607: \item We assumed that only the amplitude changes with time, so the frequency and phase were held fixed.
608: \end{itemize}
609:
610: The first assumption allowed us to test the stability of the mode.
611: We found that within the uncertainties both the frequency and phase
612: do not change over the time span of the observations.
613: However, the amplitude increases monotonically, and
614: this result is confirmed by the analysis under the second assumption.
615: A comparison of the measured amplitudes under the two assumptions
616: give nearly identical amplitudes for all subsets.
617:
618: In Fig.~\ref{fig:ampvar} we show the result obtained
619: under the second assumption, \ie\ for fixed frequency and phase.
620: The \smeisymb\ symbols are the \smei\ data,
621: the \specsymb\ symbols are the \spec\ data,
622: and \wiresymb\ are for the three \wire\ runs.
623: The dashed line is a weighted linear fit to the \smei\ data yielding
624: the amplitude change in mmag: $A_{\rm phot}(t) = (10.94\pm0.08) + (2.56\pm0.13)\cdot10^{-3} (t-t_0) $,
625: where $t$ is the HJD with zero point $t_0=2,453,000$.
626: The weighted fit to the \spec\ data (solid line) gives:
627: $A_{RV}(t) = (0.90\pm0.01) + (1.45\pm0.15)\cdot10^{-4} (t-t_0)$\,\kms.
628: From this analysis we find that from 2003--7 the amplitude measured
629: in flux and radial velocity has increased by $34\pm2$\% and $24\pm3$\%, respectively.
630: The single mode in \polaris\ has very low amplitude and from linear theory it is
631: expected that the rate of increase is the same in photometry and radial velocity.
632: The measured increase over four years is only marginally different ($8.1\pm3.1$\% or $2.6\,\sigma$),
633: and continued monitoring is required to confirm this tentative result.
634:
635: %% The rates of increase are only marginally different ($8.1\pm3.1$\% or $2.6\,\sigma$),
636: %% and continued monitoring is required to confirm this tentative result.
637:
638: % Confirm the
639: % a = 1.34 & b = 1.24 & ea = 0.02 & eb = 0.03
640: % rate = (a-b)/b & erate2 = (ea/b)^2. + (a*eb/b^2)^2.
641: % print,rate,sqrt(erate2),rate/sqrt(erate2),format='(3F6.3)'
642:
643: % Rate of increase:
644: % y4 = 4 * 365.24
645: % print,'Spec:',(0.90+ y4 * 1.45e-4)/0.90,'+-',sqrt(0.01^2.+(y4*0.15e-4)^2.)/0.90,format='(A8,F5.2, A2,F5.2)'
646: % print,'Smei:',(10.94 + y4 * 2.56e-3)/10.94,'+-',sqrt(0.08^2.+(y4*0.13e-3)^2.)/10.94,format='(A8,F5.2, A2,F5.2)'
647:
648:
649: %
650: % print, ' Spec: ', (0.89 + y4 * 2.01e-4) / 0.89
651: % print, ' Smei: ', (10.9 + y4 * 2.31e-3) / 10.9
652:
653: % Fit to \smei\ (top) and AST (bottom)
654: % 0.0109238 2.31193e-06
655: % 0.0106597 2.41239e-06
656:
657: In Fig.~\ref{fig:kamper} the measured peak-to-peak amplitudes
658: in radial velocity from \spec\ are compared with results from the literature.
659: \cite{kamper98} found a monotonic decrease in the amplitude
660: over the past 100 years\footnote{\cite{kamper98} converted
661: their radial velocity amplitudes
662: to photometric amplitudes using an empirical factor 50\,\kms\,mag$^{-1}$.
663: The ratio of the amplitudes measured in our radial velocity and the \smei\ photometric data
664: is $A_{\rm RV}/A_{\rm phot}=79.0\pm1.4$\,\kms.
665: The reason for the value being higher is that the filter response
666: of \smei\ is roughly Johnson $R$ while \cite{kamper98} used Johnson $V$.}
667: To avoid confusion about the conversion factor, we compare our results directly in radial velocity,
668: and we have reproduced their fit as the solid black line.
669: The dashed line is an extrapolation, and as noted by \cite{kamper98},
670: zero amplitude would be predicted in the year 2007.
671: However, from four years of monitoring (1993.15--1996.96) in radial velocity,
672: \cite{kamper98} found that the peak-to-peak amplitude was nearly constant.
673: This is in agreement with the amplitudes we find from the analysis (see Sect.~\ref{sec:specres})
674: of the datasets by \cite{dinshaw89} and \cite{hatzes00}.
675: From the most recent high-precision data, it is evident that the amplitude in \polaris\
676: was constant in the period 1987--1997, while the increase we report here
677: marks a new era in the evolution of \polaris.
678:
679: %
680: %--------------------------------------------------------------
681: % Program: .r wire_polaris_ampvar -- run program twice to get: Polaris_Ampl_Change2.ps
682: %--------------------------------------------------------------
683: % \clearpage
684: \begin{figure}
685: \centering
686: \includegraphics[width=8.9cm]{f5.eps}
687: \caption{Peak-to-peak radial velocity amplitude for the main 4\,d mode.
688: The solid line is the fit from \cite{kamper98}.
689: The \inset\ shows the most recent data in detail.
690: Our measurements of the amplitude
691: based on the \cite{dinshaw89} and \cite{hatzes00} are marked by arrows.
692: \label{fig:kamper}}
693: \end{figure}
694: % \clearpage
695: %
696: %______________________________________________________________
697:
698:
699: \subsection{Uncertainties in frequency, amplitude and phase\label{sec:uncert}}
700:
701: The uncertainties in the frequency, amplitude and phase in Table~\ref{tab:per}
702: are determined from realistic simulations of each dataset.
703: The times of observation are taken from the observations and
704: the simulations take into account the increase in noise towards low frequencies.
705: We used the approach described by \cite{bruntt07m67} (their Appendix~B)
706: to make 100 simulations of each dataset,
707: and the uncertainties are the \rms\ value on the extracted frequency, amplitude and phase.
708:
709: We made simulations of the \wire, \smei\ and \spec\ datasets
710: and also the radial velocity datasets from \cite{dinshaw89} and \cite{hatzes00},
711: which we use for comparison in Sect.~\ref{sec:specres}.
712: In all cases the uncertainties are somewhat higher than theoretical
713: estimates that assume the noise to be white \citep{mont99}.
714: This is especially the case for the \wire\ photometry due
715: to low-amplitude variation on timescales comparable to the 4\,d period,
716: as will be discussed in Sect.~\ref{sec:wireres}.
717:
718:
719: %
720: % --------------------------------------------------------------
721: % Program:
722: % .r wire_lc_residuals.pro (last plot)
723: % --------------------------------------------------------------
724: % \clearpage
725: \begin{figure}
726: \centering
727: \includegraphics[width=8.8cm]{f6.eps}
728: \caption{Residual amplitude spectra of the three radial velocity datasets
729: from \spec\ (black), Hatzes \& Cochran (2000; blue), and Dinshaw et al. (1989; red).
730: The low frequency peaks claimed in the previous studies are marked by arrows.
731: \label{fig:specres}}
732: \end{figure}
733: % \clearpage
734: %
735: % --------------------------------------------------------------
736: %
737:
738:
739:
740:
741: % --------------------------------------------------------------
742: \section{Evidence for additional intrinsic variation}
743: \label{sec:rv}
744: % --------------------------------------------------------------
745:
746:
747: In the following we will look
748: at the evidence for variation in \polaris\ beyond the 4\,d main mode.
749: The analysis is based on the \spec\ and \wire\ datasets and
750: data from two previous radial velocity campaigns.
751: The \spec\ dataset is the most accurate at low frequencies,
752: and can be used to look for long periods (Sect.~\ref{sec:specres}).
753: The \wire\ dataset consists of three week-long runs with very high precision
754: and can be used to study low-amplitude variation (Sect.~\ref{sec:wireres}).
755: The noise in the \smei\ dataset is distinctly non-white,
756: as mentioned briefly in Sect.~\ref{sec:obs}.
757: Therefore, we did not use this dataset in the following investigations,
758: except for the subtraction of the 4\,d mode, where \smei\ provides the
759: most accurate frequency and phase.
760:
761: \subsection{Long-period variation\label{sec:specres}}
762:
763: In addition to the 4\,d mode,
764: a few studies have claimed variation at long periods, \eg\
765: $P=9.75$\,d \citep{kamper84}, $45.3\pm0.2$\,d \cite{dinshaw89},
766: and $40.2\pm0.7$\,d and $17.2$\,d \citep{hatzes00}.
767: To confirm these claims we reanalysed
768: the original datasets of \cite{dinshaw89} and \cite{hatzes00}.
769: The basic properties of these datasets are listed in Table~\ref{tab:data}.
770:
771: In Fig.~\ref{fig:specres} we compare the amplitude spectra of the spectroscopic
772: data from \spec, \cite{dinshaw89} and \cite{hatzes00},
773: after subtracting the 4\,d period and the long-period trend due to the binary orbit.
774: The \spec\ spectrum was calculated taking
775: the amplitude increase into account (\cf\ Sect.~\ref{sec:fou}).
776: As a result, the residual double peak seen at $\sim 0.252$\,\cday\ is no longer visible
777: (compare Fig.~\ref{fig:specres} with the \bott\ \panel\ in Fig.~\ref{fig:amp}).
778: In the \spec\ amplitude spectrum we see no significant peaks
779: from $0.02$--$0.3$\,\cday\ ($P=3$--$50$\,days).
780: We set an upper limit on the amplitude at 100\,\msec,
781: which is four times the average level in the amplitude spectrum in this frequency interval.
782:
783:
784: The dataset by \cite{hatzes00} consists
785: of 42 data points distributed unevenly over 640 days,
786: providing a very complicated spectral window.
787: The dataset comprises very precise radial velocities collected with an iodine cell
788: as a reference (\rms\ residuals are 70\,\msec).
789: \cite{hatzes00} detected two peaks with almost equal amplitude at
790: $f_{\rm H00}^{\rm A}=0.0249$ and $f_{\rm H00}^{\rm B}=0.0581$\,\cday.
791: In the amplitude spectrum in Fig.~\ref{fig:specres}
792: we find only one of these peaks to be significant
793: ($f_{\rm H00}^{\rm B}=0.05834\pm0.00014$\,\cday\ with amplitude $A=163\pm25$\,\msec).
794: We made 100 simulations of the data including this frequency,
795: white noise, and a $1/f$ noise component consistent with the observations.
796: The inserted $f_{\rm H00}^{\rm B}$ frequency was only recovered in 49 of the 100 simulations.
797: Interestingly, in the other simulations the highest peaks
798: were clustered close to $f_{\rm H00}^{\rm A}$ ($0.025\pm0.007$\,\cday\ with amplitude $A=167\pm22$\,\msec),
799: indicating that the two peaks detected by \cite{hatzes00} are due to the complicated spectral window.
800:
801: The dataset by \cite{dinshaw89} comprises 175 velocities distributed over 241 days.
802: The precision is 440\,\msec,
803: as estimated from the \rms\ of the residuals after subtracting the 4\,d mode.
804: In addition to this mode we detect a peak at $1.02199\pm0.00024$\,\cday\ with strong aliases at $\pm1$\,\cday.
805: \cite{dinshaw89} found one of the alias peaks ($f_{\rm D89}=0.0221\pm0.0001$\,\cday)
806: to be the highest and suggested that it was intrinsic to \polaris.
807: We measure the amplitude to be $A=0.542\pm0.067$\,\kms, but
808: such a strong signal is not seen in the more recent dataset
809: by \cite{hatzes00} or in any of our datasets (see Figs.~\ref{fig:amp} and \ref{fig:specres}).
810: For these reasons we believe that this peak is a 1\,\cday\ artifact,
811: likely caused by instrumental drift in combination with the spectral window.
812:
813:
814: %%% The dataset by \cite{dinshaw89} comprises 175 velocities distributed over 241 days.
815: %%% The precision is 440\,\msec,
816: %%% as estimated from the \rms\ of the residuals after subtracting the 4\,d mode.
817: %%% In addition to this mode we detect a peak at $1.02199\pm0.00024$\,\cday\
818: %%% with strong aliases at $\pm1$\,\cday. The amplitude is $A=0.542\pm0.067$\,\kms.
819: %%% Such a strong signal was not seen by \cite{hatzes00} or in our datasets
820: %%% (see Figs.~\ref{fig:amp} and \ref{fig:specres}).
821: %%% \cite{dinshaw89} made a thorough analysis of the data and claimed
822: %%% that the alias peak at $f_{\rm D89}=0.0221\pm0.0001$\,\cday\
823: %%% was intrinsic to \polaris.
824: %%% We inserted this long-period signal in 100 simulations and recovered it in all cases.
825: %%% We cannot exclude that this variation was present in \polaris\ in 1987,
826: %%% but it is not recovered in the two other (more precise) radial velocity datasets we have analysed.
827: %%% We are inclined to believe that the variation
828: %%% is due to a 1\,\cday\ artifact, likely caused by
829: %%% the spectral window in combination with instrumental drift.
830:
831:
832: %
833: % --------------------------------------------------------------
834: % Program:
835: % Se POLARIS_RAPPORT.txt ``17. December 2007: new plots''
836: % --------------------------------------------------------------
837: % \clearpage
838: \begin{figure}
839: \centering
840: \includegraphics[width=8.8cm]{f7x.eps}
841: \caption{Residuals in the three \wire\ light curves of \polaris\
842: and the comparison star HD~166205 (large $\bullet$ symbols offset by 3\,mmag).
843: The dashed curve is a sinusoid with the 4\,d period and an amplitude
844: set arbitrarily to 2\,mmag.
845: \label{fig:wireres}}
846: \end{figure}
847: % \clearpage
848: %
849: % --------------------------------------------------------------
850: %
851:
852:
853: %
854: % --------------------------------------------------------------
855: \subsection{Low-amplitude variation \label{sec:wireres}}
856: % --------------------------------------------------------------
857: %
858:
859: The residuals in the \wire\ light curves,
860: after having subtracted the 4\,d period, are shown in Fig.~\ref{fig:wireres}.
861: The drifts seen in the \wire\ light curves correspond to periods from 2--6\,d,
862: \ie\ similar to the 4\,d main mode. Before interpreting this variation,
863: it is important to investigate if these variations
864: are due to improper subtraction of the main period or due to instrumental drift.
865:
866: To test the first caveat, we compared the residuals
867: when subtracting the fit to the \smei\ data and
868: when subtracting fits to the individual \wire\ datasets.
869: We find that the residuals are quite similar, and the conclusions reached here
870: are not affected by the adopted approach.
871: In the following, we have used the most accurate
872: value for the frequency and phase, which is from the fit of the main mode to the \smei\ data,
873: while we used the amplitudes fitted to the individual \wire\ datasets.
874:
875: The second caveat is instrumental drift and we tested this by using a comparison star.
876: During the observations with the \wire\ star tracker
877: four other stars were monitored on the same CCD.
878: However, not all are suited as comparison stars:
879: HD~5848 is a bright ($V=4.3$) K~giant star clearly showing solar-like oscillations \citep{stello08}.
880: HD~51802 and HD~174878 are faint ($V=5.1$ and $6.6$) M~giants showing variation with long periods.
881: The only suitable comparison star is the A1\,V star HD~166205 ($\delta$~UMi; $V=4.4$).
882: In Fig.~\ref{fig:wireres} we also show the
883: light curve of this star with large $\bullet$ symbols.
884: The star is two magnitudes fainter than \polaris,
885: so we binned the data collected within each orbit ($P_{\rm orbit}\simeq93$\,min)
886: to be able to see any low-amplitude variation.
887:
888: % --------------------------------------------------------------
889: % Program:
890: % .r wire_polaris_granulation.pro
891: % --------------------------------------------------------------
892: % \clearpage
893: \begin{figure}
894: \centering
895: \includegraphics[width=8.8cm]{f8.eps}
896: \caption{Power density spectrum (PDS)
897: of the residual light curves from \wire\ shown in Fig.~\ref{fig:wireres}.
898: The dashed line is a scaling of the granulation measured in the Sun.
899: The arrows mark the 4\,d main period and
900: frequencies that are due to the \wire\ orbit.
901: \label{fig:density}}
902: \end{figure}
903: % \clearpage
904: % --------------------------------------------------------------
905:
906: The clear variation in the residuals of the \wire\ photometry is not
907: seen in the hot comparison star HD~166205. The fact that it is seen in three
908: light curves obtained with \wire\ at three different epochs spanning one year gives
909: us some confidence that the signal is intrinsic to \polaris.
910: We will discuss two possible explanations for the observed variation here,
911: namely granulation and star spots.
912:
913: To investigate in detail how the variation in the \wire\ light curves
914: depends on frequency, we show their power density spectra (PDS) in Fig.~\ref{fig:density}
915: in a logarithmic plot.
916: The virtue of the PDS is that one can directly compare the properties of
917: datasets, which have different temporal coverage and noise characteristics.
918: Each PDS shows a clear increase from the white noise level around $10$\,mHz
919: towards low frequencies.
920: The arrows mark the harmonics of the orbital frequency of \wire\ ($f_{\rm WIRE}=178$\,\mhz),
921: one third of the \wire\ orbital frequency, and 2 c/day. These frequencies
922: are observed in almost all \wire\ datasets, and are due to a combination
923: of a low duty cycle (typically 20--40\%) and scattered light from earth shine.
924: Also marked is the location of the 4\,d mode ($f=2.91$\,\mhz), which has been subtracted.
925:
926: To see if granulation could explain the increase in the PDS towards low
927: frequencies, we have used a scaling of the granulation observed in the Sun.
928: This is based on \virgo\ satellite observations of the Sun-as-a-star,
929: and the scaling is done both in terms of amplitude and timescale,
930: following Kjeldsen \& Bedding (in preparation; see also \citealt{stello07}).
931: This prediction is shown as the dashed line in Fig.~\ref{fig:density}.
932: This is a scaling over three orders of magnitude in luminosity from
933: the Sun to the super giant \polaris, so it is intriguing
934: that the prediction of the granulation signal
935: agrees with the observations within a factor of $3.0\pm0.5$ in amplitude.
936:
937: Both \cite{dinshaw89} and \cite{hatzes00} discussed evidence for
938: star spots on \polaris. We argue that this is not the cause of the observed
939: variation in the \wire\ data, since the time-scale is only a few days.
940: That would imply very rapid rotation, which is unlikely for such an evolved star.
941:
942: % --------------------------------------------------------------
943: \section{Discussion and conclusion}
944: % --------------------------------------------------------------
945:
946: From $4.5$ years of intensive monitoring of \polaris\ we find that
947: the amplitude of the 4-day main mode has increased steadily by about 30\% in
948: both radial velocity and flux amplitude.
949: The rate of increase in the amplitude from 2003--2007 is slightly steeper than the decrease from the
950: fit by \cite{kamper98} for the period 1980--1994.
951: Other sources have also found a recent increase in the amplitude of \polaris\
952: (\citealt{engle04}; Turner 2007, private communication).
953:
954: The result that the amplitude of \polaris\ is now increasing has implications
955: for earlier explanations of the change in amplitude as a cessation of pulsation
956: due to the Cepheid's evolution towards the edge of the instability strip.
957: At the very least, whatever the process is,
958: it is not a simple monotonic progression through the HR diagram,
959: unless this stage of evolution is more complex than previously thought.
960: The recovery of the amplitude suggests that the phenomenon is cyclic.
961: As such, it is likely to be associated in some way with pulsation
962: rather than with evolution.
963:
964: A possible explanation for the increase in amplitude could be the beating
965: of two closely spaced modes \citep{breger06}. To exhibit a beat period as long
966: as the amplitude variation of \polaris\ (likely a few 100 years), the modes
967: would have to be very closely spaced indeed. Among classical Cepheids,
968: amplitude variation is extremely uncommon. The only case where it is
969: firmly established is in V\,473~Lyrae \citep{burki86}, where the variation
970: occurs over only a few years. In RR~Lyrae stars,
971: the Blazhko effect is well known, and thought to be the result of mode beating,
972: although it is not completely understood. Since the amplitude changes in \polaris\
973: are well established, they require further observations and
974: consideration theoretically to unravel the cause.
975:
976: There are other interesting aspects to the main oscillation mode.
977: The characterization of the period change is puzzling
978: since analysis of $O-C$ diagrams
979: spanning more than a century reveal that the period change is not linear.
980: There may have been a ``glitch" in both the period change and the
981: amplitude in the mid-1960's \citep{turner05}, rather than smooth changes.
982: We measured 239 epochs of maximum light but the time span of the observations
983: of $4.5$ years is too short to investigate the period change.
984: This is another aspect of the pulsation that demands further observation.
985:
986: A few radial velocity campaigns have claimed the presence of
987: additional long-period variation in \polaris. We have analysed the original
988: datasets by \cite{dinshaw89} and \cite{hatzes00}
989: and we conclude that these long periods
990: are likely spurious detections caused by instrumental drifts \citep{dinshaw89}
991: or insufficient data leading to a complicated spectral window \citep{hatzes00}.
992: From our 3.8\,yr of monitoring with \spec, we set an upper limit
993: on the variation in radial velocity at 100\,\msec\ for periods in the range 3--50\,d
994: (except for the 4\,d main mode).
995:
996: In the \wire\ data we find evidence of low-amplitude variation (peak-to-peak $2$\,mmag)
997: at time scales of 2--6 days, which are likely intrinsic to \polaris.
998: We applied a simple scaling of observed solar values for the characteristic
999: timescale and amplitude of the granulation.
1000: Although this is a scaling over three orders of magnitude in luminosity,
1001: the prediction agrees with the observed variation in \polaris\ within a factor $3.0\pm0.5$.
1002: Thus, we conclude that
1003: the variation in the \wire\ data could be due to granulation.
1004:
1005: % --------------------------------------------------------------
1006: \acknowledgments
1007: % --------------------------------------------------------------
1008:
1009: HB and DS are supported by the Australian Research Council.
1010: NRE and AJP acknowledge support from NASA contracts NAS8-03060 and
1011: NNG~05GA41G, respectively. Operation of the Tennessee State University
1012: Automatic Spectroscopic Telescope was supported by grants from NASA
1013: (NCC5-511) and NSF (HRD 9706268). \smei\ was designed and constructed by
1014: a team of scientists and engineers from the at University of California
1015: at San Diego, Boston College, Boston University, and the University of
1016: Birmingham.
1017: We thank A.~Buffington, C.~J.~Eyles and S.~J.~Tappin for advice
1018: on the \smei\ data reduction.
1019: The following internet-based resources were used for this paper:
1020: the \nasa\ Astrophysics Data System
1021: and the ar$\chi$iv scientific paper preprint service operated by Cornell University.
1022:
1023: % --------------------------------------------------------------
1024: \bibliography{ms}
1025: % --------------------------------------------------------------
1026:
1027: % --------------------------------------------------------------
1028: \end{document}
1029: % --------------------------------------------------------------
1030: