1: \documentclass[12pt,preprint]{aastex}
2: \begin{document}
3: \title{Stellar Activity and the Str\"{o}mgren Photometric Metallicity
4: Calibration of Intermediate-Type Dwarf Stars}
5:
6: \bigskip
7: \author{Sarah L. Martell and Graeme H. Smith}
8: \affil{University of California Observatories/Lick Observatory, University of
9: California}
10: \affil{Santa Cruz, California 95064}
11: \bigskip
12: \begin{abstract}
13: \noindent
14: We consider the effect of stellar activity, as measured by X-ray
15: luminosity, on metallicities of Solar-neighborhood F and G dwarfs derived
16: from Str\"{o}mgren photometry. Rocha-Pinto \& Maciel found evidence
17: that Str\"{o}mgren colors systematically underpredict [Fe/H] for stars
18: with extremely high \ion{Ca}{2} H \& K emission. We investigate
19: whether a recent photometric metallicity calibration derived by
20: Martell \& Laughlin might be subject to this effect, and whether the
21: amount of underprediction could reliably be expressed as a function of
22: $\log (L_{\rm X}/L_{\rm bol})$. Among those calibration stars used by
23: Martell \& Laughlin which are also in the Bright Star Catalogue and
24: detected in the \textit{ROSAT} All-Sky Survey there is no evidence for
25: a correlation between photometric metallicity and stellar activity.
26: However, many of the ``very active stars'' on which the Rocha-Pinto \&
27: Maciel result was based are members of interacting binaries or are
28: very young in age, and are not included in the X-ray sample that we
29: are using. Among normal dwarf stars it appears that stellar activity
30: has little effect on the metallicity calibration of Str\"{o}mgren colors.
31:
32: \end{abstract}
33:
34: \keywords{stars: activity --- stars: metallicity}
35:
36:
37: \section{Introduction}
38:
39: The effect of chromospheric activity on photometric techniques for measuring
40: the metallicities of stars has been explored by several authors commencing
41: with Giampapa, Worden, \& Gilliam (1979). With recent
42: large photometric and spectroscopic surveys the question can be addressed
43: in a statistically meaningful way (see e.g., West et al. 2004).
44: Rocha-Pinto \& Maciel (1998) correlated the calcium emission line
45: index $\log R_{HK}^{\prime}$ against metallicities calculated from
46: Str\"{o}mgren photometry, and found that their most-active stars had
47: surprisingly low values of inferred ${\rm [Fe/H]_{phot}}$.
48:
49: In highly-active stars, the equivalent width of metallic absorption lines
50: can be reduced by chromospheric emission in the lines (see e.g., Basri
51: et al. 1989). For extremely-active stars this effect may reduce the
52: Str\"{o}mgren $m_1$ index, which is intended to measure line
53: blanketing (Crawford 1975), leading to a falsely low photometric
54: metallicity. Rocha-Pinto \& Maciel (1998) appealed to an
55: activity-$m_1$ correlation to explain the apparent low photometric
56: metallicity (and corresponding apparent old age) of the most-active
57: nearby stars as an artifact of their chromospheric activity. In this
58: paper we make a similar comparison, adopting as an indicator of
59: stellar activity the soft X-ray luminosity measured by the
60: \textit{ROSAT} satellite, and using a photometric metallicity
61: calibration (Martell \& Laughlin 2002, hereafter ML02) which was
62: developed to be more accurate for higher-metallicity stars than the
63: Schuster \& Nissen (1989, hereafter SN89) calibration.
64:
65:
66: \section{The metallicity data set}
67:
68: In ML02 the authors used a set of 664 F, G, and K dwarfs located within
69: 100 pc of the Sun to derive an empirical relation between Str\"{o}mgren
70: photometric indices and metallicity.
71: The selection criteria for the ML02 ``calibration stars'' are as follows:
72: they are the members of the Cayrel de Strobel, Soubiran,
73: \& Ralite (2001) compilation of [Fe/H] abundances which have absolute
74: magnitudes $M_V > +1.0$, \textit{Hipparcos} parallaxes
75: greater than 0.01\arcsec, and which are also in the Hauck-Mermilliod (1998)
76: compilation of Str\"{o}mgren photometry. The calibration used a
77: Levenberg-Marquardt algorithm (e.g., Press et al. 1992) to find the
78: coefficients for a general third-order polynomial relating [Fe/H] to
79: the Str\"{o}mgren indices $(b-y)$, $m_1$, and $c_1$. When the
80: distributions of the residuals (i.e., ${\rm [Fe/H]_{spec} -
81: [Fe/H]_{phot}}$) for the ML02 and SN89 calibrations were compared,
82: the former was found to be more accurate. This can be seen in
83: Figure 1, which shows a Gaussian fit to each distribution. Both the
84: central offset and the half-width at half-maximum of the fits are
85: smaller for the ML02 calibration.
86:
87: We have modified slightly the methodology of ML02: in the Cayrel de
88: Strobel et al. (2001) metallicity catalog, many stars have multiple [Fe/H]
89: measurements, which were treated as separate objects for the purposes of
90: the ML02 polynomial fitting. For the present work we
91: averaged such multiple [Fe/H] values together, both to reduce the effect of
92: outlying measurements, and to prevent multiple-counting of stars in
93: our histograms and plots. For stars with multiple observations, we
94: took the observational error in [Fe/H] to be the standard deviation in
95: the mean, calculated from those multiple measurements. For stars with
96: single observations, or where the standard deviation was zero, we
97: adopted as the error the mean of the standard deviation for all of the
98: multiply-observed stars. That quantity has a value of 0.0824 dex. We
99: then refitted the ML02 Str\"{o}mgren photometry-metallicity relation,
100: and while the values of the coefficients did change, the overall
101: quality of the fit stayed roughly constant.
102: The same residual-distribution test was done as in ML02,
103: and the results are also shown in Figure 1. The Gaussian fits to the
104: new residual distribution (hereafter MS04) and that of ML02 are almost
105: equivalent: the center falls at $-0.0248$ for the ML02 calibration,
106: and at $-0.0266$ for the MS04 calibration. The HWHM for ML02 is
107: $0.0868$, and $0.0890$ for MS04. In the residual distribution for the
108: SN89 calibration, the center of the Gaussian fit is at $-0.0517$, and
109: the HWHM is $0.1097$, values which are clearly different from the ML02
110: and MS04 calibrations.
111:
112: The resulting calibration is
113:
114: \begin{eqnarray}
115: [{\rm Fe/H}]_{\rm phot} = &-&41.836891+153.92203(b-y)+53.678346m_1+129.01008c_1 \nonumber \\
116: &-&101.47843(b-y)^2+161.87500m_1^2-150.07528c_1^2-412.75949(b-y)m_1
117: \nonumber \\
118: &-&370.84617(b-y)c_1+52.187608m_1c_1-103.14707(b-y)^3+81.084037m_1^3
119: \nonumber \\
120: &+&53.244338c_1^3+651.10576(b-y)^2m_1+204.52658(b-y)^2c_1 \nonumber \\
121: &-&452.44692m_1^2(b-y)-80.536525m_1^2c_1+247.37448c_1^2(b-y) \nonumber \\
122: &-&90.169531c_1^2m_1+128.07586(b-y)m_1c_1 \nonumber \\
123: \end{eqnarray}
124:
125: The differences between photometric metallicities derived from this new
126: calibration and those of ML02 and Schuster \& Nissen (1989) are shown in
127: Figures 2 and 3 respectively as a function of the spectroscopic
128: metallicity of the calibrating stars. These diagrams give an
129: appreciation for the uncertainty in using these fitting functions to
130: derive [Fe/H] from Str\"{o}mgren colors. We should point out that the
131: calibration stars for ML02 and MS04 do not extend to metallicities as
132: low as the calibrators used by Schuster \& Nissen (1989), and these
133: former calibrations should only be employed over the range in [Fe/H]
134: shown in Figures 2 and 3.
135:
136: The change in coefficients between the ML02 and MS04 calibrations lead
137: us to investigate how strongly the coefficients depend on the assumed
138: errors in the spectroscopic [Fe/H] values of the calibrating stars.
139: If the errors are assumed to be the same for all stars, the
140: coefficients are insensitive to the value of the assumed error: they
141: are the same whether the assumed error is 0.05 dex, 0.10 dex, or even
142: 0.20 dex.
143:
144: By contrast, more complicated behavior resulted when we allowed the errors
145: in [Fe/H] to vary among the calibration stars. For each star with multiple
146: [Fe/H] measurements we calculated the mean [Fe/H] value, the standard
147: deviation $\sigma$ in these [Fe/H] values, and the standard deviation
148: in the mean $\sigma_m$. We then ran two different fits for the
149: calibration stars, assuming in both cases that the error in [Fe/H] for
150: each multiply-observed star was equal to the individual value of
151: $\sigma_m$ calculated for that star. These two fits differed in the
152: error adopted for all of the singly-observed stars; in the first case
153: this error was taken to be the average value of $\sigma$ from the
154: multiply-observed stars, while in the second case an error twice this
155: amount was adopted. It is the former of these fits that corresponds to
156: equation (1). The coefficients, as a rule, were larger for the
157: second fit. Most of the terms involving $(b-y)$ and $m_{1}$ stayed fairly
158: constant, although two of the largest changes were in the coefficients
159: of $(b-y)^{2}m_{1}$ and $(b-y)m_{1}^{2}$, which are the two largest
160: coefficients in the calibration.
161:
162: We also experimented with setting to zero the four coefficients whose
163: values varied the most with changes in the errors. This caused the
164: other coefficients to all decrease. However, the quality of the fits,
165: as measured by Gaussian parameters, stayed roughly constant as we
166: varied the assumed values of the errors, and all had centers closer to
167: zero and smaller HWHMs than the SN89 calibration.
168:
169: \section{The ROSAT data set}
170:
171: The \textit{ROSAT} observatory conducted an All-Sky Survey (RASS) of X-ray
172: sources (Voges et al. 1999) during 1990 and 1991 using the onboard
173: PSPC imaging detector (Pfeffermann et al. 1987). H\"{u}nsch, Schmitt, \&
174: Voges (1998) searched the RASS data for detections at the locations of all
175: main sequence and subgiant stars listed in the Bright Star Catalogue
176: (BSC; Hoffleit \& Warren 1991).
177: We sifted the ``calibration stars'' used by ML02
178: for those also included in the catalog of H\"{u}nsch, Schmitt, \& Voges
179: (1998). This produced a set of 146 stars that
180: we refer to as the ``RASS-BSC-calibration'' sample.
181: Figures 4 and 5 show the distribution of distances and spectroscopic
182: metallicities respectively for these stars.
183: Distances derived from the \textit{Hipparcos} Catalogue (ESA 1997),
184: together with a table of bolometric corrections from
185: Allen's Astrophysical Quantities (Cox 2000),
186: were used to convert the RASS fluxes in the 0.1 - 2.4 keV energy range
187: into X-ray luminosities and to calculate $\log (L_{\rm X}/L_{\rm bol})$.
188:
189: According to H\"{u}nsch et al. (1998), the RASS has a typical flux limit of
190: $10^{-13}$ erg cm$^{-2}$ s$^{-1}$.
191: This can readily be seen in Figure 6, which shows a plot of the X-ray
192: luminosities of the RASS-BSC-calibration stars versus distance.
193: Very few of these stars are more than 60 pc distant, and the
194: majority are within 25 pc of the Sun. We note for comparison with
195: Figure 6 that a solar-like dwarf with $M_V = 4.7$ at a distance of 20 pc
196: will have an apparent magnitude of $V = 6.2$, at the limit of the
197: Bright Star Catalogue. A star at this distance would require an X-ray
198: luminosity of $L_{\rm X} > 5 \times 10^{27}$ ergs s$^{-1}$ to be detected
199: in the RASS. Since the goal of this paper is to investigate whether the
200: Str\"{o}mgren metallicity calibration is compromised among
201: stars with high $\log (L_{\rm X}/L_{\rm bol})$,
202: the RASS flux limit may not adversely bias our analysis:
203: the most-active of our calibration
204: stars are above the RASS flux limit all the way
205: out to 100 pc. In the next section, we test for possible bias by
206: showing that the photometric trends for
207: the RASS-BSC-calibration stars within
208: 25 pc are the same as for the full sample.
209: However, we refrain from using the calibration data to form conclusions
210: about any possible dependency of X-ray activity on stellar metallicity: the
211: RASS flux limit biases against the presence of low-metallicity
212: stars in our sample because of their relatively low space density in the
213: Solar neighborhood compared to near-solar abundance stars.
214: Indeed, nearly all of the RASS-BSC stars found in our
215: Str\"{o}mgren-[Fe/H] calibration sample have metallicities of
216: [Fe/H] $> -0.5$.
217:
218:
219: \section{ Discussion }
220:
221: To investigate the question of whether chromospheric activity affects
222: the metallicity derived from Str\"{o}mgren photometry, we looked for
223: trends involving $[{\rm Fe/H]_{phot}}$, $[{\rm Fe/H]_{spec}}$, and
224: $\log (L_{\rm X}/L_{\rm bol})$ among the RASS-BSC-calibration stars.
225: Figure 7 shows the difference between $[{\rm Fe/H]_{spec}}$ and $[{\rm
226: Fe/H]_{phot}}$ derived from equation (1) as a function of $\log
227: (L_{\rm X}/L_{\rm bol})$ for the full set of RASS-BSC-calibration
228: stars out to 100 pc, and a best-fit line obtained by a regression of
229: $[{\rm Fe/H]_{spec}}$ -- $[{\rm Fe/H]_{phot}}$versus $\log (L_{\rm
230: X}/L_{\rm bol})$. The slope of the best-fit line is
231: 0.0121$\pm$0.0144, which is fairly consistent with there being no
232: trend. The linear Pearson correlation coefficient calculated for the
233: data is 0.0696. Figure 8 shows the same quantities for those
234: RASS-BSC-calibration stars within 25 pc, and the slope of that
235: best-fit line is 0.0210$\pm$0.0179. The correlation coefficient for
236: these data is 0.1287. Both correlation coefficients are quite small;
237: there appears to be no significant evidence that metallicities derived
238: from equation (1) are compromised by stellar activity among normal solar
239: neighborhood dwarf stars.
240:
241: We have conducted a similar analysis using the same set of stars but the
242: photometric metallicity calibration of Schuster \& Nissen (1989) rather
243: than equation (1). The results are shown in Figures 9 and 10, again for stars
244: within 100 pc and 25 pc of the Sun respectively. Once again there is no
245: evidence that the photometric metallicities depart from the spectroscopic
246: values in any way correlated with stellar activity. The correlation
247: coefficient is 0.0915 for the 100-pc set and 0.1775 for the 25-pc set.
248:
249: The lack of trends in Figures 7-10 is not surprising: the
250: H\"{u}nsch et al. (1998) stars in our calibration set have little overlap with
251: the Rocha-Pinto \& Maciel (1998) ``very active'' stars, among which the
252: authors find that stellar activity may affect the Str\"{o}mgren colors. Those
253: ``very active'' stars have $\log R_{HK}^{\prime} > -4.3$, and tend to be
254: either very young, or in close or interacting binary systems. We find that the
255: lower limit on X-ray luminosity for their ``very active'' category of stars is
256: $\log (L_{\rm X}/L_{\rm bol}) \approx -4.1$. There are
257: only five stars (out of 146) with such high X-ray luminosity in our
258: RASS-BSC-calibration set,
259: none of which are within 25 pc of the Sun.
260: Hence the sample of RASS-BSC stars that we have been using, as selected
261: from the compilation of H\"{u}nsch et al. (1998), avoids the uppermost end of
262: the stellar X-ray luminosity function.
263:
264: In summary, we find no evidence for a trend in
265: ${\rm [Fe/H]_{spec}-[Fe/H]_{phot}}$ with $\log (L_{\rm X}/L_{\rm bol})$,
266: regardless of whether ${\rm [Fe/H]_{phot}}$ is based on equation (1) or
267: the previous widely-used calibration of Schuster \& Nissen (1989).
268: We conclude that for most normal single stars, there is no need to apply a
269: correction for chromospheric activity to metallicity calibrations based on
270: Str\"{o}mgren photometry.
271:
272:
273: \newpage
274: \begin{thebibliography}{DUM}
275:
276: \bibitem[]{}
277:
278: \bibitem[]{}
279: Basri, G., Wilcots, E., \& Stout, N. 1989, \pasp, 101, 528
280:
281: \bibitem[]{}
282: Cayrel de Strobel, G., Soubiran, C., \& Ralite, N. 2001, \aap, 373, 159
283:
284: \bibitem[]{}
285: Cox, A.N. 2000, Allen's Astrophysical Quantities, 4th edition
286: (New York: Springer Verlag)
287:
288: \bibitem[]{}
289: Crawford, D. L. 1975, \aj, 80, 955
290:
291: \bibitem[]{}
292: ESA. 1997, The {\it Hipparcos} Catalogue ESA SP-1200 (Noordwijk: ESA)
293:
294: \bibitem[]{}
295: Giampapa, M. S., Worden, S. P., \& Gilliam, L. B. 1979, \apj, 229, 1143
296:
297: \bibitem[]{}
298: Hauck, B., \& Mermilliod, M. 1998, \aaps, 129, 431
299:
300: \bibitem[]{}
301: Hoffleit, D. E., \& Warren, W. H. Jr. 1991, The Bright Star Catalogue, 5th Rev.
302: (New Haven, Yale Univ.)
303:
304: \bibitem[]{}
305: H\"{u}nsch, M., Schmitt, J. H. M. M., Voges, W. 1998, \aaps, 132, 155
306:
307: \bibitem[]{}
308: Martell, S. L. \& Laughlin, G. 2002, \apjl, 577, L45
309:
310: \bibitem[]{}
311: Pfeffermann, E., et al. 1987, SPIE, 733, 519
312:
313: \bibitem[]{}
314: Press, W. H., et al. 1992, Numerical Recipes in Fortran: The Art of Scientific
315: Computing (Cambridge: Cambridge Univ. Press)
316:
317: \bibitem[]{}
318: Rocha-Pinto, H. J. \& Maciel, W. J. 1998, \mnras, 298,332
319:
320: \bibitem[]{}
321: Schuster, W. J. \& Nissen P. E. 1989, \aap, 221, 65
322:
323: \bibitem[]{}
324: Voges, W., et al. 1999, \aap, 349, 389
325:
326: \bibitem[]{}
327: West, A. A. et al. 2004, astro-ph/0403486
328:
329: \end{thebibliography}
330:
331: \newpage
332: \begin{center}
333: \large Figure Captions
334: \end{center}
335:
336: \figcaption{
337: Histograms of ${\rm [Fe/H]_{phot} - [Fe/H]_{spec}}$, for the current
338: calibration (MS04, solid line), ML02 (dotted line), and SN89 (dashed
339: line), together with Gaussian fits to each. These histograms provide
340: a visual representation of how well the various photometric
341: calibrations reproduce the spectroscopic metallicities. The
342: calibrations of MS04 and ML02 are quite similar, but that of SN89 fits
343: the calibration stars less well.
344: \label{fig1}}
345:
346: \figcaption{
347: The difference between
348: photometric metallicities derived from the current calibration (MS04) and
349: that of Martell \& Laughlin (ML02) versus
350: spectroscopic metallicity ${\rm [Fe/H]_{spec}}$ for the calibration stars.
351: \label{fig2}}
352:
353: \figcaption{
354: The difference between photometric metallicities derived from
355: the current calibration (MS04) and the standard calibration of Schuster \&
356: Nissen (SN89) versus ${\rm [Fe/H]_{spec}}$ for the MS04 calibration stars.
357: \label{fig3}}
358:
359: \figcaption{
360: Distribution of distances for the RASS-BSC-calibration stars.
361: \label{fig4}}
362:
363: \figcaption{
364: Distribution of ${\rm [Fe/H]_{spec}}$ for the RASS-BSC-calibration stars.
365: \label{fig5}}
366:
367: \figcaption{
368: Distance versus $\log L_{\rm X}$ for the RASS-BSC-calibration stars. The
369: solid line shows the maximum distance at which a star of a given $L_{\rm X}$
370: would have been detected in the \textit{ROSAT} All-Sky Survey.
371: \label{fig6}}
372:
373: \figcaption{
374: Difference between ${\rm [Fe/H]_{spec}}$ and photometric
375: metallicity, calculated using the MS04 calibration,
376: versus $\log (L_{\rm X}/L_{\rm bol})$ for
377: RASS-BSC-calibration stars.
378: \label{fig7}}
379:
380: \figcaption{
381: Difference between ${\rm [Fe/H]_{spec}}$ and photometric metallicity,
382: calculated using the MS04 calibration, versus $\log (L_{\rm X}/L_{\rm bol})$
383: for RASS-BSC-calibration stars within 25 pc.
384: \label{fig8}}
385:
386: \figcaption{
387: Difference between ${\rm [Fe/H]_{spec}}$ and photometric metallicity,
388: calculated using the SN89 calibration, versus $\log (L_{\rm X}/L_{\rm bol})$
389: for RASS-BSC-calibration stars.
390: \label{fig9}}
391:
392: \figcaption{
393: Difference between ${\rm [Fe/H]_{spec}}$ and photometric metallicity,
394: calculated using the SN89 calibration, versus $\log (L_{\rm X}/L_{\rm bol})$
395: for RASS-BSC-calibration stars within 25 pc.
396: \label{fig10}}
397:
398:
399: \end{document}
400:
401:
402: