1: \documentclass[12pt,preprint]{aastex}
2:
3: \newcommand{\ds}{\displaystyle}
4: \newcommand{\be}{\begin{equation}}
5: \newcommand{\ee}{\end{equation}}
6: \newcommand{\opr}{{\Lambda_{\rm op}}}
7:
8:
9: \begin{document}
10: \title{An Improved uvby-Metallicity Calibration for Metal-Rich Stars}
11:
12: \bigskip
13: \author{Sarah Martell, Gregory Laughlin}
14:
15: \bigskip
16: \affil{UCO/Lick Observatory, Department of Astronomy and Astrophysics, UCSC, Santa Cruz, CA 95064}
17:
18: \begin{abstract}
19: We present an improved uvby-metallicity relation calibrated for
20: F,G, and early K dwarfs,
21: and an analogous uvby-$T_{eff}$ relation, both derived using a
22: Levenberg-Marquardt minimization scheme. Our calibrations are based on 1533
23: stars which appear in both the Cayrel de Strobel (2001) metallicity
24: compilation, and in the Hauck-Mermilliod (1998) catalog of uvby photometry.
25: We also examine the speculative possibility
26: of using uvby photometry to produce a uvby-planeticity calibration.
27: We conclude that while there is likely no strong photometric indicator
28: of the presence or absence of short-period planets, the possibility of a
29: spectroscopic indicator of planeticity is well worth examining.
30:
31: \end{abstract}
32:
33: \keywords{stars: planetary systems ---
34: stars: metallicity}
35:
36: \section{Introduction}
37: The number of known extrasolar planets is approaching 100, and more planets
38: are being discovered every month.\footnote {For the latest catalog of planets,
39: see http://www.exoplanets.org}
40: A remarkable empirical correlation has emerged from the aggregate of planets:
41: The parent stars of the detected extrasolar planets appear to be significantly
42: metal-rich in comparison to field FGK dwarfs in the solar neighborhood.
43: This correlation was first extensively discussed by Gonzalez (1997),
44: and has been further studied by a number of authors; a partial list might
45: include Gimenez (2000), Laughlin (2000), Santos, Israelian \& Mayor (2001),
46: Murray \& Chaboyer (2001), and Reid (2002). These studies all agree that
47: the fraction of stars containing readily detectable extrasolar planets
48: (e.g. having $K>15$ m/s, and $P<5$ yr) increases substantially
49: with increasing stellar metallicity. A search strategy geared toward the
50: most metal-rich stars should thus detect short-period planets at a
51: far greater rate than a conventional volume-limited survey.
52:
53: One might argue that a metallicity-based planet search strategy will
54: result in a haphazard accumulation of short-period planets,
55: while simultaneously
56: skewing the overall census of planets in the solar neighborhood. Indeed, it is
57: important to continue with volume-limited surveys in order to round out the
58: unbiased statistical distribution of planetary systems.
59: The discovery of additional
60: short-period planets, however, has important ramifications. For the following
61: reasons, it is desirable to locate as many short-period planets as possible:
62: ({\bf 1}) The presence of a short-period planet is an excellent
63: indicator that additional planets are present in the system.
64: Currently, for example, 5 of 12 short-period planets
65: observed at Lick Observatory show evidence for additional companions
66: (Fischer et al 2001).
67: Recently, it has been realized that some multiple planet systems
68: such as GJ 876
69: allow for unambiguous determination of planet masses and orbital
70: parameters
71: (Laughlin \& Chambers 2001, Rivera \& Lissauer 2001).
72: ({\bf 2}) The current
73: census of short-period extrasolar planets shows an interesting concentration of
74: objects in the period range between roughly 2.98 and 3.52 days
75: (8 out of 20 planets with periods $P<20$ d). While
76: this pile-up is probably a consequence of disk-protoplanet tidal migration
77: (Lin, Bodenheimer \& Richardson 1996), the stopping mechanism is not at all
78: well understood. It is thus worthwhile to find more short-period planets to
79: better delineate the minimum planetary period. ({\bf 3})
80: Geometric arguments indicate
81: that 10\% of Hot Jupiter-type planets transit the face of the parent star.
82: A large increase in the rate of short-period planet detections
83: thus translates into
84: a significant increase in the number of transits discovered.
85: Transiting planets permit
86: accurate mass and radius determinations, and are extremely
87: important data points within
88: the overall theory of giant planets.
89:
90: High-resolution spectroscopy provides the most precise method for flagging
91: the metal-rich stars which have a high probability of bearing a detectable planet.
92: Unfortunately, the number of field stars with spectroscopically determined metallicities
93: is relatively small.
94: The most recent compilation from the literature by Cayrel de Strobel, Soubiran,
95: \& Ralite (2001; hereafter CSR) contains 6534 metallicity determinations for 3356 stars.
96: The majority of single, metal-rich, main sequence and subgiant FGK stars in
97: the CSR compilation are already under radial velocity survelliance.
98: A larger pool of high metallicity candidates is needed.
99:
100: A study by Laughlin (2000) used the Hipparcos database (Perryman et al
101: 1997), the Hauck-Mermilliod (1998) compilation of uvby photometry, and
102: the Schuster \& Nissen (1989; hereafter SN) uvby-[Fe/H] calibration to
103: to identify 200 metal rich
104: stars from a parent population of roughly 10,000 K,G,
105: and late F-type main sequence stars (having $7.8<V<10$)
106: in the immediate ($d<100{\rm pc}$) solar neighborhood. Nineteen
107: of these stars were added to the Keck Radial Velocity survey
108: in the Summer of 2000.
109: Three of these nineteen stars soon proved to be accompanied by planets
110: (Vogt et al 2001, Butler et al 2002).
111: Metallicity-based search strategies are
112: therefore clearly optimizing the planet-detection efficiency.
113:
114: The SN calibrations for F and G spectral types were based on 219 calibration
115: stars spanning a range of metallicities from
116: $-3.5 < [{\rm Fe/H}] < 0.4$. This very useful calibration,
117: which has an overall standard
118: deviation of 0.16 dex per star, is wholly empirical, and has the
119: advantage of using only the standard uvby indices, rather than the
120: differential indices, $\delta m_{1}=m_{1}^{hyades}-m_{1}^{star}$, and
121: $\delta c_{1} = c_{1}^{hyades}-c_{1}^{star}$, which are used in the
122: earlier calibrations of Crawford (1975) and Olsen (1984). The SN
123: calibration has been widely used to investigate metallicity trends
124: among solar neighborhood stars (e.g. Rocha-Pinto \& Maciel (1996), or
125: Haywood (2001). As recently emphasized by Twarog, Anthony-Twarog and Tanner
126: (2002), however, the SN calibration systematically underestimates
127: $[{\rm Fe/H}]$ by $\sim0.2$ dex for stars with spectroscopic metallicities
128: $[{\rm Fe/H}]>0.1$. The SN calibration is thus not optimally suited for
129: compiling an accurately ranked catalog of high metallicity stars.
130:
131: Our main goal in this letter is to make the uvby-metallicity calibration
132: more discerning and accurate among the highest metallicity
133: stars. We also use this forum to briefly introduce the
134: concept of a ``planeticity calibration'', in which stellar properties
135: are correlated directly with the detectable presence of an extrasolar
136: planet.
137:
138: \section{New Photometric Calibrations}
139:
140: To produce our new calibration, we
141: we queried the CSR database for all stars which have (1)
142: Hipparcos parallaxes, (2) uvby measurements in the Hauck-Mermilliod (1998)
143: compilations, (3) ${\rm M_V} < 1.0$, and (4) lie within 100 pc of the Sun
144: (to reduce the problems posed by reddening). This subset of the CSR
145: catalog contains 1533 measurements of 664 stars. Using a Levenberg-Marquardt
146: scheme (see e.g. Press et al 1992), we then optimized the coefficients,
147: $A_{n}$, of the most general third-order polynomial in the Str\"omgren
148: indices $(b-y)$, $m_{1}$, and $c_{1}$ so as to best reproduce the
149: spectroscopically determined $[{\rm Fe/H}]$ values. Because the Levenberg-Marquardt
150: routine provides local convergence, it is prudent to start the
151: iterative sequence with different sets of initial guesses for ${A_{n}}_{i}$
152: to better insure that ${A_{n}}$, the set of final coefficients produced by the
153: routine, corresponds to the global minimum. We ran the Levenberg-Marquardt
154: procedure using (1) ${A_{n}}_{i}$ from the SN ``F star'' calibration
155: (ignoring the logarithmic terms), (2) ${A_{n}}_{i}$ from the SN ``G star''
156: calibration, and (3) by setting all ${A_{n}}_{i}$ to unity. In each
157: case, the routine converged to an identical set of final coefficients,
158: suggesting that a global minimum has been achieved. Our calibration is:
159:
160: \begin{eqnarray}
161: [{\rm Fe/H}]_{phot} = &-&10.424602+31.059003(b-y)+42.184476m_1+15.351995c_1 \nonumber \\
162: &-&11.239435(b-y)^2-29.218135m_1^2-11.457610c_1^2-138.92376(b-y)m_1 \nonumber \\
163: &-&52.033290(b-y)c_1+11.259341m_1c_1-46.087731(b-y)^3+26.065099m_1^3 \nonumber \\
164: &-&1.1017830c_1^3+138.48588(b-y)^2m_1+39.012001(b-y)^2c_1 \nonumber \\
165: &+&23.225562m_1^2(b-y)-69.146876m_1^2c_1+20.456093c_1^2(b-y) \nonumber \\
166: &-&3.3302478c_1^2m_1+70.168761(b-y)m_1c_1 \nonumber \\
167: \end{eqnarray}
168:
169: Figure 1 shows our estimates of $[{\rm Fe/H}]_{phot}$ arising from this calibration
170: plotted versus $[{\rm Fe/H}]_{spec}$ for all 1533 calibration measurements.
171:
172: The CSR database has been compiled from a diverse array of metallicity
173: determinations, some of which have reported uncertainties, and some of
174: which do not. Furthermore, many of the 664 stars which survived our cuts
175: for inclusion into the calibration
176: are represented by multiple spectroscopic measurements. We circumvented
177: the non-uniform property of the CSR compilation by assuming that every
178: spectroscopic metallicity
179: measurement has an associated uncertainty of 0.05 dex. With
180: this assumption, each of the 1533 measurements contributes an equal weight to
181: the calibration. We compute a reduced $\chi^{2}$ statistic of 6.68 for
182: the overall fit, which is superior to the reduced $\chi^{2}$ value of
183: 9.33 which results from comparing $[{\rm Fe/H}]_{phot}$ obtained with
184: the SN calibrations with the spectroscopic metallicities.
185:
186: In Figure 2, we plot the distribution of metallicity differences between
187: the photometric and spectroscopic determinations,
188: $[{\rm Fe/H}]_{phot}-[{\rm Fe/H}]_{spec}$, for both our calibration
189: and also for the SN calibration. Both distributions are well
190: represented by Gaussians. The new calibration is superior to SN
191: in that it has a narrower half-width at half maximum (0.10 dex
192: as opposed to 0.13 dex), and is centered more closely to zero
193: (-0.027 dex as opposed to -0.049 dex). Figure 3 shows a similar
194: diagram in which we plot only the results of measurements which
195: yielded a spectroscopic metallicity
196: $[{\rm Fe/H}]>0.0$. Here again, our calibration is
197: significantly superior to SN, which is not surprising, as very few
198: high metallicity stars were included in the SN calibration, which was
199: geared primarily toward studies of population II stars.
200: Our calibration still tends to underestimate the metallicity of
201: the most metal-rich stars in comparison to spectroscopic determinations,
202: but the discrepancy has been significantly reduced. As a specific
203: example, the planet-bearing star HD 145675 (14 Her), which is the
204: highest metallicity star in our subset of the CSR catalog, with
205: $[{\rm Fe/H}]=0.5$ (Gonzalez, Wallerstein \& Saar 1999), receives a
206: metallicity estimate of [Fe/H]=0.47 from our calibration, and [Fe/H]=0.13
207: from SN.
208:
209: Our metallicity calibration should provide a useful sieve for selecting
210: candidate stars to be added to radial velocity surveys. We
211: note here a handful of class IV and V stars for which (1) our new calibration
212: finds $[{\rm Fe/H}]_{phot}>0.3$, which (2) do not appear in the Laughlin (2000)
213: catalog, and which (3) have no previously announced planet: HD 88176,
214: HD 112164, HD 15942, HD 144585.
215:
216: In addition to providing spectroscopic metallicities, the
217: CSR catalog also lists effective temperature estimates for each
218: of the cited entries. Following the same procedure and applying the
219: same cuts as for our our $[{\rm Fe/H}]_{phot}$ calibration, we have also derived
220: a $uvby-T_{eff}$ calibration,
221:
222: \begin{eqnarray}
223: [{\rm T_{eff}}]_{phot} = &&4434.8364+29839.391(b-y)-13967.518m_1-3041.0698c_1 \nonumber \\
224: &-&98624.604(b-y)^2-8963.3232m_1^2+10620.269c_1^2+63289.783(b-y)m_1 \nonumber \\
225: &-&10099.516(b-y)c_1+16204.483m_1c_1+91256.827(b-y)^3-45093.846m_1^3 \nonumber \\
226: &-&10580.081c_1^3-122571.68(b-y)^2m_1+16993.421(b-y)^2c_1 \nonumber \\
227: &+&111551.98m_1^2(b-y)-43418.471m_1^2c_1-5856.0002c_1^2(b-y) \nonumber \\
228: &+&15492.020c_1^2m_1-127.04874(b-y)m_1c_1 \nonumber \\
229: \end{eqnarray}
230:
231: If we assume that the average measurement uncertainty of the spectroscopic
232: $T_{eff}$ determinations is 50K, the reduced $\chi^{2}$ of this calibration
233: is 3.17, indicating that $uvby$ photometry is quite adequate to providing
234: effective temperature estimates for a given star. As before, we also
235: construct a distribution of differences ${T_{eff}}_{phot}-{T_{eff}}_{spec}$,
236: and construct a gaussian fit to the distribution.
237: The half-width at half maximum for this distribution is 80.4 K, whereas the
238: center lies at -11.5 K.
239:
240: \section{Toward a Planeticity Calibration}
241:
242: Given the demonstrated connection between the metallicity of a parent star and
243: the detectable presence of a planetary companion, and given the power of the
244: Levenberg-Marquardt minimization technique in producing empirical calibrations,
245: one can ask a rather provacative question: Is it possible to construct a
246: ``planeticity'' calibration in which the chance of finding a readily detectable
247: planet around a given star is estimated directly from uvby photometry?
248:
249: Cumming, Butler and Marcy (1999), have published a detailed analysis of 11
250: years of precision radial velocity measurements for 73 nearby stars. Within
251: this list of stars, there are currently 10 announced planets. We assign a planeticity ${\cal P}=1.0$ to stars with planets, and ${\cal P}=0.0$ to stars without. We then
252: use the Levenberg-Marquardt method to produce a uvby-planeticity calibration in
253: exact analogy to producing the metallicity and effective temperature calibrations.
254: This exercise leads to the following empirical calibration:
255: \begin{eqnarray}
256: [{\cal P}_{phot}] = &&168.92038-1367.1865(b-y)+1215.0485m_1-625.29367c_1 \nonumber \\
257: &+&2450.1524(b-y)^2+251.91349m_1^2+686.49412c_1^2-3024.1263(b-y)m_1 \nonumber \\
258: &+&3776.9231(b-y)c_1-3394.6622m_1c_1-838.40082(b-y)^3-347.90246m_1^3 \nonumber \\
259: &-&238.02961c_1^3+698.73217(b-y)^2m_1-3804.6842(b-y)^2c_1 \nonumber \\
260: &+&802.37174m_1^2(b-y)-454.17907m_1^2c_1-2429.7350c_1^2(b-y) \nonumber \\
261: &+&2468.8122c_1^2m_1+4574.5147(b-y)m_1c_1 \nonumber \\
262: \end{eqnarray}
263:
264: The results of this calibration are shown in Figure 4, and indicate that
265: the planeticity calibration is not particularly convincing. Nevertheless,
266: it is instructive to test the relation on a larger set of stars.
267: We queried the Hipparcos catalog for stars lying within 50 pc of the sun,
268: retaining those for which (1) uvby photometry exists, (2) fall
269: in the range $0.8 M_{\odot} < M_{\star} < 1.5 M_{\odot}$ (using the
270: stellar evolution models of Allende Prieto \& Lambert 1999), and which (3)
271: were not among the 73 Lick Survey stars used to build the planeticity calibration given above.
272: These cuts left 2056 stars, of which 46 have reported planets. We expect
273: that the majority of the remaining 2010 are currently under radial
274: velocity surveillance, and that relatively few easily detected planets
275: will emerge from this aggregate. Using our calibration, we compute
276: an average planeticity of 0.271 for the 46 known planet-bearing stars,
277: in comparison to an average planeticity of 0.113 for the 2010 stars
278: without reported planets.
279:
280: We conclude that uvby photometry likely holds little planeticity
281: information beyond that provided by the well-recognized metallicity
282: proxy. We are very optimistic, however, regarding the possibility
283: of producing a planeticity calibration which is based on spectroscopically
284: determined differential metal abundances within a particular star. Recent work, in particular by Gonzalez et al (2001), indicates that planet-bearing stars may have systematically unusual abundance ratios for $[{\rm Al/Fe}]$, $[{\rm Mg/Fe}]$, and $[{\rm Na/Fe}]$. Debra Fischer (2002, personal communication) has recently completed detailed spectroscopic abundance analyses of more than 500 stars which are currently maintained in the Keck and Lick planet search lists. In a future project, we are hoping to replace the uvby indices with spectroscopically determined differential abundances, and thus derive a more definitive planeticity calibration.
285:
286:
287: \subsection{Acknowledgements}
288:
289: We would like to thank Debra Fischer and Geoff Marcy for
290: useful conversations. We would also like to thank Craig Markwardt
291: for advice regarding his IDL routines {\it mpfit} \& {\it mpfitfun}. This
292: work was supported by faculty research funds granted by the
293: University of California, Santa Cruz.
294:
295: \begin{thebibliography}{DUM}
296:
297: \bibitem[]{}
298: Allende Prieto, C., \& Lambert, D. L. 1999, AA, 352, 555
299: `
300: \bibitem[]{}
301: Butler, R. P. et al. 2002, submitted to ApJ
302:
303: \bibitem[]{}
304: Cayrel de Stobel, G., Soubiran, C., \& Ralite, N. 2001, AA, 373, 159
305:
306: \bibitem[]{}
307: Crawford, D. L. 1975, AJ, 80, 955
308:
309: \bibitem[]{}
310: Cumming, A., Marcy, G. W., \& Butler, R. P. 1999, ApJ, 526, 916
311:
312: \bibitem[]{}
313: Fischer, D. 2002, personal communication
314:
315: \bibitem[]{}
316: Gim\'enez A. 2000, AA, 356, 213
317:
318: \bibitem[]{}
319: Gonzalez, G. 1997, MNRAS, 285, 403
320:
321: \bibitem[]{}
322: Gonzalez, G., et al. 2001, AJ, 121, 432
323:
324: \bibitem[]{}
325: Gonzalez, G., Wallerstein, G., \& Saar, S. H. 1999, ApJ, 511, L111
326:
327: \bibitem[]{}
328: Hauck, B. \& Mermilliod, M. 1998, AAS, 129, 431
329:
330: \bibitem[]{}
331: Haywood, M. 2001, MNRAS, 325, 1365
332:
333: \bibitem[]{}
334: Laughlin, G. 2000, ApJ, 545, 1064
335:
336: \bibitem[]{}
337: Laughlin, G. \& Chambers, J. E. 2001, ApJ, 551, L109
338:
339: \bibitem[]{}
340: Lin, D. N. C., Bodenheimer, P., \& Richardson, D. C. 1996, Nature, 380, 606
341:
342: \bibitem[]{}
343: Marcy, G.W., Cochran, W.D. \& Mayor, M. 2000,
344: in Protostars and Planets IV, ed. V. Mannings, A. P. Boss \&
345: S. S. Russell (Tucson: University of Arizona Press), p.1285.
346:
347: \bibitem[]{}
348: Murray, N., \& Chaboyer, B. 2002, ApJ, 566, 442
349:
350: \bibitem[]{}
351: Olsen, E. H. 1984, AAS, 57, 43
352:
353: \bibitem[]{}
354: Perryman, M.A.C., et al. 1997, The Hipparcos Catalog, A\&A, 323, L49
355:
356: \bibitem[]{}
357: Press, W. H., et al. 1992, Numerical Recipes in Fortran: The Art of Scientific
358: Computing (Cambridge: Cambridge Univ. Press)
359:
360: \bibitem[]{}
361: Reid, I. N. 2002, PASP, 114, 306
362:
363: \bibitem[]{}
364: Rivera, E. J, Lissauer, J. J. 2001, ApJ, 558, 392
365:
366: \bibitem[]{}
367: Rocha-Pinto, H. J., \& Maciel, W. J. 1996, MNRAS, 279, 447
368:
369: \bibitem[]{}
370: Santos, N. C., Israelian, G., \& Mayor, M. 2001, AA, 373, 1019
371:
372: \bibitem[]{}
373: Schuster, W. J., \& Nissen, P. E. 1989, A\&A, 221, 65
374:
375: \bibitem[]{}
376: Twarog, B. A., Anthony-Twarog, B., \& Tanner, D. 2002, AJ, 123, 2715
377:
378: \bibitem[]{}
379: Vogt, S. S., et al. 2002, ApJ, 568, 352
380:
381: \end{thebibliography}
382:
383: \newpage
384: \begin{center}
385: \large Figure Captions
386: \end{center}
387:
388: \figcaption{
389: $uvby$ metallicities for planet-bearing stars
390: obtained with our photometric calibration
391: in comparison with spectroscopically determined metallicities
392: from the CSR 2001 compilation; solid line is a 1:1 correlation
393: \label{fig1}}
394:
395: \figcaption{
396: Distribution of residuals between photometrically and spectroscopically
397: determined metallicities, with Gaussian fits to the distributions overplotted
398: in the same line style. Our new photometric calibration is shown as a
399: solid line; the SN calibration is shown as a dashed line.
400: \label{fig2}}
401:
402: \figcaption{
403: Distribution of residuals between photometrically and spectroscopically
404: determined metallicities, with Gaussian fits to the distributions overplotted.
405: Only measurements yielding a spectroscopic determination of
406: ${\rm [Fe/H]}>0.0$ are plotted.
407: Our new photometric calibration is shown as a
408: solid line; the SN calibration is shown as a dashed line.
409: \label{fig3}}
410:
411: \figcaption{
412: Distribution of calculated values of planeticity for the 73 nearby Lick Survey stars. Stars without planets are shown as a solid line; stars with confirmed planets are shown as a filled histogram.
413: \label{fig4}}
414:
415: \end{document}
416:
417:
418:
419: