1: \documentstyle[11pt,newpasp,twoside]{article}

2: \markboth{Seager}{Sodium Detection in HD~209458~b}

3: \pagestyle{myheadings} \nofiles

4: 

5: % Some definitions I use in these instructions.

6: 

7: \def\emphasize#1{{\sl#1\/}}

8: \def\arg#1{{\it#1\/}}

9: \let\prog=\arg

10: 

11: \def\edcomment#1{\iffalse\marginpar{\raggedright\sl#1\/}\else\relax\fi}

12: \marginparwidth 1.25in

13: \marginparsep .125in

14: \marginparpush .25in

15: \reversemarginpar

16: 

17: %my definitions

18: \def\spose#1{\hbox to 0pt{#1\hss}}

19: \def\rsun{R_{\odot}}

20: \def\msun{M_{\odot}}

21: \def\rhosun{\rho_{\odot}}

22: \def\rstar{R_{*}}

23: \def\mstar{M_{*}}

24: \def\rplanet{R_{p}}

25: \def\mplanet{M_{p}}

26: \def\mp{M_p}

27: \def\rp{R_p}

28: \def\sqdepth{\sqrt{\Delta F}}

29: \def\tf{t_F}

30: \def\tt{t_T}

31: \def\rhostar{\rho_*}

32: 

33: \begin{document}

34: \title{The Significance of the Sodium Detection in the Extrasolar 

35: Planet HD~209458~b Atmosphere}

36: 

37: \author{S.\ Seager}

38: \affil{Institute for Advanced Study, School of Natural Sciences, 1

39: Einstein Drive, Princeton, NJ 08540}

40: 

41: \affil{The Carnegie Institution of Washington, Dept. of

42: Terrestrial Magnetism, 5241 Broad Branch Rd. NW, Washington, DC

43: 20015}

44: 

45: \begin{abstract}

46: The Hubble Space Telescope (HST) detection of an extrasolar planet

47: atmosphere in 2001 was a landmark step forward for the

48: characterization of extrasolar planets. HST detected the trace

49: element sodium, via the neutral atomic resonance doublet at 593

50: nm, in the transiting extrasolar giant planet HD~209458~b. In this

51: paper I discuss the significance of this first ever extrasolar

52: planet atmosphere detection. I explain how the sodium measurement

53: can be used as a constraint on HD~209458~b atmosphere models and

54: review recent interpretations of the lower-than-expected sodium

55: line strength.

56: \end{abstract}

57: \vspace{-4mm}

58: \section{Introduction}

59: The first---and only to date---atmosphere detection of an extrasolar

60: planet was made by Charbonneau et al. (2002) with the Hubble Space

61: Telescope (HST). Charbonneau et al. (2002) observed the parent star

62: HD~209458~A, during and outside of the planet transit. During a transit,

63: some of the stellar intensity passes through the optically thin parts

64: of the planet atmosphere. Thus the star's spectrum taken during

65: transit is expected to contain some weak signature of the planet

66: atmosphere. The planet atmosphere signature measured in this way is

67: called the ``transmission spectrum''.

68: 

69: The strength of the planetary transmission spectrum can be

70: estimated as the ratio of the planet atmosphere annulus area

71: to the star area. We can very roughly

72: estimate the maximum annulus area by assuming the atmosphere is 10

73: ``pressure scale heights'' thick. The atmospheric pressure scale

74: height, $H$, is the characteristic vertical dimension in pressure:

75: the e-folding distance for pressure,

76: \begin{equation}

77: H = \frac{P}{g\rho} = \frac{kT}{mg},

78: \end{equation}

79: where $P$ is pressure, $g$ is surface gravity, $\rho$ is density, and

80: $m$ is atomic or molecular mass. With appropriate numbers for

81: HD~209458~b ($T=1000$~K, $g=8.6$~m s$^{-2}$, $m$ the mass of H$_2$),

82: $H\sim$~500~km\footnote{The scale height is constant only for an

83: isothermal atmosphere and is otherwise depth dependent; i.e. it

84: changes throughout the atmosphere. For our estimate it is sufficient

85: to use an average number.}. Ten times the pressure scale height is

86: 5000 km, which is 5\% of the planetary radius. Note that the large

87: scale height works to great advantage; the much cooler

88: planet Jupiter with a higher surface gravity has

89: a scale height of 24 km; ten times this is only 0.3\% percent of

90: Jupiter's radius.

91: 

92: The above estimate for atmosphere size can be used to estimate the

93: maximum strength of the planet atmosphere transmission spectrum in the

94: combined star + planet spectrum. This maximum estimate uses the

95: assumption that at some wavelengths the gas is strongly absorbing

96: (i.e., opaque) out to 10 pressure scale heights.  Comparing the area

97: of the atmosphere annulus to the stellar disk, ignoring stellar limb

98: darkening, we get $1.5\times10^{-3}$.

99: 

100: Because the planet transmission spectrum is so weak, the planet

101: transmission spectrum must be searched for in the residuals of the

102: in-transit minus out-of-transit stellar observations. Furthermore,

103: the weakness of the planet signature requires that observers are

104: unable to take a broad spectrum to search for atomic and molecular

105: absorption features, but instead must concentrate observations on

106: a pre-chosen narrow wavelength region in order to get a high

107: enough signal-to-noise. Charbonneau et al. (2002) describe a very

108: careful analysis resulting in a measurement of the sodium (Na)

109: line strength of $2.3 \pm 0.57 \times$ 10$^{-4}$. Even so, the Na

110: detection is a 4-$\sigma$ result and follow-up confirming

111: measurements of sodium other absorption features (such as the

112: resonance doublet of potassium (K) (767.0~nm), and near-IR H$_2$O,

113: CO, and possibly CH$_4$) would be reassuring. Na---expected to be

114: the strongest absorber---found to be less than the maximum

115: estimate implies that the other species may have much smaller

116: signatures than expected.

117: 

118: The planet atmospheric transmission spectrum is wavelength-dependent;

119: at wavelengths where no absorbers are present the stellar intensity

120: will pass through the atmosphere unimpeded, whereas at (possibly)

121: neighboring wavelengths a strong absorber will allow no stellar

122: intensity to be transmitted. This wavelength dependency can be thought

123: of a planet radius being different sizes at different

124: wavelengths. Comparing the continuum to an absorption line was used to

125: make the Na line measurement.

126: 

127: 

128: \section{The Significance of the Sodium Detection}

129: The first ever atmosphere detection is a landmark in extrasolar

130: resesarch because atmosphere studies open a whole new window to

131: extrasolar planet characterization. Three main reasons why the

132: sodium detection in the atmosphere of HD~209458~b is so important

133: are described in the following subsections.

134: 

135: \subsection{CEGPs Are What we Expect to First Order}

136: First and foremost the detection of neutral Na confirms the very

137: basic postulate that CEGPs have atmospheres expected for their

138: equilibrium effective temperatures ($T_{eq}$). This is because

139: HD~209458~b is a representative of the class of the close-in

140: extrasolar giant planets (CEGPs) with semi-major axes $<0.05$~AU.

141: Physically, $T_{eq}$ is is the effective temperature

142: attained by an isothermal planet after it has reached complete

143: equilibrium with its star.

144: \begin{equation}

145: T_{eq} = T_* \left( R_*/2a\right)^{1/2}[f(1-A_B)]^{1/4},

146: \end{equation}

147: where $T_*$ is the stellar temperature, $R_*$ is the stellar

148: radius, $a$ is the semi-major axis, $A_B$ is the Bond albedo, and

149: $f$ is a parameter to describe the heat redistribution from the

150: day side where $f=1$ if the heat is evenly distributed or $f=2$ if

151: only the day side reradiates the energy. With the unknown Bond

152: albedo the CEGPs are expected to range in temperature from

153: approximately 1000 to 1500~K; Na is expected to exist in

154: neutral gaseous form in atmospheres for any $T_{eq}$ in this

155: range. In other words, for HD~209458~b, for a given distance from

156: the parent star, star temperature, and solar composition the

157: HD~209458~b atmosphere is to first order what is expected.

158: 

159: Even though at 2 x 10$^{-6}$ solar abundance Na is essentially a trace

160: element, the neutral Na spectral signature is extremely strong at

161: optical wavelengths due to a very strong resonance line. In addition

162: there are expected to be no other strong optical absorption lines;

163: most atoms are locked into molecules which tend to have transitions at

164: either UV wavelengths (electronic transitions) or at IR wavelengths

165: (rotational-vibrational transitions). In fact since cool T dwarfs with

166: similar effective temperatures to HD~209458~b have extremely deep and

167: broad Na and K resonance lines it is not a surprise that neutral Na

168: was measurable in the HD~209458~b atmosphere. See Figure~2 in Liebert

169: et al. (2000) for the first observational confirmation of Na and K as

170: the major optical absorbers in T dwarfs, from a spectrum of the T

171: dwarf SDSS1624.

172: 

173: \subsection{The Sodium Line was Predicted in Advance}

174: The presence of neutral Na in the atmosphere of HD~209458~b was first

175: predicted for transmission spectra by Seager \& Sasselov

176: (2000). Before the atmosphere detection the Na line doublet was also

177: studied by Brown (2001) and Hubbard et al.  (2001). The HST atmosphere

178: detection of the transiting planet is reassuring because it involved a

179: specialized observation that required advance knowledge of what

180: atmospheric feature to look for, as described in Sec. 1. The success

181: of the Na detection shows that applications of atmospheric physics to

182: planets in new environments can be successful and that model results

183: as guidelines to experimental design are reliable.

184: 

185: \subsection{The Sodium Line Strength is Weaker than Expected} The

186: measured value of the Na absorption line feature was lower than

187: predicted. This is not surprising because the models did not include

188: accurate treatment of secondary effects and inputs, almost all of

189: which should reduce the strength of the Na line. To illuminate the

190: significance of the low value of detected Na I used simple inputs to

191: my atmosphere model and computed the change in transit depth in

192: adjacent bands as was done with the real data in Charbonneau et

193: al. (2002; see Figure~1 in this reference).  The simple inputs

194: include: solar composition, cloud-free, and that the heat from the

195: incident stellar intensity is instantaneously redistributed around the

196: planet. The ``secondary effects'' are all omitted: no photoionization,

197: no photochemistry, no atmospheric circulation.  For this reason I

198: emphasize that the model is just one out of a large range of parameter

199: space, but that it is sufficient for the illustrative purposes here.

200: 

201: The computed transit depth in the Na line from the simple model is

202: $\sim9 \times 10^{-4}$, 12 sigma away from the observational

203: measurement.  This simple model is therefore completely ruled out by

204: the data.  In this simple, homogeneous model the abundance of Na would

205: have to be approximately 1000 times less than solar to match the

206: observed line strength. In other words, the low measured value of the

207: Na line is significant.

208: 

209: \section{Using the Sodium Observation as a Constraint on Model

210: Atmospheres}

211: \subsection{Transmission Spectra Models}

212: The transmission spectrum from the transiting extrasolar planet

213: HD~209458~b is straightforward to compute. Stellar light rays travel

214: along the line of sight through the planet atmosphere towards the

215: observer. These rays from the star suffer exponential attenuation

216: due to wavelength-dependent absorption from the planetary

217: atmosphere\footnote{Refraction in the planet's atmosphere is

218: negligible (Hui \& Seager 2002).},

219: \begin{equation}

220: I_{\lambda} = I_{\lambda,0}\exp(-\tau_{\lambda}),

221: \end{equation}

222: where $I_{\lambda}$ is intensity and $I_{\lambda,0}$ is initial

223: intensity originating from the star. $\tau_{\lambda}$ is the

224: optical depth

225: \begin{equation}

226: \tau_{\lambda} = \int_0^L n(l) \sigma_{\lambda} dl,

227: \end{equation}

228: where $n(l)$ is the number density of a given species,

229: $\sigma_{\lambda}$ is the absorption cross section, and $L$ is the

230: path length through the planetary atmosphere. $\tau_{\lambda}$ can

231: include many different absorption species and $n(l)$ and

232: $\sigma_{\lambda}$ depend on $T$ and $P$.

233: For a resonance line transition where $n_1$ is

234: the number density of atoms with electrons in the ground state and

235: subscript 2 refers to the $n=2$ state,

236: \begin{equation}

237: n_1 \sigma_{\lambda} = n_1 \frac{h \nu_0}{4 \pi} B_{12} \phi(\nu),

238: \end{equation}

239: where $B_{12}$ is an Einstein probability and $\phi(\nu)$ is the

240: line profile.

241: 

242: While the transmission spectra, via exponential attenuation shown in

243: equation~(3), is straightforward to calculate, it relies on the

244: underlying radial atmosphere structure via $T$, and $n_1$ (which is

245: related to $P$ and to the radiation field $J$); see Figure~1. This

246: underlying radial atmosphere structure is difficult to calculate. The

247: radiative transfer methods used to self consistently compute the

248: atmospheric structure in proximity to the central star and results can

249: be found in a variety of sources (e.g., Seager 1999; Barman et

250: al. 2002; Burrows \& Sudarsky 2003; Sudarsky et al. 2003a, 2003b).

251: The atmospheric temperature-pressure structure is required to

252: determine: the species abundance (from chemical equilibrium) that

253: factors into the number density for the opacity in equation (4); and

254: the temperature required for absorption coefficients, e.g.  equation

255: (5). There are many uncertainties in these atmosphere models. Most

256: notably, none of them consider atmospheric circulation which is

257: crucial for the redistribution of stellar irradiation (Showman \&

258: Guillot 2002; Showman \& Guillot 2003; Cho et al. 2003) that

259: determines the atmospheric temperature-pressure structure. In addition

260: photochemistry is omitted as is non-equilibrium chemistry. Steps are

261: being made to advance the models and this is an ongoing process.

262: 

263: Different parts of a transmission spectrum absorption line are formed

264: at different parts in the atmosphere. The transmission spectral line

265: is created by removing photons from the stellar beam. This absorption

266: process can be thought of as additive over many layers of the

267: atmosphere (see Figure~1). Deep in the atmosphere, the number density

268: of absorbing atoms is high and at the line center all of the stellar

269: photons will be absorbed, i.e., the part of the line formed deep in

270: the atmosphere is saturated.  In addition, deep in the atmosphere

271: pressure broadening will be strong, causing absorption away from the

272: line center---the line wings. High in the atmosphere there is little

273: pressure broadening and hence little contribution to the line

274: wings. Because the number density of absorbers is much lower compared

275: to the deep atmosphere, the line core is not saturated and actually is

276: formed in the upper atmosphere. Although the Charbonneau et al. (2002)

277: measurement does not resolve the spectral line, some information can

278: be garnered about the gross line shape because the measurement is deeper

279: in the narrow bands than the medium bands, and is absent from the wide

280: bands,

281: 

282: 

283: \begin{figure}[h]

284: \plotfiddle{transit2.eps}{2.5in}{0}{40}{40}{-130}{0} \caption{A

285: schematic cross section of HD~209458~b to illustrate transmission

286: spectrum modeling. Light rays (indicated by the lines with arrows)

287: travel from the star through the planetary atmosphere. The total

288: transmission spectrum is additive over the different rays. Because

289: the rays sample different radial atmosphere depths, the underlying

290: temperature-pressure structure is crucial to the transmission

291: spectrum calculation. This simple day/night picture for HD~209458~b,

292: however, is in error (Showman \& Guillot 2002, 2003; Cho et al.

293: 2003). Note that for the transmission spectrum calculation

294: $R_p$---the measured planetary radius at optically thick

295: wavelengths---must be chosen.} \label{fig:2D}

296: \end{figure}

297: 

298: \subsection{Possible Interpretations of the Weak Sodium Line}

299: 

300: The Na line strength measured by Charbonneau et al. (2002)

301: provides the first constraint on extrasolar giant planet

302: atmosphere models. Notably the measured Na abundance is much lower

303: than predicted. This weaker than expected Na line strength can be

304: matched by only a few extreme cases in the parameter space modeled

305: by Charbonneau et al. (2002). In their discovery paper,

306: Charbonneau et al. (2002) outline four possible explanations for

307: the weaker than expected Na line: a high cloud; photoionization;

308: low Na abundance; or chemical equilibrum depletion of atomic Na

309: into molecules and solids.

310: 

311: Many causes have been proposed since the discovery paper, with

312: calculations to back them up, to explain the lower than expected

313: strength of the Na line.  The author finds it interesting that each

314: group finds the reason to be what their code can calculate. The

315: proposed explanations are described in the rest of this subsection.

316: 

317: {\bf Non-LTE effects} could make the Na line weaker and a different

318: shape compared to the simple model (Sec. 2.3).  LTE, an abbreviation

319: for ``local thermodynamic equilibrium'', is a simplification used in

320: most extrasolar planet atmosphere models (and many stellar atmosphere

321: models). The LTE assumption is that the state of the gas can be

322: described by only two variables, temperature and pressure. State of

323: the gas refers to the chemical equilibrium partitioning of molecules,

324: the ionization states, and, most relevant for this case, the atomic

325: level populations (used in equation (5)). In order to compute the

326: transmission spectrum of the Na line, the ground state level

327: population needs to be known (see equation~(4)). LTE is valid at high

328: densities and where the radiation field is a black body. In the upper

329: layers of HD~209458~b, where part of the transmission spectrum forms,

330: neither of these is valid and non-LTE is expected to prevail. The

331: atomic levels must be computed under non-LTE in order to compute the

332: line strength.  Because the upper atmosphere is most senstive to

333: non-LTE conditions, the non-LTE calculation mostly affects the line

334: core, even producing emission in the line core with absorption in line

335: wings. See Barman et al. (2002, 2003) for a more detailed description

336: of the non-LTE effect.

337: 

338: {\bf A high cloud} was first suggested in Seager \& Sasselov

339: (2000). We assumed that the cloud would be optically thick, and the

340: cloud top would correspond to $R_p$ in Figure~1. In this case the only

341: part of the planet atmosphere that is probed by the transmitting

342: stellar rays is the part of the atmosphere above the cloud tops. If

343: the cloud is high in the atmosphere, then the stellar rays probe a

344: small part of the atmosphere that is not very dense, with low pressure

345: broadening. The resulting Na transmission spectrum line is expected to

346: be fairly weak with no broad line wings. This is in contrast to the

347: low or no cloud case where $R_p$ is low in the atmosphere so that the

348: stellar rays probe a large fraction of the atmosphere where the high

349: number densities cause high pressure broadening. The resulting Na line

350: is expected to be strong and wide from pressure broadening.

351: 

352: {\bf Photoionization} is an obvious suggestion since due to the

353: proximity of the planet to the parent star the planet is receiving

354: huge amounts of stellar UV photons compared to solar system planets

355: where photoionization is known to play a role. Detailed

356: photoionization models have yet to be produced; in addition to Na they

357: would involve many atoms and molecules which could serve as sinks for

358: the UV radiation and sources of free electrons for Na$^+$ to

359: recombine. Using a simple photoionization model Fortney et al. (2003)

360: show that the weak Na line can be reproduced with photoionization

361: together with a high cloud.

362: 

363: {\bf Atmospheric circulation} models (Guillot \& Showman 2002; Cho et

364: al. 2003) are needed to describe the transfer of stellar radiation

365: throughout the planet atmosphere. HD~209458~b is expected to be

366: tidally locked (Guillot et al. 1996) to its parent star, thus one side

367: is permanently heated while the other side is in permanent

368: darkness. Strong winds are expected to redistribute the stellar

369: radiation, based on a comparison of the advective and radiative time

370: scales. Thus atmospheric circulation will determine the atmospheric

371: structure (see Figure~1) and hence the transmission spectrum. Guillot

372: \& Showman (2002, 2003) suggest that winds could transport atomic Na

373: from the planetary day side to the night side. They propose that

374: temperatures on the night side could be colder, allowing Na to

375: condense into solid NaCl where it could sink out of the atmosphere and

376: not be available as Na for transmission spectra. In this scenario, Na

377: is depeleted and a weak line results.

378: 

379: {\bf Post-accretionary extraplanetary origin of sodium} has been

380: suggested by Atreya et al. (2003).  Atreya et al. (2003) propose that

381: primordial Na is not present in HD~209458~b's atmosphere due to

382: differentiation to the planetary interior and subsequent depletion of

383: atomic Na into molecular species. They propose that the small amount

384: of Na present in HD~209458~b's atmosphere comes from later influx of

385: material from meteorite or comet impacts, planetary rings, or a

386: volcanically active satellite.

387: 

388: {\bf Low sodium abundance.} It could be difficult to attribute the

389: weakness of the Na line to a low Na abundance until all of the

390: above effects have been ruled out. As suggested in Charbonneau et

391: al. (2002), however, if observations of H$_2$O and CO show

392: ``normal'' abundances compared to Na, then we may be able to

393: attribute the weak Na line to low primordial abundance.

394: 

395: {\bf Can anything be ruled out} from the Na measurement? There is no

396: detailed information about line shape from the Na measurement.

397: However, based on the fact that the transit in the Na line was weaker

398: in the medium band than the narrow band, and not detected in the wide

399: band, we can say that a very broad absorption line is not present.

400: Therefore we have an indication that either the deep atmosphere is not

401: being probed or the Na atomic number density is very low. Furthermore,

402: we know that photoionization alone is unlikely to be the only

403: explanation of the weak Na signal. Photoionization affects only the

404: upper atmosphere, and hence mostly the line core. In this case the

405: line should still be broad.  Emission line models (see Barman et

406: al. (2002)) under extreme assumptions in the non-LTE scenario are also

407: ruled out because a net increase---as opposed to a net decrease---in

408: Na line flux during transit would have been detected.

409: \vspace{-4mm}

410: \section{Other Observational Diagnostics}

411: The Na measurement is a very useful constraint on HD~209458~b's

412: atmosphere. Yet it is not enough to get anywhere near a clear picture

413: of the atmosphere. Many observations are ongoing to try to detect

414: other atmospheric constituents. We expect the H$_2$O near-IR

415: absorption bands to be prominent spectral features. CO or CH$_4$ or

416: both should also be present, the relative abundance is a good

417: temperature diagnostic (Seager, Whitney, and Sasselov 2000).

418: HD~209458~b's albedo and orbital light curve will be measured with the

419: upcoming space satellite MOST (Micro Oscillations of

420: STars)\footnote{http://www.astro.ubc.ca/MOST/2002/index.html}.

421: HD~209458~b's day and night side temperatures will be measured with

422: SIRTF. Following is a table of key features that are obtainable within

423: the next two years. The combination of all of these observational

424: diagnostics will provide powerful constraints on theoretical

425: atmosphere models.

426: 

427: \acknowledgements{This work was supported by the W. M. Keck

428: Foundation and by the Carnegie Institution of Washington.}

429: 

430: %\vspace{5mm}

431: \begin{tabular}{l l l}

432: %\renewcommand{\arraystretch}{0.8}

433: \hline \\

434: Bulk Property & Observational Diagnostic & Method \\

435: \hline

436: \hline

437: Temperature (T) &  CH$_4$ vs. CO  &   ground\\

438: \\

439: T of day & IR photometry & ground or SIRTF \\

440: and night side & & \\

441: \\

442: Albedo & Secondary eclipse photometry &  MOST \\

443: \\

444: General properties & Photometry &  MOST \\

445: of atmospheric scattering &during the planet's orbit \\

446: particles & & \\

447: \\

448:   Other spectral lines & Transit transmission spectra &

449: ground, HST \\

450: \\

451: \hline

452: %Expected observations of HD~209458~b within the next two years.

453: \end{tabular}

454: 

455: \vspace{-2mm}

456: \begin{references}

457: \vspace{-1mm}

458: \reference {} Atreya, S. K., Mahaffy, P. R., Niemann, H. B., Wong, M. H., \& Woen, T. C. 2003, Planetary and Space Science, 51, 105

459: 

460: \reference {} Barman et al. 2003, these proceedings

461: 

462: \reference {} Barman, T. S., Hauschildt, P. H., Schweitzer, A., 

463: Stancil, P. C., Baron, E., \& Allard, F. 2002, ApJ, 569, L51

464: 

465: \reference {} Brown, T. M. 2001, ApJ, 553, 1006

466: 

467: \reference {} Burrows, A. \& Sudarsky, S. 2003, these proceedings

468: 

469: \reference {} Charbonneau, D., Brown, T., Noyes, R. W., \& Gilliland

470: R. 2002, ApJ, 568, 377

471: 

472: \reference {} Cho, J. Y-K., Menou, K., Hansen, B., \& Seager, S. 

473: 2003, ApJ, 587, L117

474: 

475: \reference {} Fortney, J. J., Sudarsky, D., Hubbard W. B., Burrows, 

476: A., Lunine, J. I., \& Cooper, C. S. 2003, submitted to ApJ, 

477: astro-ph/0208263

478: 

479: \reference {} Guillot, T. \& Showman, A. P. 2003, these

480: proceedings

481: 

482: \reference {} Guillot, T. \& Showman, A. P. 2002 A\&A, 385, 156

483: 

484: \reference {} Guillot, T., Burrows, A., Hubbard, W. B., Lunine, J. 

485: I., \& Saumon, D. 1996, \apj, 459, L35

486: 

487: \reference Guillot, T. et al. 1996, \apj, 459, L35

488: 

489: \reference {} Hubbard, W. B., Fortney, J. J., Lunine, J. I., Burrows, 

490: A., Sudarsky, D., \&

491: Pinto, P. 2001, ApJ, 560, 413

492: 

493: \reference {} Hui, L. \& Seager, S. 2002, ApJ, 574, 1004

494: 

495: \reference{} Liebert, J., Reid, I. N., Burrows, A., Burgasser, A.

496: J., Kirkpatrick, J. D., \& Gizis, J. E. 2000, ApJ, 533, L155

497: 

498: \reference{} Seager, S. 1999, PhD Thesis, Harvard University

499: 

500: \reference {} Seager, S. \& Sasselov, D. D. 2000, ApJ, 537, 916

501: 

502: \reference {} Seager, S., Whitney, B. A., \& Sasselov, D. D. 2000,

503: ApJ, 540, 504

504: 

505: \reference {} Showman, A. P. \& Guillot T. 2003, these proceedings

506: 

507: \reference {} Showman, A. P. \& Guillot, T. 2002 A\&A, 385, 166

508: 

509: \reference {} Sudarsky et al. 2003a, these proceedings

510: 

511: \reference {} Sudarsky, D., Burrows, A., \& Hubeny, I. 2003b, 

512: ApJ in press, astro-ph/0210216

513: 

514: \end{references}

515: 

516: \end{document}

517: 

518: