0712.0761/ms.tex
1: \documentclass[12pt,preprint]{aastex}
2:  
3: \shorttitle{Detection of the HD189733b Exoplanet Atmosphere}
4: \shortauthors{Redfield et al.}
5:  
6:  
7: \begin{document}
8:  
9: \newcommand{\php}[0]{\phantom{--}}
10: \newcommand{\kms}[0]{km~s$^{-1}$}
11:  
12: \title{Sodium Absorption From the Exoplanetary Atmosphere of HD189733b
13: Detected in the Optical Transmission Spectrum\footnote{Based on
14: observations obtained with the Hobby-Eberly Telescope, which is a
15: joint project of the University of Texas at Austin, the Pennsylvania
16: State University, Stanford University,
17: Ludwig-Maximilians-Universit\"{a}t M\"{u}nchen, and
18: Georg-August-Universit\"{a}t G\"{o}ttingen.}}
19:  
20: 
21: \author{Seth Redfield\altaffilmark{2,3}, Michael Endl\altaffilmark{2}, William
22: D. Cochran\altaffilmark{2}, and Lars Koesterke\altaffilmark{2,4}}
23: \altaffiltext{2}{Department of Astronomy and McDonald Observatory,
24: University of Texas, Austin, TX, 78712; {\tt
25: sredfield@astro.as.utexas.edu}}
26: \altaffiltext{3}{Hubble Fellow}
27: \altaffiltext{4}{Texas Advanced Computing Center, University of Texas,
28: Austin, TX, 78758}
29: 
30: 
31: \begin{abstract}
32: 
33: We present the first ground-based detection of sodium absorption in
34: the transmission spectrum of an extrasolar planet.  Absorption due to
35: the atmosphere of the extrasolar planet HD189733b is detected in both
36: lines of the \ion{Na}{1} doublet.  High spectral resolution
37: observations were taken of eleven transits with the High
38: Resolution Spectrograph (HRS) on the 9.2 meter Hobby-Eberly Telescope
39: (HET).  The \ion{Na}{1} absorption in the transmission spectrum due to
40: HD189733b is $\left( -67.2 \pm 20.7 \right) \times 10^{-5}$ deeper in
41: the ``narrow'' spectral band that encompasses both lines relative to
42: adjacent bands.  The 1$\sigma$ error includes both random and
43: systematic errors, and the detection is $>$3$\sigma$.  This amount of
44: relative absorption in \ion{Na}{1} for HD189733b is $\sim$3$\times$ larger
45: than detected for HD209458b by \citet{charbonneau02}, and indicates
46: these two hot-Jupiters may have significantly different atmospheric
47: properties.
48: 
49: 
50: \end{abstract}
51:  
52: \keywords{atmospheric effects --- line: profiles --- planetary systems
53: --- planets and satellites: individual (HD189733b) --- stars:
54: individual (HD189733) --- techniques: spectroscopic}
55: 
56: \section{Introduction}
57: 
58: 
59: High signal-to-noise ($S/N$) and high spectral resolution observations
60: of transiting exoplanets provide an opportunity to measure the
61: properties of exoplanet atmospheres and exospheres.  Early
62: theoretical models predicted that absorption, due to exoplanetary
63: atmospheres, should be detectable in the strong resonance lines of
64: \ion{Na}{1} and other alkali metals in optical transmission spectra
65: \citep{seager00,brown01}.  \citet{charbonneau02} used medium
66: resolution spectra ($R \equiv \lambda/\Delta\lambda \sim$ 5,540) of
67: \ion{Na}{1} from the Space Telescope Imaging Spectrograph (STIS)
68: onboard the {\it Hubble Space Telescope} ({\it HST}) to make the first
69: detection of absorption in the transmission spectrum caused by the
70: exoplanetary atmosphere of HD209458b during transit.  The measured
71: \ion{Na}{1} absorption, made in the ``narrow'' spectral window
72: (5887--5899\,\AA) since the doublet is not fully resolved, was
73: $\left(-23.2 \pm 5.7 \right) \times 10^{-5}$ deeper than in adjacent
74: bands during transit, and weaker than model predictions by a factor of
75: $\sim$3.  Adjustments to cloud levels, metallicity, rainout of
76: condensates, distribution of stellar flux, and photoionization of
77: sodium could account for the discrepancy between the models and
78: observations \citep{barman07,fortney03}.  Exospheric absorption was
79: also detected for HD209458b in several UV lines by
80: \citet{vidalmadjar03,vidalmadjar04}.
81: 
82: Previous ground-based attempts to detect absorption in the optical
83: transmission spectra of HD209458 and other bright transiting systems
84: have measured only upper limits, likely due to limited observations of
85: only a single effective transit
86: \citep[e.g.,][]{bundy00,moutou01,winn04,narita05,bozorgnia06}.  In
87: addition, contaminants, such as telluric lines and, in some cases, the
88: use of an iodine cell, hindered or prevented measurements of several
89: spectral features, most notably the \ion{Na}{1} lines.
90: 
91: %Radial velocity search spectra of
92: %HD209458, obtained by the High Resolution Echelle Spectrometer (HIRES)
93: %on Keck with an iodine cell, were used by \citet{bundy00} to place
94: %upper limits on calcium, hydrogen, and iron lines.  A partial transit
95: %of HD209458 observed with the Very Large Telescope (VLT) UV-Visual
96: %Echelle Spectrograph (UVES; $R\,\sim\,$58,000) was used by
97: %\citet{moutou01} to place limits on atmospheric and exospheric
98: %absorption.  HD209458 was observed with the High Dispersion
99: %Spectrograph (HDS) on Subaru ($R\,\sim\,$90,000) over a single transit
100: %and limits were placed on H$\alpha$ exospheric absorption
101: %\citep{winn04}, and absorption of \ion{Li}{1}, \ion{Fe}{1},
102: %\ion{Na}{1}, H$\beta$, H$\gamma$, \ion{Ca}{1} \citep{narita05}.
103: %Another bright transiting system, HD149026, was observed by
104: %\citet{bozorgnia06} with HIRES on Keck ($R\,\sim\,$55,000) over a
105: %single transit with an iodine cell, and limits were set on \ion{K}{1}
106: %and \ion{Li}{1} absorption.
107: 
108: \citet{bouchy05} announced the identification of the short-period
109: ($P=2.2$ days) transiting exoplanet around the bright ($V = 7.7$) K0V
110: star, HD189733.  Subsequent work on this system tightly constrained
111: the orbital parameters \citep{hebrard06,bakos06}, determined that the
112: axis of stellar rotation is aligned with the planetary orbital axis
113: using the Rossiter-McLaughlin effect \citep{winn06}, found evidence
114: for photometric stellar variability and a stellar rotational period of
115: $\sim$12 days \citep{henry07,winn07}, and identified transit
116: signatures of starspots \citep{pont07}.  Infrared studies of this
117: system have been successful in characterizing the atmospheric
118: temperature distribution \citep{deming06,knutson07}.
119: 
120: \section{Observations and Data Reduction}
121: 
122: Based on the previous ground-based attempts to detect exoplanetary
123: absorption in transmission spectra, our program strategy was designed
124: to observe multiple transits and to specifically target
125: the strong \ion{Na}{1} lines.  Observations of HD189733 were taken
126: over the course of a year (11 Aug 2006 to 11 Aug 2007) using the
127: 9.2~meter Hobby-Eberly Telescope (HET; \citealt{ramsey98}) High
128: Resolution Spectrograph (HRS; \citealt{tull98}).  These observations
129: cover the spectral range from 5000--9000~\AA, with a resolving power
130: of $R \sim$ 60,000.  No iodine cell was used.  A bright, rapidly
131: rotating, B star ($\alpha$ Del) was observed immediately after
132: HD189733 in order to characterize the telluric absorption.
133: We obtained spectra during 11 in-transit visits, and 25 out-of-transit
134: visits.  Each visit is typically comprised of several ($\sim$3.2
135: during transits and $\sim$5.1 outside of transits) 600 second
136: exposures, and the $S/N$ at \ion{Na}{1} of each exposure is
137: $\sim$320 per resolution element.  Observations are considered
138: in-transit, if the entire exposure is obtained within first and fourth
139: contact.  
140: 
141: %The fixed elevation-axis design of the HET necessitates
142: %queue-scheduling, which is well-suited to obtaining sporadic
143: %in-transit observations, together with the out-of-transit observations
144: %which can be acquired at the convenience of the telescope schedule.
145: 
146: The data were reduced using Image Reduction and Analysis Facility
147: \citep[IRAF;][]{tody93} and Interactive Data Language (IDL) routines
148: to subtract the bias, flat field the images, remove scattered light
149: and cosmic ray contamination, extract the echelle orders, calibrate
150: the wavelength solution, and convert to heliocentric velocities.
151: Local interstellar medium (LISM) and stellar lines were removed from
152: the telluric standard before being applied to the target spectra.
153: LISM \ion{Na}{1} absorption is detected in the spectra of $\alpha$ Del
154: ($\sim$74 pc), our telluric standard, consistent with LISM detections
155: toward other nearby stars \citep{redfield02} and previous \ion{Ca}{2}
156: and \ion{Na}{1} observations of $\alpha$ Del by \citet{vallerga93} and
157: \citet{welsh91}.  Stellar lines in our telluric standard, a rapidly
158: rotating hot B-star, are very weak, and were removed using a stellar
159: model \citep{koesterke07,fitzpatrick05,castelli06}.  Observations of
160: HD189733 were divided by the cleaned telluric standard spectra to
161: remove the contaminating telluric features.
162: 
163: To create the in-transit and out-of-transit templates, additional
164: processing was required to match the blaze response and wavelength
165: calibration of all observations.  Over the short spectral windows
166: ($\sim$30 \AA) of interest, fluctuations in the blaze response are
167: smooth and slowly varying, and were easily characterized by a low
168: order polynomial.  Likewise, precise spectral alignment required small
169: adjustments ($\sim$0.1--0.2 pixels) to the wavelength solution based
170: on the cross correlation of nearby stellar lines.
171: 
172: 
173: \section{\ion{Na}{1} Transmission Absorption}
174: 
175: Figure~\ref{fig:na} shows the observed \ion{Na}{1} region of the
176: stellar spectrum.  The doublet is fully resolved and accompanied by
177: many weak stellar lines.  The bottom plot shows the transmission
178: spectrum, or the difference of the relative fluxes, $\left({\cal
179: F}_{\rm in} - {\cal F}_{\rm out}\right)/{\cal F}_{\rm out}$, of the
180: ultra high $S/N$ in-transit template (${\cal F}_{\rm in}$; $S/N \sim
181: 1600$) and the even higher $S/N$ out-of-transit template (${\cal
182: F}_{\rm out}$; $S/N \sim 3400$).  The signal predicted due to
183: differential stellar limb-darkening is shown, and has been removed
184: from the observations, although the impact is minimal.
185: 
186: Excess absorption is clearly detected in the in-transit spectrum
187: (${\cal F}_{\rm in}$) compared to the out-of-transit spectrum (${\cal
188: F}_{\rm out}$) for both \ion{Na}{1} lines.  Using the same ``narrow''
189: spectral region defined by \citet{charbonneau02}, 5887--5899~\AA, we
190: measure an excess of absorption $\left( -67.2 \pm 7.2 \right) \times
191: 10^{-5}$, when compared to the adjacent spectrum.  This is $\sim$3$\times$
192: larger than measured for HD209458b \citep[{[$-23.2 \pm 5.7] \times
193: 10^{-5}$;}][]{charbonneau02}.  We measure a random error of $7.2 \times
194: 10^{-5}$, comparable to that achieved by the STIS/{\it HST}
195: observations.
196: 
197: We also analyze a neighboring strong stellar line that is predicted to
198: show no significant exoplanetary atmospheric absorption.  In
199: Figure~\ref{fig:ca}, we present the same analysis near the \ion{Ca}{1}
200: line at 6122\,\AA.  No absorption is detected in this
201: region, and over a narrow spectral range from 6119--6125\,\AA, we
202: measure an insignificant amount of emission $\left( +15.4 \pm 6.8
203: \right) \times 10^{-5}$, consistent with there being no difference
204: between the in-transit and out-of-transit templates.
205: 
206: \subsection{Stellar Limb Darkening}
207: 
208: Differential limb-darkening, the relative difference in the
209: limb-darkening response of the cores of spectral lines and the
210: continuum, contributes to the transmission spectrum.  Since we are
211: dealing with a narrow spectral window, and the spectra are normalized
212: by the adjacent continuum, these observations are not sensitive to the
213: broadband color dependency expected due to limb-darkening
214: \citep{brown01etal,ballester07}.
215: 
216: Essentially, the in-transit template is a collection of spectra where
217: a small fraction, ($\sim$2.5\%; \citealt{pont07}), of the stellar disk
218: is blocked.  This blocked component causes a loss of flux at a
219: particular radial velocity, and leads to the Rossiter-McLaughlin
220: effect \citep{winn06}.  In addition, the blocked component is
221: associated with a specific $\mu$-angle, and therefore causes the loss
222: of a specifically limb-darkened fraction of the stellar disk.  The
223: observational distribution of $\mu$ angles, all necessarily $<$0.73
224: since the HD189733b transits at an angle slightly inclined from
225: precisely edge-on \citep[$i=85.7^{\circ}$;][]{pont07}, will produce a
226: unique differential limb-darkening signature.  Indeed, with sufficient
227: $S/N$, a transiting exoplanet could be used as a blocking obstruction,
228: where the resulting differential limb-darkening spectrum would
229: constrain, among other things, the temperature profile of the stellar
230: atmosphere.  In our case, we are concerned about the contribution this
231: signal may have on the absorption feature in the transmission
232: spectrum.
233: 
234: We use a stellar model of HD189733, based on literature values,
235: $T_{\rm eff} = 5050$ K, $\log g = 4.5$, and Fe/H$ = -0.03$
236: \citep{bouchy05}, to calculate the stellar flux blocked by the
237: exoplanet for each observation \citep{koesterke07}.  We assume a 0.755
238: $R_{\odot}$ star and a 1.154 $R_{\rm Jup}$ planet \citep{pont07}.  The
239: blocked flux is subtracted from the integrated disk flux and the
240: continuum is normalized.  The colored spectrum in the bottom panel of
241: the left plot in Figures~\ref{fig:na} and \ref{fig:ca}, shows the
242: relative difference spectrum due to differential limb-darkening for
243: our distribution of in-transit observations.  The contribution of
244: differential stellar limb-darkening is much smaller than the observed
245: absorption, and as for HD209458 \citep{charbonneau02}, at \ion{Na}{1},
246: it is actually a net emission, where it contributes $+1.46 \times
247: 10^{-5}$ in the ``narrow'' spectral window.  At \ion{Ca}{1} the
248: contribution is $-0.10 \times 10^{-5}$.  In both cases, the
249: differential limb-darkening contribution is well below the 1$\sigma$
250: random error.
251: 
252: \subsection{Empirical Monte Carlo \label{sec:emc}}
253: 
254: Given the weakness of any atmospheric absorption in a transmission
255: spectrum, it is critical to assess the contribution of systematic
256: errors.  Systematic errors associated with data reduction may arise
257: from the continuum normalization, wavelength calibration, 
258: removal of the stellar and interstellar lines from the telluric
259: standard, and the differential limb-darkening calculation.  In
260: addition, there are systematic contributions that are unrelated to
261: data reduction procedures, including stellar variability, the changing
262: distribution of starspots, and transits over active regions.  These
263: issues are particularly important since HD189733 is a moderately
264: active star that is known to exhibit photometric variability due to
265: star spots \citep{henry07,pont07}.  Furthermore, because our dataset
266: is comprised of many observations, spread over a year, it is important
267: to quantify the stability of the individual observations that
268: contribute to our in-transit and out-of-transit templates.
269: 
270: We have constructed a diagnostic that we call an empirical Monte
271: Carlo, or bootstrapping, analysis.  It involves randomly selecting
272: from the $\sim$200 individual exposures a test sample of
273: ``in-transit'' exposures, running through our entire data analysis
274: algorithm, and measuring any signal in the transmission spectrum.  We
275: repeat this exercise thousands of times in order to establish
276: statistical significance.  Three different scenarios are explored: (1)
277: an ``out-out'' comparison, where a subset of out-of-transit exposures,
278: equal to the number of our in-transit exposures ($\sim$36), is
279: selected and compared to the rest of the out-of-transit exposures; (2)
280: an ``in-in'' comparison, where a subset of in-transit exposures
281: ($\sim$8), set at the same ratio as the number of in-transit exposures
282: to the number of out-of-transit exposures (36/165), is compared to the
283: rest of the in-transit exposures; and (3) an ``in-out'' comparison,
284: where an increasing number of in-transit exposures (up to $\sim$half
285: of the sample) is removed from the in-transit template and compared to
286: the out-of-transit template.  The distribution of the ``narrow'' band
287: difference of the relative flux measurements for all realizations of
288: the three scenarios are shown for \ion{Na}{1} in the right panel of
289: Figure~\ref{fig:na} and for \ion{Ca}{1} in Figure~\ref{fig:ca}.
290: 
291: Both the ``in-in'' and ``out-out'' distributions are centered on zero
292: net signal, indicating that there is no significant difference between
293: the different out-of-transit exposures (``out-out'' scenario), and no
294: significant difference between the various in-transit exposures
295: (``in-in'' scenario).  Since we have fewer in-transit exposures in
296: order to do an ``in-in'' comparison, the $S/N$ is greatly reduced, and
297: the distribution is much wider than the ``out-out'' distribution.  For
298: \ion{Na}{1}, the mean value ($\mu$) of the ``out-out'' distribution is
299: $-3.5 \times 10^{-5}$ with a standard deviation ($\sigma$) of $20.7
300: \times 10^{-5}$, whereas the ``in-in'' distribution $\mu = 3.9 \times
301: 10^{-5}$ and $\sigma = 35.5 \times 10^{-5}$.  The results are similar
302: for the \ion{Ca}{1} line where for the ``out-out'' distribution $\mu =
303: -0.5 \times 10^{-5}$ and $\sigma = 13.7 \times 10^{-5}$, and the
304: ``in-in'' distribution $\mu = -4.0 \times 10^{-5}$ and $\sigma = 32.8
305: \times 10^{-5}$.  The \ion{Ca}{1} line is located more centrally on
306: the blaze and therefore is acquired with slightly higher $S/N$, which
307: translates into a narrower distribution of measurements.
308: 
309: As expected, the ``in-out'' distributions are centered on the value we
310: obtained from the full dataset.  The width of this distribution
311: provides a consistency check that a handful of in-transit exposures
312: are not dominating the measurement.  This demonstrates, along with the
313: ``in-in'' distribution being centered at zero, that all the in-transit
314: exposures, over 11 transits, consistently show excess absorption from
315: the transiting exoplanetary atmosphere.
316: 
317: The width of the ``out-out'' distribution measures directly the
318: propagation of systematic errors through our analysis.  These
319: realizations include the contribution of both reduction systematics,
320: as well as any astrophysical systematics which are not correlated with
321: the exoplanet orbital period.  For example, if changing distributions
322: of starspots (i.e., different phases of the stellar variability cycle)
323: could mimic an absorption feature in the transmission spectrum, we
324: would expect to see a significant fraction of ``out-out'' realizations
325: result in such measurements.  Because no such realizations are able to
326: replicate the level of the detection, and because the orbital period
327: is completely decoupled from the stellar variability period
328: \citep{henry07} such that our 11 in-transit visits are randomly
329: distributed in stellar rotational phase, stellar variability cannot be
330: responsible for the observed absorption in the transmission spectrum.
331: The standard deviation of our ``out-out'' distribution is adopted as
332: the appropriate 1$\sigma$ error for our transmission spectrum
333: measurements since it encompasses both random and systematic errors.
334: Our \ion{Na}{1} absorption detection is $\left( -67.2 \pm 20.7 \right)
335: \times 10^{-5}$, while the signal at \ion{Ca}{1} is $\left( +15.4 \pm
336: 13.7 \right) \times 10^{-5}$.
337: 
338: \section{Properties of the HD189733b Atmosphere}
339: 
340: The absorption feature in the transmission spectrum in
341: Figure~\ref{fig:na} is a direct measurement of the effective radius of
342: the exoplanetary atmosphere due to the high opacity at the center of
343: the strong \ion{Na}{1} lines.  The right axis of Figure~\ref{fig:na}
344: shows the magnitude of the absorption in units of exoplanetary radius.
345: The cores of the \ion{Na}{1} lines extend to effective radii of
346: $\sim$1.06~$R_{\rm pl}$.
347: 
348: The spectra have not been corrected for the Doppler shift of the
349: exoplanetary absorption.  During conjunction, (e.g., mid-transit), the
350: radial velocity of the planet is 0 km\,s$^{-1}$, although the rate of
351: change in radial velocity is at its maximum.  Over the duration of the
352: transit, the radial velocity of HD189733b will change by $\pm$19.2
353: km\,s$^{-1}$.  This will cause a smearing of the absorption profile.
354: The observed feature appears to be significantly blueshifted from the
355: stellar line center, by $\sim$38 km~s$^{-1}$, and may be a combination of the planetary orbital motion and high speed winds flowing from the hot dayside, predicted by several dynamical atmospheric models \citep[e.g.,][]{cho03,cooper05}.
356: 
357: %The blueshift is
358: %apparent in all transmission spectra created from subsets of the full
359: %in-transit dataset (all realizations that make up the ``in-out''
360: %distribution in Figure~\ref{fig:na}), and even subsets binned as a
361: %function of transit phase.
362: 
363: Recent infrared observations of HD209458b by \citet{knutson07inver}
364: show evidence for a temperature inversion in the upper atmosphere of
365: that exoplanet.  \citet{burrows07} presented accompanying models which
366: require an additional absorber in the upper atmosphere of HD209458b,
367: which could be high-altitude clouds and explain the weaker than
368: expected \ion{Na}{1} absorption signal in the transmission spectrum
369: \citep{charbonneau02,fortney03}.  Our measurement of strong
370: \ion{Na}{1} absorption argues against the presence of such clouds high
371: in the atmosphere of HD189733b, consistent with the isothermal
372: temperature profile presented for HD189733b \citep{tinetti07}.
373: 
374: \section{Conclusions}
375: 
376: We present the first ground-based detection of absorption in the
377: transmission spectrum of a transiting extrasolar planet.  The
378: detection is based on the first high resolution ($R \sim 60$,000)
379: transmission spectrum, with the \ion{Na}{1} doublet fully resolved.
380: The \ion{Na}{1} transmission spectrum absorption for HD189733b,
381: measured in a ``narrow'' spectral band that encompasses both lines, is
382: $\left( -67.2 \pm 20.7 \right) \times 10^{-5}$.  The estimated error
383: includes both random and systematic contributions.  The amount of
384: absorption in \ion{Na}{1} due to HD189733b is $\sim$3$\times$ larger
385: than detected for HD209458b by \citet{charbonneau02}, and indicates
386: that the two exoplanets may have significantly different atmospheric
387: properties.  The excess absorption appears in a blue shifted
388: component, while the red half of the Na I doublet lines has a net
389: deficit of absorption during transit.  Future work will include the
390: analysis of other important optical lines (e.g., \ion{K}{1},
391: H$\alpha$), detailed analysis of the absorption profile and comparison
392: with atmospheric models, measurements of spectral indicators of
393: stellar variability, and observations of other transiting systems.
394: High resolution space-based spectroscopic observations will provide a
395: reliable confirmation of this detection, and access to important lines
396: in the ultraviolet.  Transmission spectra do not only have the
397: potential to probe the physical properties of hot-Jupiter and
398: hot-Neptune exoplanets, but lay the foundation toward detailed
399: physical diagnostics of Earth-sized extrasolar planetary atmospheres
400: \citep{ehrenreich06}.
401: 
402: \acknowledgements 
403: 
404: The authors would like to thank the referee, Ronald Gilliland, for his
405: insightful comments.  S.R. would like to acknowledge
406: support provided by the Hubble Fellowship grant HST-HF-01190.01
407: awarded by the STScI, which is operated by the AURA, Inc., for NASA,
408: under contract NAS 5-26555.  M.E. and W.D.C. acknowledge support
409: provided by NASA under grants NN05G107G and NNX07AL70G issued through
410: the TPF Foundation Science and the Origins of Solar Systems
411: programs. The Hobby-Eberly Telescope (HET) is a joint project of the
412: University of Texas at Austin, the Pennsylvania State University,
413: Stanford University, Ludwig-Maximilians-Universit\"{a}t M\"{u}nchen,
414: and Georg-August-Universit\"{a}t G\"{o}ttingen. The HET is named in
415: honor of its principal benefactors, William P. Hobby and Robert
416: E. Eberly.
417: 
418: {\it Facilities:} \facility{HET (HRS)}
419: 
420: \begin{thebibliography}{37}
421: \expandafter\ifx\csname natexlab\endcsname\relax\def\natexlab#1{#1}\fi
422: 
423: \bibitem[{{Bakos} {et~al.}(2006)}]{bakos06}
424: {Bakos}, G.~{\'A}., {et~al.} 2006, \apj, 650, 1160
425: 
426: \bibitem[{{Ballester} {et~al.}(2007){Ballester}, {Sing}, \&
427:   {Herbert}}]{ballester07}
428: {Ballester}, G.~E., {Sing}, D.~K., \& {Herbert}, F. 2007, \nat, 445, 511
429: 
430: \bibitem[{{Barman}(2007)}]{barman07}
431: {Barman}, T. 2007, \apjl, 661, L191
432: 
433: \bibitem[{{Bouchy} {et~al.}(2005)}]{bouchy05}
434: {Bouchy}, F., {et~al.} 2005, \aap, 444, L15
435: 
436: \bibitem[{{Bozorgnia} {et~al.}(2006){Bozorgnia}, {Fortney}, {McCarthy},
437:   {Fischer}, \& {Marcy}}]{bozorgnia06}
438: {Bozorgnia}, N., {Fortney}, J.~J., {McCarthy}, C., {Fischer}, D.~A., \&
439:   {Marcy}, G.~W. 2006, \pasp, 118, 1252
440: 
441: \bibitem[{{Brown}(2001)}]{brown01}
442: {Brown}, T.~M. 2001, \apj, 553, 1006
443: 
444: \bibitem[{{Brown} {et~al.}(2001){Brown}, {Charbonneau}, {Gilliland}, {Noyes},
445:   \& {Burrows}}]{brown01etal}
446: {Brown}, T.~M., {Charbonneau}, D., {Gilliland}, R.~L., {Noyes}, R.~W., \&
447:   {Burrows}, A. 2001, \apj, 552, 699
448: 
449: \bibitem[{{Bundy} \& {Marcy}(2000)}]{bundy00}
450: {Bundy}, K.~A., \& {Marcy}, G.~W. 2000, \pasp, 112, 1421
451: 
452: \bibitem[{{Burrows} {et~al.}(2007){Burrows}, {Hubeny}, {Budaj}, {Knutson}, \&
453:   {Charbonneau}}]{burrows07}
454: {Burrows}, A., {Hubeny}, I., {Budaj}, J., {Knutson}, H., \& {Charbonneau}, D.
455:   2007, \apjl, in press (ArXiv: 0709.3980)
456: 
457: \bibitem[{{Castelli} \& {Kurucz}(2006)}]{castelli06}
458: {Castelli}, F., \& {Kurucz}, R.~L. 2006, \aap, 454, 333
459: 
460: \bibitem[{{Charbonneau} {et~al.}(2002){Charbonneau}, {Brown}, {Noyes}, \&
461:   {Gilliland}}]{charbonneau02}
462: {Charbonneau}, D., {Brown}, T.~M., {Noyes}, R.~W., \& {Gilliland}, R.~L. 2002,
463:   \apj, 568, 377
464: 
465: \bibitem[{{Cho} {et~al.}(2003){Cho}, {Menou}, {Hansen}, \& {Seager}}]{cho03}
466: {Cho}, J.~Y.-K., {Menou}, K., {Hansen}, B.~M.~S., \& {Seager}, S. 2003, \apjl,
467:   587, L117
468: 
469: \bibitem[{{Cooper} \& {Showman}(2005)}]{cooper05}
470: {Cooper}, C.~S., \& {Showman}, A.~P. 2005, \apjl, 629, L45
471: 
472: \bibitem[{{Deming} {et~al.}(2006){Deming}, {Harrington}, {Seager}, \&
473:   {Richardson}}]{deming06}
474: {Deming}, D., {Harrington}, J., {Seager}, S., \& {Richardson}, L.~J. 2006,
475:   \apj, 644, 560
476: 
477: \bibitem[{{Ehrenreich} {et~al.}(2006){Ehrenreich}, {Tinetti}, {Lecavelier Des
478:   Etangs}, {Vidal-Madjar}, \& {Selsis}}]{ehrenreich06}
479: {Ehrenreich}, D., {Tinetti}, G., {Lecavelier Des Etangs}, A., {Vidal-Madjar},
480:   A., \& {Selsis}, F. 2006, \aap, 448, 379
481: 
482: \bibitem[{{Fitzpatrick} \& {Massa}(2005)}]{fitzpatrick05}
483: {Fitzpatrick}, E.~L., \& {Massa}, D. 2005, \aj, 129, 1642
484: 
485: \bibitem[{{Fortney} {et~al.}(2003){Fortney}, {Sudarsky}, {Hubeny}, {Cooper},
486:   {Hubbard}, {Burrows}, \& {Lunine}}]{fortney03}
487: {Fortney}, J.~J., {Sudarsky}, D., {Hubeny}, I., {Cooper}, C.~S., {Hubbard},
488:   W.~B., {Burrows}, A., \& {Lunine}, J.~I. 2003, \apj, 589, 615
489: 
490: %\bibitem[{{Hartkopf} {et~al.}(1989){Hartkopf}, {McAlister}, \&
491: %  {Franz}}]{hartkopf89}
492: %{Hartkopf}, W.~I., {McAlister}, H.~A., \& {Franz}, O.~G. 1989, \aj, 98, 1014
493: 
494: \bibitem[{{H{\'e}brard} \& {Lecavelier Des Etangs}(2006)}]{hebrard06}
495: {H{\'e}brard}, G., \& {Lecavelier Des Etangs}, A. 2006, \aap, 445, 341
496: 
497: \bibitem[{{Henry} \& {Winn}(2007)}]{henry07}
498: {Henry}, G.~W., \& {Winn}, J.~N. 2007, \apj, submitted (ArXiv: 0709.2142)
499: 
500: \bibitem[{{Knutson} {et~al.}(2007{\natexlab{a}}){Knutson}, {Charbonneau},
501:   {Allen}, {Burrows}, \& {Megeath}}]{knutson07inver}
502: {Knutson}, H.~A., {Charbonneau}, D., {Allen}, L.~E., {Burrows}, A., \&
503:   {Megeath}, S.~T. 2007{\natexlab{a}}, \apj, in press (ArXiv: 0709.3984)
504: 
505: \bibitem[{{Knutson} {et~al.}(2007{\natexlab{b}})}]{knutson07}
506: {Knutson}, H.~A., {et~al.} 2007{\natexlab{b}}, \nat, 447, 183
507: 
508: \bibitem[{{Koesterke} {et~al.}(2007){Koesterke}, {Allende Prieto}, \&
509:   {Lambert}}]{koesterke07}
510: {Koesterke}, L., {Allende Prieto}, C., \& {Lambert}, D.~L. 2007, \apj,
511:   submitted
512: 
513: \bibitem[{{Moutou} {et~al.}(2001){Moutou}, {Coustenis}, {Schneider}, {St
514:   Gilles}, {Mayor}, {Queloz}, \& {Kaufer}}]{moutou01}
515: {Moutou}, C., {Coustenis}, A., {Schneider}, J., {St Gilles}, R., {Mayor}, M.,
516:   {Queloz}, D., \& {Kaufer}, A. 2001, \aap, 371, 260
517: 
518: \bibitem[{{Narita} {et~al.}(2005)}]{narita05}
519: {Narita}, N., {et~al.} 2005, \pasj, 57, 471
520: 
521: \bibitem[{{Pont} {et~al.}(2007)}]{pont07}
522: {Pont}, F., {et~al.} 2007, \aap, submitted (ArXiv: 0707.1940)
523: 
524: \bibitem[{{Ramsey} {et~al.}(1998)}]{ramsey98}
525: {Ramsey}, L.~W., {et~al.} 1998, in Proc. SPIE Vol. 3352, Advanced Technology
526:   Optical/IR Telescopes VI, ed. L.~M. {Stepp}, 34
527: 
528: \bibitem[{{Redfield} \& {Linsky}(2002)}]{redfield02}
529: {Redfield}, S., \& {Linsky}, J.~L. 2002, \apjs, 139, 439
530: 
531: \bibitem[{{Seager} \& {Sasselov}(2000)}]{seager00}
532: {Seager}, S., \& {Sasselov}, D.~D. 2000, \apj, 537, 916
533: 
534: \bibitem[{{Tinetti} {et~al.}(2007)}]{tinetti07}
535: {Tinetti}, G., {et~al.} 2007, \nat, 448, 169
536: 
537: \bibitem[{{Tody}(1993)}]{tody93}
538: {Tody}, D. 1993, in ASP Conf. Ser. 52: Astronomical Data Analysis Software and
539:   Systems II, ed. R.~J. {Hanisch}, R.~J.~V. {Brissenden}, \& J.~{Barnes} (San
540:   Francisco: ASP), 173
541: 
542: \bibitem[{{Tull}(1998)}]{tull98}
543: {Tull}, R.~G. 1998, in Proc. SPIE Vol. 3355, Optical Astronomical
544:   Instrumentation, ed. S.~{D'Odorico}, 387
545: 
546: \bibitem[{{Vallerga} {et~al.}(1993){Vallerga}, {Vedder}, {Craig}, \&
547:   {Welsh}}]{vallerga93}
548: {Vallerga}, J.~V., {Vedder}, P.~W., {Craig}, N., \& {Welsh}, B.~Y. 1993, \apj,
549:   411, 729
550: 
551: \bibitem[{{Vidal-Madjar} {et~al.}(2003){Vidal-Madjar}, {Lecavelier des Etangs},
552:   {D{\'e}sert}, {Ballester}, {Ferlet}, {H{\'e}brard}, \&
553:   {Mayor}}]{vidalmadjar03}
554: {Vidal-Madjar}, A., {Lecavelier des Etangs}, A., {D{\'e}sert}, J.-M.,
555:   {Ballester}, G.~E., {Ferlet}, R., {H{\'e}brard}, G., \& {Mayor}, M. 2003,
556:   \nat, 422, 143
557: 
558: \bibitem[{{Vidal-Madjar} {et~al.}(2004)}]{vidalmadjar04}
559: {Vidal-Madjar}, A., {et~al.} 2004, \apjl, 604, L69
560: 
561: \bibitem[{{Welsh} {et~al.}(1991){Welsh}, {Vedder}, {Vallerga}, \&
562:   {Craig}}]{welsh91}
563: {Welsh}, B.~Y., {Vedder}, P.~W., {Vallerga}, J.~V., \& {Craig}, N. 1991, \apj,
564:   381, 462
565: 
566: \bibitem[{{Winn} {et~al.}(2004){Winn}, {Suto}, {Turner}, {Narita}, {Frye},
567:   {Aoki}, {Sato}, \& {Yamada}}]{winn04}
568: {Winn}, J.~N., {Suto}, Y., {Turner}, E.~L., {Narita}, N., {Frye}, B.~L.,
569:   {Aoki}, W., {Sato}, B., \& {Yamada}, T. 2004, \pasj, 56, 655
570: 
571: \bibitem[{{Winn} {et~al.}(2006)}]{winn06}
572: {Winn}, J.~N., {et~al.} 2006, \apjl, 653, L69
573: 
574: \bibitem[{{Winn} {et~al.}(2007)}]{winn07}
575: {Winn}, J.~N., {et~al.} 2007, \aj, 133, 1828
576: 
577: \end{thebibliography}
578: 
579:  
580: %figures
581: \clearpage
582: \begin{figure}
583: \epsscale{1.2}
584: \plottwo{f1a.eps}{f1b.eps}
585: \caption{{\it top left:} Spectrum of HD189733 near the \ion{Na}{1}
586: doublet.  {\it bottom left:} The difference of the relative flux of
587: the in-transit template (${\cal F}_{\rm in}$) and the out-of-transit
588: template (${\cal F}_{\rm out}$).  Excess absorption in the in-transit
589: template is clearly detected in both \ion{Na}{1} lines.  Also shown is
590: the contribution of differential limb darkening (red line and filled
591: circles), which has been removed from the data, but the impact is
592: minimal.  The vertical dashed lines indicate the narrow bandpass used
593: to calculate the relative transmission spectrum absorption, and is
594: identical to that used by \citet{charbonneau02}.  The righthand axis
595: gives the effective radius of the planet as a function of wavelength,
596: assuming stellar and planetary properties from \citet{pont07}.  {\it
597: right:} Distributions of the empirical Monte Carlo analysis, which
598: demonstrate the stability of the observations and the contribution of
599: systematic errors.  The full dataset transmission spectrum absorption
600: measurement is indicated by the vertical dashed line.  Given the
601: random and systematic errors measured by the standard deviation of the
602: ``out-out'' distribution, this exoplanetary absorption detection is
603: $>$3$\sigma$.\label{fig:na}}
604: \end{figure}
605: 
606: %figures
607: \clearpage
608: \begin{figure}
609: \epsscale{1.2}
610: \plottwo{f2a.eps}{f2b.eps}
611: \caption{Same as Figure~\ref{fig:na}, but for the \ion{Ca}{1} line,
612: which like \ion{Na}{1} is a strong line, but predicted not to show
613: absorption in the transmission spectrum.\label{fig:ca}}
614: \end{figure}
615: 
616: \end{document}
617: