1: \documentclass[10pt,preprint2]{aastex}
2: \parskip .7em % sets inter para spacing at .7ems
3: \textheight 10.0in % sets textheight
4: \textwidth 7.5 in % sets textwidth
5: \hoffset -0.50in % sets vert. and horiz offsets
6: \voffset -1.2in
7:
8: \usepackage{wasysym}
9:
10: \def\Rsat{$\mathrm{R}_{\mathrm{\saturn}}$}
11:
12: \shorttitle{Eccentric Transiting Planets}
13: \shortauthors{Barnes}
14:
15: \begin{document}
16:
17: \title{Effects of Orbital Eccentricity on Extrasolar Planet Transit
18: Detection and Lightcurves}
19: \author{Jason W. Barnes}
20: \affil{NASA Ames Research Center}
21: \affil{M/S 244-30}
22: \affil{Moffett Field, CA 94035}
23: \email{jason@barnesos.net}
24:
25: \newpage
26:
27: \begin{abstract}
28:
29: It is shown herein that planets with eccentric orbits are more likely to transit
30: than circularly orbiting planets with the same semimajor axis by a factor of
31: $(1-e^2)^{-1}$. If the orbital parameters of discovered transiting planets are
32: known, as from follow-up radial velocity observations, then the transit-detected
33: planet population is easily debiased of this effect. The duration of a planet's
34: transit depens upon of its eccentricity and longitude of periastron; transits near
35: periastron are shorter, and those near apoastron last longer, for a given impact
36: parameter. If fitting for the stellar radius with the other transit parameters,
37: this effect causes a systematic error in the resulting measurements. If the
38: stellar radius is instead held fixed at a value measured independently, then it is
39: possible to place a lower limit on the planet's eccentricity using photometry
40: alone. Orbital accelerations cause a difference in the planet's ingress and egress
41: durations that lead to an asymmetry in the transit lightcurve that could be used
42: along with the transit velocity measurement to uniquely measure the planet's
43: eccentricity and longitude of periapsis. However, the effect is too small to be
44: measured with current technology. The habitability of transiting terrestrial
45: planets found by \emph{Kepler} depends on those planets' orbital eccentricities.
46: While \emph{Kepler} will be able to place lower limits on those planets' orbital
47: eccentricity, the actual value for any given planet will likely remain unknown.
48:
49:
50: \end{abstract}
51:
52: \keywords{occultations --- planets and satellites: individual (HD80606b,
53: HD147506b) --- techniques: photometric}
54:
55:
56:
57: \section{INTRODUCTION}
58:
59: There are presently 21 extrasolar planets known to transit their parent stars ({\tt
60: http://exoplanet.eu/}). Radial velocity measurements of all but one of them are
61: consistent (within errors) with circular orbits; i.e., zero eccentricity
62: \citep[\emph{e.g.},][]{2005ApJ...629L.121L}. Presumably, any initial eccentricity
63: in those orbits has since been damped by tidal circularization
64: \citep{2000ApJ...537L..61T}. In light of the discovery of the first transiting
65: extrasolar planet with an eccentric orbit, HD147506b, I explore the effect that
66: orbital eccentricity has on transit lightcurves with an eye toward the data to come
67: from \emph{CoRoT} and \emph{Kepler}.
68:
69: Since tidal circularization is most effective at short planet-star distances, as
70: transit search programs extend into longer-period regimes the prospects for
71: detecting non-circularly orbiting planets grows. Perhaps not coincidentally, the
72: first known transiting planet on an eccentric orbit, HD147506b ($e=0.507$), is the
73: transiting planet with the longest period (5.63 days) \citep{2007arXiv0705.0126B}.
74: Recently, \emph{Spitzer} measurements of the relative timing of the secondary
75: eclipse of GJ436b have confirmed that planet's nonzero orbital eccentricity and
76: measured it to be $e=0.14\pm.01$ \citep{2007arXiv0707.3809D}.
77:
78: With the space-based transit searches of \emph{CoRoT} \citep{2003A&A...405.1137B}
79: and particularly \emph{Kepler} \citep{2005NewAR..49..478B}, hundreds of transiting
80: planets will be found that will not have been tidally circularized. Based on the
81: findings from radial velocity planet searches \citep{2006ApJ...646..505B}, many of
82: these newly-discovered transiting planets are likely to follow eccentric orbits.
83:
84: Orbital eccentricity has several effects on planetary transits. The timing of the
85: transit, relative to that of the secondary eclipse, is a strong function of the
86: planet's orbital eccentricity and longitude of periastron
87: \citep[\emph{e.g.},][]{2005ApJ...629L.121L,2007AJ....133.1828W}. Planets with
88: eccentric orbits, if sufficiently close to their parent stars, can have their
89: rotations brought into tidal equilibrium at rotation rates greater than their mean
90: motions \citep{oblateness.2003}; this affects planetary transit lightcurves via
91: the planet's oblateness \citep{Seager.oblateness,oblateness.2003}. These effects
92: and others have been explored in the context of eclipsing binary stars on eccentric
93: orbits as well \citep[\emph{e.g.}][]{1972ApJ...174..617N}.
94:
95: In this paper, I explore three additional ways that orbital eccentricity affects
96: the transits of extrasolar planets. First, I calculate the increased transit
97: probability for planets on eccentric orbits. Next I point out the variability in
98: transit duration that results from planets moving faster near periastron and
99: slower near apoastron. The third effect that I explore is the possibility for
100: asymmetric transit lightcurves induced as the planet's trajectory evolves between
101: transit ingress and egress. I then use least-squares fits of synthetic transit
102: lightcurves to determine whether or not these effects can be used to constrain the
103: orbital elements of transiting planets from photometry alone.
104:
105: Though orbital eccentricity is not inherent to the planet itself, I will sometimes
106: refer to planets on eccentric orbits as `eccentric planets' for brevity, following
107: \citep{Lissauer.1997.RotationfromEccentricAccretion}.
108:
109:
110: \section{EFFECTS}\label{section:effects}
111:
112: An extrasolar planet on an eccentric orbit has three primary differences
113: relative to that same planet on a circular orbit with the same semimajor axis.
114: The eccentric planet is more likely to transit, and, if it does, then the
115: transit duration depends upon both the impact parameter $b$ and the orbital true
116: anomaly $f$, and the
117: transit lightcurve may be asymmetric due to accelerations during the transit.
118:
119: \subsection{Transit Probability}\label{section:probability}
120:
121: Planets on eccentric orbits are more likely to transit than equivalent planets
122: with the same semimajor axis but circular orbits. Though these planets spend a
123: majority of their \emph{time} at greater asterocentric distances than their
124: semimajor axes, they spend a majority of their \emph{true anomalies} at smaller
125: asterocentric distances. The probability for a planet on a circular orbit to
126: transit was derived by \citet{1984Icar...58..121B} based on the solid angle swept
127: out by a planet's shadow: \begin{equation}\label{eq:borucki} p~=~\frac{R_*}{r_\mathrm{p}}
128: \end{equation} where $p$ is the transit probability, $R_*$ is the stellar radius,
129: and $r_\mathrm{p}$ is the distance between the planet and the star.
130:
131: Using the method of \citet{1984Icar...58..121B}, then, the transit
132: probability for an extrasolar planet
133: is equal to the solid angle swept out by the planet's shadow, a
134: function of both $f$ and the polar angle from the orbit plane $\theta$, normalized
135: to $4\pi$ steradians:
136: \begin{equation}
137: p~=~\frac{1}{4\pi}\int_0^{2\pi}\int_{\theta_0}^{\theta_1}\cos\theta~\mathrm{d}\theta~\mathrm{d}f
138: \end{equation}
139: using $\theta_0$ and $\theta_1$ for the angular
140: extent of the shadow below and above the orbital plane. Due to the symmetry of
141: the problem $\theta_0~=~-\theta_1$. Geometry sets
142: $\theta_1~=~\sin^{-1}\frac{R_*}{r_\mathrm{p}}$ (see Figure \ref{figure:geometry}).
143: Hence, integrating over $\theta$,
144: \begin{equation}
145: p~=~\frac{1}{4\pi}\int_0^{2\pi}~\sin\theta\big|^{\theta_1}_{\theta_0}~\mathrm{d}f
146: \end{equation}
147: and plugging in $\theta_0$ and $\theta_1$
148: \begin{equation}
149: p~=~\frac{1}{4\pi}\int_0^{2\pi}~\frac{2R_*}{r_\mathrm{p}}~\mathrm{d}f
150: \end{equation}
151:
152: For planets on eccentric orbits, $r_\mathrm{p}$ varies with time. However, the variation
153: of $r_\mathrm{p}$ as a function of $f$ is all that matters for
154: determining the solid angle over which the planet will transit.
155: \begin{equation}\label{eq:orbit}
156: r_\mathrm{p}~=~\frac{a_\mathrm{p}(1-e^2)}{1+e~\cos f}
157: \end{equation}
158: \citep{SolarSystemDynamics} where $e$ is the planet's orbital eccentricity and
159: $a_\mathrm{p}$ its semimajor axis.
160:
161: Plugging in $r_\mathrm{p}$ from Equation \ref{eq:orbit} and integrating over $f$
162: \begin{equation}
163: p~=~\int_0^{2\pi}{\frac{2R_*~(1+e~\cos f)}{a_\mathrm{p}(1-e^2)}}~\mathrm{d}f .
164: \end{equation}
165: leads to
166: \begin{equation}
167: p~=~\frac{1}{4\pi}~\frac{2R*}{a_\mathrm{p}(1-e^2)}~\int_0^{2\pi}{1+e\cos(f)}~\mathrm{d}f .
168: \end{equation}
169: which leads to the result that for planets on eccentric orbits
170: \begin{equation}\label{eq:eccentric.probability}
171: p~=~\frac{R*}{a_\mathrm{p}(1-e^2)}
172: \end{equation}
173: since $\int_0^{2\pi}~\mathrm{d}f~=~2\pi$ and
174: $\int_0^{2\pi}\cos f~\mathrm{d}f~=~0$.
175:
176: The above derivation is valid for transit impact parameters $b<=1$. To exclude
177: all grazing transits, replace the $R_*$ in the numerator of Equation
178: \ref{eq:eccentric.probability} with $R_*-R_\mathrm{p}$, where $R_\mathrm{p}$ is
179: the radius of the transiting planet. Similarly, to include all transits, no
180: matter how grazing, the numerator of Equation \ref{eq:eccentric.probability} would
181: be $R_*+R_\mathrm{p}$.
182:
183: The increased transit probability for even a planet with a significantly eccentric
184: orbit with $e=0.5$, similar to that for HD147506b \citep{2007arXiv0705.0126B}, is
185: modest: 33\%. However, the increased probability for planets on extremely
186: eccentric orbits like HD80606b \citep{2001A&A...375L..27N} with $e=0.93$ is
187: 640\%! Given that 28 of the 224 planets with radial velocity orbits have $e\geq
188: 0.5$ ({\tt http://exoplanet.eu/}), transit surveys should detect a significant
189: number of planets on eccentric orbits. Nearly half, 110 out of 224, of radial
190: velocity planets are more eccentric than our solar system's most eccentric planet,
191: Mercury ($e=0.2056$). Planets with extreme orbital eccentricities will be
192: detected
193: at a rate decidedly higher than their occurrance would predict given Equation
194: \ref{eq:borucki}.
195:
196: \begin{figure}
197: \plotone{f1.eps}
198: \caption{Geometry of a planetary orbit and the solid angle swept out by
199: its shadow, modified from \citet{1984Icar...58..121B} to account for orbital eccentricity.
200: \label{figure:geometry}}
201: \end{figure}
202:
203: This excess will lead to a bias in the raw planet incidence as a function of
204: semimajor axis derived from \emph{CoRoT} and \emph{Kepler} discoveries. The
205: bias can be easily corrected by accounting for the $(1-e^2)^{-1}$ detection
206: increase factor from Equation \ref{eq:eccentric.probability}, or, more formally,
207: by using the original \citet{1984Icar...58..121B} probability (Equation
208: \ref{eq:borucki}) while substituting the instantaneous planet-star distance
209: at mid-transit for the orbital radius for each detected planet.
210:
211: Debiasing requires knowledge of the planet's orbital eccentricity
212: and longitude of periapsis, which would be difficult to ascertain for planets too
213: small to induce detectable radial velocity variations (Section
214: \ref{section:detectability}). Although this extra step adds complexity, it is
215: heartening to note that the orbital eccentricity bias induced in planet
216: distrubutions as derived using the transit method is precisely calculable and
217: removable. The orbital eccentricity bias in radial velocity planet surveys (as
218: results from data gaps near periastron passage, for instance) is known less
219: precisely and is therefore more challenging to remove.
220:
221: The differing transit probability at periastron and apoastron can lead to planets
222: that transit, but have no secondary eclipse if the planet is sufficiently
223: eccentric, inclined, and transits near periastron. Alternately there can exist
224: planets that show secondary eclipses but no primary transit. In the case where
225: giant planets have just secondary eclipses, the small secondary eclipse depth
226: could be mistaken for the primary transit of a terrestrial-sized planet. Careful
227: monitoring of reflected-light phase variability (as described in
228: \citet{2002ApJ...575..493J}) may eliminate this source of systematic error.
229: Secondary-eclipse-only planets would need to be very near their parent stars
230: during secondary eclipse in order to have sufficient reflected light so as to
231: mimic the transit of a terrestrial-sized planet. Faint secondary stars in similar
232: orbits could mimic the terrestrial planet transit regardless of their distance
233: from the primary star at secondary eclipse.
234:
235:
236:
237:
238: \subsection{Transit Duration}\label{section:duration}
239:
240: For a given star mass, planets with the same orbital semimajor axes have the same
241: energy per unit mass ($-GM_*/2a_p$); those with higher eccentricities have lower
242: specific angular momenta. As such, those more highly eccentric objects move
243: faster near periastron, and slower near apoastron. If such an eccentric planet
244: were to transit, it would then have transits of shorter or longer duration than
245: the equivalent planet on a circular orbit transiting with the same impact
246: parameter ($b$).
247:
248: The velocity of a circularly orbiting planet $V_\mathrm{circ}$ is
249: \begin{equation}
250: V_\mathrm{circ}~=~\sqrt{\frac{GM_*}{a_p}}~.
251: \end{equation}
252: Using conservation of energy, the
253: periapsis velocity $V_\mathrm{peri}$ of a planet with orbital eccentricity
254: $e$ can be shown to be
255: \begin{equation}\label{eq:periastron}
256: V_\mathrm{peri}~=~\sqrt{\frac{1+e}{1-e}}\sqrt{\frac{GM_*}{a_p}}~=~
257: \sqrt{\frac{1+e}{1-e}}~V_\mathrm{circ}~=~\sqrt{1+e}~V_\mathrm{pericirc}~.
258: \end{equation}
259: In the case of planets discovered by their transits, $a_p$ is set by the
260: planet's known orbital period, and hence the comparison to
261: $V_\mathrm{circ}$ is most relevant.
262: For convenience I also compare $V_\mathrm{peri}$ to
263: $V_\mathrm{pericirc}$, the velocity of
264: a planet orbiting circularly at the eccentric planet's periastron.
265: Similarly, at apoapsis
266: \begin{equation}\label{eq:apoastron}
267: V_\mathrm{apo}~=~\sqrt{\frac{1-e}{1+e}}\sqrt{\frac{GM_*}{a_p}}~=~
268: \sqrt{\frac{1-e}{1+e}}~V_\mathrm{circ}~=~\sqrt{1-e}~V_\mathrm{apocirc}
269: \end{equation}
270: for similarly named variables (see Figure \ref{figure:eccorbit}).
271:
272: \begin{figure}
273: \plotone{f2.eps}
274: \caption{This figure illustrates the variables described in Section
275: \ref{section:duration} of the text, as applied to a hypothetical planet
276: with $e~=~0.5$.
277: \label{figure:eccorbit}}
278: \end{figure}
279:
280: The pericenter and apocenter velocities behave as expected in their
281: extremes. $V_\mathrm{peri}$ approaches escape velocity
282: ($\sqrt{2}~V_\mathrm{pericirc}$) as $e\to1$ but is undefined for non-closed
283: orbits where $e\geq1$. $V_\mathrm{apo}\to0$ as $e\to1$, in absolute terms, as a
284: function of $V_\mathrm{circ}$, and as a function of $V_\mathrm{apocirc}$.
285:
286: A significantly eccentric planet with $e=0.5$ (similar to HD147506b for which
287: $e=0.507$) travels $\sqrt{3}$ ($\sim1.73$) times faster at periapsis than you
288: would expect given its period and assuming a circular orbit, and $\sqrt{3}$ times
289: more slowly at apoapsis. This planet's transit duration, if it were to transit
290: at periastron, would be 58\% as long for a particular impact parameter as the
291: transit of its circularly orbiting equivalent. If transiting at apoastron (three
292: times less probable than a periastron transit; see Section
293: \ref{section:probability}), the transit would last 73\% longer than the
294: equivalent circularly orbiting planet.
295:
296: A planet on an extremely eccentric orbit like HD80606b ($e=0.93$)
297: \citep{2001A&A...375L..27N} would have a periastron transit duration only 19\% that of
298: its same-semimajor-axis circular equivalent, and 72\% that of a planet orbiting
299: circularly at HD80606b's periastron. Conversely if HD80606b were to transit near
300: apoastron, such a transit would last 5.25 times longer than if HD80606b were to
301: transit in a circular orbit with the same semimajor axis, and 3.78 times longer
302: than if HD80606b were to transit in a circular orbit at its true apoastron.
303:
304: The duration of a exoplanetary transit depends on the chord length of the planet's
305: apparent passage in front of the star ($2R_*\cos b$) and the planet's azimuthal
306: velocity $V_{f}~=~r\dot{f}$. In the cases discussed above, when a planet is at
307: periastron and apoastron, $V_\mathrm{peri}$ and $V_\mathrm{apo}$ are equal to
308: $V_f$. The rest of the time, $V_f$ if not equal to the planet's full velocity as
309: the planet will also have a velocity component radial to the star ($V_r$).
310: According to
311: \citet{SolarSystemDynamics}, $V_f$ varies sinusoidally with $f$:
312: \begin{equation} \label{eq:velocity}
313: V_f~=~V_\mathrm{circ}\frac{1+e\cos f}{\sqrt{1-e^2}}~\mathrm{.}
314: \end{equation}
315: Hence the planet's velocity is greater than the equivalent circular orbit velocity
316: for more than half of the orbit as a function of the true anomaly.
317:
318:
319: \subsection{Transit Symmetry}\label{section:asymmetry}
320:
321: The dependence of the planet's azimuthal velocity on $f$ as shown in Equation
322: \ref{eq:velocity} belies to another effect that eccentric orbits have on
323: transits. Because $V_f$ changes slightly between ingress and egress (unless the
324: planet is at periastron or apoastron mid-transit), eccentric planets can produce
325: asymmetric transit lightcurves. If a planet transits after periastron and before
326: apoastron, the time that it takes for the planet to ingress across the star's limb
327: will be shorter than the time that it takes to egress. Similarly if the planet is
328: between apoastron and periastron, then the ingress will be longer than the egress.
329:
330: To calculate the velocity difference between ingress and egress, I first take
331: $V_f$ from Equation \ref{eq:velocity} and differentiate it with respect to $f$:
332: \begin{equation}
333: \frac{\mathrm{d}V_f}{\mathrm{d}f}~=~-\frac{e~V_\mathrm{circ}}{\sqrt{1-e^2}}~\sin f \mathrm{.}
334: \end{equation}
335: The total ingress-egress velocity difference, $\Delta V$, is equal to
336: $\frac{\mathrm{d}V_f}{\mathrm{d}f}$ times $\Delta f$, the difference of the true
337: anomaly between ingress and egress, under the assumption that
338: $\frac{\mathrm{d}V_f}{\mathrm{d}f}$ varies negligibly across the transit. Taking
339: $f_0$ to be the true anomaly of the planet at mid-transit,
340: \begin{equation}
341: \Delta f~=~\frac{2R_*}{r_\mathrm{p}(f_0)}
342: \end{equation}
343: assuming that $r_\mathrm{p}$ varies only slowly during the transit. Now, plugging in $r_\mathrm{p}$
344: from Equation \ref{eq:orbit},
345: \begin{equation}\label{eq:deltaV}
346: \Delta V~=~\frac{\mathrm{d}V_f}{\mathrm{d}f}~\Delta f~=~-
347: \frac{2R_*eV_\mathrm{circ}}{a(1-e^2)^{3/2}}\sin f_0~(1+e\cos f_0)~\mathrm{.}
348: \end{equation}
349: I show a plot of the varying component of $\Delta V$ for various values of $e$ in
350: Figure \ref{figure:deltaV}.
351: The \emph{fractional} difference in velocity, $\Delta V/V$, is a bit simpler:
352: \begin{equation}
353: \frac{\Delta V}{V}~=~-\frac{2R_*e}{a(1-e^2)}\sin f_0~\mathrm{,}
354: \end{equation}
355: and has an evident maximum where $f_0=\pm\pi/2$. Hence, the greatest
356: \emph{fractional} variation in ingress and egress duration will occur ninety degrees
357: away from periastron and apoastron, as measured in the planet's true anomaly.
358:
359: \begin{figure}
360: \plotone{f3.eps}
361: \caption{Velocity difference between egress and ingress as a function of the
362: planet's orbital true anomaly at mid-transit, $f_0$. I have only plotted that
363: portion of Equation \ref{eq:deltaV} that is a function of $f$ and $e$; to convert to
364: true $\Delta V$, multiply by $\frac{2R_*V_\mathrm{circ}}{a}$. As an example, the
365: multiplier for a planet in an $a_p=0.1$AU orbit around a $1M_\odot$ star is 8.76
366: km/s.
367: \label{figure:deltaV}}
368: \end{figure}
369:
370: To determine where the \emph{absolute} velocity difference $\Delta V$ is maximized, I
371: differentiate $\Delta V$ with respect to $f$ and set the result equal to zero:
372: \begin{equation}
373: 0~=~\frac{\mathrm{d}\Delta
374: V}{\mathrm{d}f}~=~-\frac{2R_*eV_\mathrm{circ}}{a(1-e^2)^\frac{3}{2}}
375: \frac{\mathrm{d}\big[\sin f_0+e\sin f_0\cos f_0\big]}{\mathrm{d}f}~\mathrm{.}
376: \end{equation}
377: Differentiating and using the double-angle formula $\sin (2x)=2\sin x \cos x$, I
378: determine that
379: \begin{equation}\label{eq:maxdeltaV}
380: \cos f_0+e\cos (2f_0)~=~0~\mathrm{.}
381: \end{equation}
382: I show the solution to this equation in Figure \ref{figure:maxdeltaV}.
383: The maximum velocity difference occurs at $f_0=\pm\pi/2$ for infinitessimal
384: eccentricities, and approaches $f_0=\pm\pi/3$ as $e\to1$.
385:
386: \begin{figure}
387: \plotone{f4.eps}
388: \caption{The true anomaly at mid-transit
389: ($f_0$, here in units of radians divided by $\pi$) for which
390: the difference between a planet's transit ingress and egress velocity is maximized
391: as a function of the planet's orbital eccentricity, $e$.
392: \label{figure:maxdeltaV}}
393: \end{figure}
394:
395: \begin{figure}
396: \plotone{f5.eps}
397: \caption{To-scale diagram of the HD147506 system. Ingress and egress times listed
398: are in minutes~:~seconds format. Figure concept inspired by Gregory Laughlin's
399: {\tt http://oklo.org/} website.
400: \label{figure:HD147506b}}
401: \end{figure}
402:
403: To illustrate the consequences of this effect, I apply the results derived above
404: to newly-discovered eccentric transiting planet HD147506b
405: \citep{2007arXiv0705.0126B}. HD147506b was determined by its discoverers,
406: \citet{2007arXiv0705.0126B}, to have a radius of $1.18~\mathrm{R_{Jup}}$, an
407: orbital semimajor axis of $0.0685~\mathrm{AU}$, an orbital eccentricity of
408: $0.507$, and a longitude of periastron of $184.6^\circ$. The planet's parent
409: star, HD147506, was determined to have a radius of $1.8~\mathrm{R_\odot}$ and a
410: mass of $1.35~\mathrm{M_\odot}$ (note that refined system parameters were
411: published by \citet{2007arXiv0707.0679L} while this paper was in review -- the new
412: values do not change the qualitative results that I describe here, but future work
413: should employ these newer values). I show a to-scale schematic of the system in
414: Figure \ref{figure:HD147506b}.
415:
416: From Equation \ref{eq:velocity} HD147506b's $V_\mathrm{circ}$ is
417: $132.2~\mathrm{km~s^{-1}}$. This planet's orbital parameters are particularly
418: favorable with respect to maximizing the magnitude of $\Delta~V_f$. HD147506b's
419: large eccentricity, fast $V_\mathrm{circ}$, and nearly ideal true anomaly at
420: midtransit ($f_\mathrm{0}=4.84$, very close to $-\pi/2$), combined with HD147506's
421: large stellar radius yield $\Delta~V_f~=~-24.6~\mathrm{km~s^{-1}}$. Calculating
422: the ingress and egress times $\tau$ using
423: \begin{equation}
424: \tau = \frac{R_\mathrm{p}}{V_f~\cos(\sin^{-1}b)}
425: \end{equation}
426: with the appropriate $V_f$ values for the transit ingress and egress of HD147506b
427: yields $\tau_\mathrm{ingress}~=~10~\mathrm{min}~36~\mathrm{sec}$ and
428: $\tau_\mathrm{egress}~=~8~\mathrm{min}~37~\mathrm{sec}$.
429:
430:
431:
432: \section{DETECTABILITY}\label{section:detectability}
433:
434: The \emph{CoRoT} and \emph{Kepler} missions will discover hundreds of new
435: transiting extrasolar planets \citep{2003A&A...405.1137B,2005NewAR..49..478B}.
436: The most massive of these will be amenable to radial velocity follow-up
437: observations to measure their masses; a more complete set of radial velocity
438: measurements (\emph{i.e.}, covering the full orbit and not just the radial
439: velocity maxima and minima, which are all that would be required to determine mass
440: assuming a circular orbit and the epoch and period as established by the transit)
441: can determine the planets' orbital eccentricities and longitudes of periastron.
442: However for planets of Neptune-mass and smaller, in larger orbits, or orbiting
443: fainter stars, radial velocity follow-up will not be practical or in some cases
444: possible. In those cases, it would be useful to attempt to constrain the orbital
445: eccentricity of those planets using transit photometry alone. Such a
446: determination would bear critically on the climatic variability of Earth-like
447: worlds; highly eccentric planets may not be habitable even if their orbital
448: semimajor axes place them within their stars' habitable zone.
449:
450: To determine the photometric detectability of the transit duration and asymmetry
451: effects of eccentric planet orbits, I create synthetic transit lightcurves that I
452: then fit as if the orbit were circular. I assume knowledge of the planet's period,
453: as that value will be measured by the time between transits in \emph{Kepler} and
454: \emph{CoRoT} data. I calculate both the synthetic lightcurves and the best-fit
455: solutions using the analytical approximation of \citet{2002ApJ...580L.171M}. The
456: \citet{2002ApJ...580L.171M} formulation assumes that the portion of the star
457: covered by the planet has uniform surface brightness, but that brightness accounts
458: for limb-darkening; hence it is least accurate for ingress and egress. However,
459: the detrimental effects on this particular calculation should be minimal since both
460: the synthetic and best-fit lightcurves should show the same systematic errors,
461: which, when subtracted, should leave a good estimate for the proper fit residuals.
462: The effects of light-travel delay illustrated by \citet{2005ApJ...623L..45L} are
463: not included.
464:
465: For larger planets, those where the ingress and egress are temporally resolved,
466: that have high signal-to-noise lightcurves, I first fit for $R_*$, $R_\mathrm{p}$,
467: the transit impact parameter $b$, and a stellar limb darkening coefficient $c_1$,
468: the treatment applied to HD209458b by \citet{2001ApJ...552..699B}. I assumed a
469: $1~R_\mathrm{Jup}$ planet orbiting a $1~R_\odot$, $1~M_\odot$ star with
470: $a_p=0.1~\mathrm{AU}$ and $e=0.5$. Since $R_{\mathrm{p}}/R_*$ is set by the
471: transit depth, and $b$ by the ingress/egress time relative to the total transit
472: duration, orbital eccentricity in this type of fit systematically affects $R_*$,
473: which governs the total transit timescale. In my test runs, the best-fit
474: $R_{*\mathrm{measured}}$ varies as $R_{*\mathrm{measured}}=R_* V_\mathrm{circ}
475: V_{f_0}^{-1}$. Hence assuming a circular orbit when the orbit is actually
476: eccentric leads to a systematic error in the measurement of $R_*$ and, by
477: extension, $R_\mathrm{p}$ ($R_\mathrm{p}/R_*$ is unaffected). The measured values
478: are smaller than the actual values if the planet is near periastron, and are
479: larger if the planet is near apoastron.
480:
481: This systematic error can be addressed by assuming a stellar radius, as could be
482: estimated by other observations such as parallax, spectral type, and stellar
483: absolute magnitude. In this case, the transit timescale can be set by fitting
484: explicitly for $V_{f_0}$ in addition to $r_\mathrm{p}$, $b$, and $c_1$. However, without
485: knowledge of the mid-transit true anomaly $f_0$, $V_{f_0}$ cannot uniquely
486: determine the orbital eccentricity. Instead, $V_{f_0}$ can \emph{constrain} $e$
487: if we allow that the planet must have a minimum eccentricity in order that $V_f$
488: reach $V_{f_0}$. If $V_{f_0}~>~V_\mathrm{circ}$, then from Equation
489: \ref{eq:periastron}
490: \begin{equation}\label{eq:eminperi}
491: e~\geq~\frac{\big(\frac{V_{f_0}}{V_\mathrm{circ}}\big)^2-1}
492: {\big(\frac{V_{f_0}}{V_\mathrm{circ}}\big)^2+1}
493: \end{equation}
494: and if $V_{f_0}~<~V_\mathrm{circ}$, then similarly from Equation
495: \ref{eq:apoastron}
496: \begin{equation}\label{eq:eminapo}
497: e~\geq~\frac{1-\big(\frac{V_{f_0}}{V_\mathrm{circ}}\big)^2}
498: {1+\big(\frac{V_{f_0}}{V_\mathrm{circ}}\big)^2}~\mathrm{.}
499: \end{equation}
500: The above lower limits can only be placed when the planet's ingress and egress are
501: resolved. In order to constrain the eccentricity of terrestrial-sized planets,
502: both high temporal cadence and high photometric precision would be necessary.
503: Resolving the ingress of the Earth, with $\tau_\mathrm{ingress}=7.04~\mathrm{minutes}$ at $b=0$,
504: will not be possible given \emph{Kepler} data alone. Coaddition of
505: multiple transits from multiple telescopes might help to constrain the ingress
506: times for detected transiting terrestrial planets.
507:
508: The degeneracy between $e$ and $f_0$, multiple cominations of which can produce
509: the same $V_{f_0}$, can be broken by measuring the transit asymmetry outlined in
510: Section \ref{section:asymmetry}. To measure the detectability of the asymmetry, I
511: produce a synthetic lightcurve for HD147506b ($r_\mathrm{p}=1.18\mathrm{R_{Jup}}$,
512: $R_*=1.8\mathrm{R_\odot}$, $a_\mathrm{p}=0.0685~\mathrm{AU}$, $e=0.507$,
513: $f_0=4.84$, and assuming \citet{2001ApJ...552..699B} limb darkening coefficient
514: $c_1=0.64$) that I fit using a model planet with a circular orbit. The
515: residuals, which I refer to as the detectability, are shown in Figure
516: \ref{figure:residual} for transits at several impact parameters.
517:
518: \begin{figure}
519: \plotone{f6.eps}
520: \caption{Detectability of the transit lightcurve asymmetry induced by the orbital
521: eccentricity of planet HD147506b, for impact parameters $b=0.0$, $b=0.3$, $b=0.7$,
522: and $b=0.85$.
523: \label{figure:residual}}
524: \end{figure}
525:
526: When faced with an ingress and egress of differing duration, the best-fit
527: circular-orbit planet model splits the difference. On ingress for HD147506b,
528: which transits before periastron ($f_0=-1.44$), the real planet is moving more
529: slowly than the model planet. In the difference lightcurve, the real planet hits
530: the stellar limb \emph{before} the model, such that at mid-ingress the real and
531: model planets are in the same place. So the real-minus-model
532: detectability initially trends negative, reaches zero at mid-ingress, and then
533: trends positive until the end of the planet's ingress. The detectability varies
534: slowly between the end of ingress and the beginning of egress, resulting from the
535: planet covering stellar areas that have been slightly differently limb-darkened.
536:
537: The process is reversed on egress. With the planet now moving more quickly than
538: the circular-orbit model, the model begins its egress early so that the real
539: planet will have caught up with it by mid-egress. Hence the real-minus-model
540: detectability is again initially negative, reaches zero by mid-egress, and then
541: trends positive until the end of egress for both the real and model planets.
542:
543: Despite the large $\Delta V$ for HD147506b, the magnitude of the transit
544: lightcurve asymmetry induced by orbital eccentricity is rather low (Figure
545: \ref{figure:residual}), peaking at only $3\times10^{-6}$ of the stellar flux.
546: This low detectability is unmeasurable given current capabilities, and will
547: probably prove challenging in the future as well given inherent stellar
548: variability. The large stellar radius of HD147506 increases $\Delta V$, but also
549: decreases $R_\mathrm{p}/R_*$, diminishing the asymmetry effect. Hence though
550: measurement of the asymmetry induced by an eccentric planet orbit can, along with
551: $V_f$, uniquely determine both $e$ and the longitude of periastron, the small
552: magnitude and duration of the effect are such that a measurement is unlikely to be
553: practical.
554:
555: For transiting terrestrial-sized planets, determination of orbital eccentricity is
556: even more difficult. If the ingress and egress are temporally resolved, that
557: duration along with the total transit duration together set the transit impact
558: parameter and $V_f$, providing a minimum $e$ constraint as per Equations
559: \ref{eq:eminperi} and \ref{eq:eminapo}. For objects with such small transit
560: depths as terrestrial planets, though, it will be difficult to temporally resolve
561: the planets' ingress and egress.
562:
563: Another way to constrain $b$ for terrestrial planets would be to use the effect of
564: stellar limb darkening. \citet{2003ApJ...585.1038S} showed that minimizing limb
565: darkening more precisely delineates the end of a planet's ingress and the
566: beginning of its egress. However strong limb darkening, if well-understood, can
567: provide a mechanism to ascertain a transit's impact parameter. Similar to the
568: analysis used by \citet{2007ApJ...655..564K} to study HD209458b, fixing limb
569: darkening coefficients based on theoretical or previousely determined values could
570: suffice to constrain $b$, which, with a previously measured $R_*$, would then
571: determine $V_f$ and allow for constraints on $e$.
572:
573:
574:
575:
576:
577:
578:
579:
580:
581: \section{CONCLUSIONS}
582:
583: For a given orbital semimajor axis, extrasolar planets on eccentric orbits are
584: more likely to transit than planets on circular orbits by a factor of
585: $(1-e^2)^{-1}$. As the space-based transit surveys \emph{CoRoT} and \emph{Kepler}
586: discover transiting planets that are far enough from their parent stars to have
587: avoided tidal circularization, more highly eccentric planets will be found
588: preferentially. This bias is straightforward to remove if the eccentricity and
589: longitude of periastron are known, as they could be given follow-up radial
590: velocity observations.
591:
592: The duration of a transit is a function of the planet's tangential velocity at
593: mid-transit, $V_{f_0}$. For eccentric planets $V_{f_0}$ is greatest at periastron
594: and smallest at apoastron. Hence if a planet transits near periastron the
595: duration is shorter than that of an equivalent planet in a circular orbit,
596: and similarly transits that occur when a planet is near apoastron is longer
597: than those of the equivalent circularly orbiting planet. It would be useful to be
598: able to use this effect to determine the orbital eccentricity of discovered
599: transiting extrasolar planets, either before or without radial velocity follow-up.
600:
601: If fitting the resulting lightcurve with a model planet on a circular orbit with
602: the known planetary period, a systematic error results in the determination of the
603: transit parameters if fitting for $R_*$, $R_\mathrm{p}$, $b$, and $c_1$ as done by
604: \citet{2001ApJ...552..699B} for HD209458b. If instead the model system uses an
605: assumed stellar radius measured by different means, then the transit velocity
606: $V_{f_0}$ can be measured. However without another, independent measurement of
607: either $e$ or the planet's longitude of periastron, knowledge of $V_{f_0}$ cannot
608: alone determine those parameters. It can set a lower limit on a planet's orbital
609: eccentricity.
610:
611: The difference in $V_f$ between a planet's ingress and egress that results from
612: the planet's orbital accelerations can resolve the $e$ / longitude of periastron
613: degeneracy. This velocity differential $\Delta V$ introduces an asymmetry into
614: the transit lightcurve: ingress is longer than egress before periastron, and
615: shorter after periastron. My model fits to synthetic eccentric planet transit
616: lightcurves show that the detectability of this asymmetry is small, of order
617: $3\times10^{-6}$ for recently discovered eccentric transiting planet HD147506b.
618: An effect that small is undetectable using present techniques. As HD147506b is
619: nearly a model candidate for which to observe this effect, it is unlikely that
620: transit lightcurve asymmetry will prove useful for determining the orbital
621: parameters of transiting planets using photometry alone.
622:
623: Determination of orbital eccentricity is of critical importance for evaluating the
624: habitability of terrestrial-sized transiting planets discovered by \emph{Kepler}.
625: As these planets have masses too low for radial velocity measurements to detect,
626: our only constraints on $e$ for these planets will be photometric. If the stellar
627: radius can be assumed from other, prior measurements, then it is possible to use
628: theoretical stellar limb-darkening coefficients within the \emph{Kepler} bandpass
629: to measure the transit impact parameter. This measurement would then constrain
630: $V_{f_0}$ and allow a lower limit to be placed on $e$.
631:
632: No techniques currently available will be able to uniquely measure the orbital
633: eccentricity of the terrestrial extrasolar planets that \emph{Kepler} will
634: discover. The lower limits on eccentricity described above will allow for a
635: statistical exploration of the eccentricity distibution of terrestrial planets.
636: That distribution will serve to constrain the formation and evolution of such
637: planets, as it has done for giant planets. However, whether or not any particular
638: \emph{Kepler} planet is truly habitable will remain unknown until its orbital
639: eccentricity can be measured.
640:
641:
642:
643:
644: \acknowledgements
645:
646: JWB acknowledges the support of the NASA Postdoctoral Program, administered for
647: NASA by Oak Ridge Associated Universities, and the support of NASA's \emph{Kepler}
648: mission for publication costs.
649:
650: \newpage
651:
652: \begin{thebibliography}{21}
653: \expandafter\ifx\csname natexlab\endcsname\relax\def\natexlab#1{#1}\fi
654:
655: \bibitem[{{Bakos} {et~al.}(2007){Bakos}, {Kovacs}, {Torres}, {Fischer},
656: {Latham}, {Noyes}, {Sasselov}, {Mazeh}, {Shporer}, {Butler}, {Stefanik},
657: {Fernandez}, {Sozzetti}, {Pal}, {Johnson}, {Marcy}, {Sipocz}, {Lazar},
658: {Papp}, \& {Sari}}]{2007arXiv0705.0126B}
659: {Bakos}, G.~A., {Kovacs}, G., {Torres}, G., {Fischer}, D.~A., {Latham}, D.~W.,
660: {Noyes}, R.~W., {Sasselov}, D.~D., {Mazeh}, T., {Shporer}, A., {Butler},
661: R.~P., {Stefanik}, R.~P., {Fernandez}, J.~M., {Sozzetti}, A., {Pal}, A.,
662: {Johnson}, J., {Marcy}, G.~W., {Sipocz}, B., {Lazar}, J., {Papp}, I., \&
663: {Sari}, P. 2007, ArXiv e-prints, 705
664:
665: \bibitem[{{Barnes} \& {Fortney}(2003)}]{oblateness.2003}
666: {Barnes}, J.~W. \& {Fortney}, J.~J. 2003, \apj, 588, 545
667:
668: \bibitem[{{Basri} {et~al.}(2005){Basri}, {Borucki}, \&
669: {Koch}}]{2005NewAR..49..478B}
670: {Basri}, G., {Borucki}, W.~J., \& {Koch}, D. 2005, New Astronomy Review, 49,
671: 478
672:
673: \bibitem[{{Bord{\'e}} {et~al.}(2003){Bord{\'e}}, {Rouan}, \&
674: {L{\'e}ger}}]{2003A&A...405.1137B}
675: {Bord{\'e}}, P., {Rouan}, D., \& {L{\'e}ger}, A. 2003, \aap, 405, 1137
676:
677: \bibitem[{{Borucki} \& {Summers}(1984)}]{1984Icar...58..121B}
678: {Borucki}, W.~J. \& {Summers}, A.~L. 1984, Icarus, 58, 121
679:
680: \bibitem[{{Brown} {et~al.}(2001){Brown}, {Charbonneau}, {Gilliland}, {Noyes},
681: \& {Burrows}}]{2001ApJ...552..699B}
682: {Brown}, T.~M., {Charbonneau}, D., {Gilliland}, R.~L., {Noyes}, R.~W., \&
683: {Burrows}, A. 2001, \apj, 552, 699
684:
685: \bibitem[{{Butler} {et~al.}(2006){Butler}, {Wright}, {Marcy}, {Fischer},
686: {Vogt}, {Tinney}, {Jones}, {Carter}, {Johnson}, {McCarthy}, \&
687: {Penny}}]{2006ApJ...646..505B}
688: {Butler}, R.~P., {Wright}, J.~T., {Marcy}, G.~W., {Fischer}, D.~A., {Vogt},
689: S.~S., {Tinney}, C.~G., {Jones}, H.~R.~A., {Carter}, B.~D., {Johnson}, J.~A.,
690: {McCarthy}, C., \& {Penny}, A.~J. 2006, \apj, 646, 505
691:
692: \bibitem[{{Demory} {et~al.}(2007){Demory}, {Gillon}, {Barman}, {Bonfils},
693: {Mayor}, {Mazeh}, {Pont}, {Queloz}, {Udry}, {Bouchy}, {Delfosse},
694: {Forveille}, {Mallmann}, {Pepe}, \& {Perrier}}]{2007arXiv0707.3809D}
695: {Demory}, B.~., {Gillon}, M., {Barman}, T., {Bonfils}, X., {Mayor}, M.,
696: {Mazeh}, T., {Pont}, F., {Queloz}, D., {Udry}, S., {Bouchy}, F., {Delfosse},
697: X., {Forveille}, T., {Mallmann}, F., {Pepe}, F., \& {Perrier}, C. 2007, ArXiv
698: e-prints, 707
699:
700: \bibitem[{{Jenkins}(2002)}]{2002ApJ...575..493J}
701: {Jenkins}, J.~M. 2002, \apj, 575, 493
702:
703: \bibitem[{{Knutson} {et~al.}(2007){Knutson}, {Charbonneau}, {Noyes}, {Brown},
704: \& {Gilliland}}]{2007ApJ...655..564K}
705: {Knutson}, H.~A., {Charbonneau}, D., {Noyes}, R.~W., {Brown}, T.~M., \&
706: {Gilliland}, R.~L. 2007, \apj, 655, 564
707:
708: \bibitem[{{Laughlin} {et~al.}(2005){Laughlin}, {Marcy}, {Vogt}, {Fischer}, \&
709: {Butler}}]{2005ApJ...629L.121L}
710: {Laughlin}, G., {Marcy}, G.~W., {Vogt}, S.~S., {Fischer}, D.~A., \& {Butler},
711: R.~P. 2005, \apjl, 629, L121
712:
713: \bibitem[{{Lissauer} {et~al.}(1997){Lissauer}, {Berman}, {Greenzweig}, \&
714: {Kary}}]{Lissauer.1997.RotationfromEccentricAccretion}
715: {Lissauer}, J.~J., {Berman}, A.~F., {Greenzweig}, Y., \& {Kary}, D.~M. 1997,
716: Icarus, 127, 65
717:
718: \bibitem[{{Loeb}(2005)}]{2005ApJ...623L..45L}
719: {Loeb}, A. 2005, \apjl, 623, L45
720:
721: \bibitem[{{Mandel} \& {Agol}(2002)}]{2002ApJ...580L.171M}
722: {Mandel}, K. \& {Agol}, E. 2002, \apjl, 580, L171
723:
724: \bibitem[{{Murray} \& Dermott(2000)}]{SolarSystemDynamics}
725: {Murray}, C.~D. \& Dermott, S.~F. 2000, {Solar System Dynamics} (New York:
726: Cambridge University Press)
727:
728: \bibitem[{{Naef} {et~al.}(2001){Naef}, {Latham}, {Mayor}, {Mazeh}, {Beuzit},
729: {Drukier}, {Perrier-Bellet}, {Queloz}, {Sivan}, {Torres}, {Udry}, \&
730: {Zucker}}]{2001A&A...375L..27N}
731: {Naef}, D., {Latham}, D.~W., {Mayor}, M., {Mazeh}, T., {Beuzit}, J.~L.,
732: {Drukier}, G.~A., {Perrier-Bellet}, C., {Queloz}, D., {Sivan}, J.~P.,
733: {Torres}, G., {Udry}, S., \& {Zucker}, S. 2001, \aap, 375, L27
734:
735: \bibitem[{{Nelson} \& {Davis}(1972)}]{1972ApJ...174..617N}
736: {Nelson}, B. \& {Davis}, W.~D. 1972, \apj, 174, 617
737:
738: \bibitem[{{Seager} \& {Hui}(2002)}]{Seager.oblateness}
739: {Seager}, S. \& {Hui}, L. 2002, \apj, 574, 1004
740:
741: \bibitem[{{Seager} \& {Mall{\'e}n-Ornelas}(2003)}]{2003ApJ...585.1038S}
742: {Seager}, S. \& {Mall{\'e}n-Ornelas}, G. 2003, \apj, 585, 1038
743:
744: \bibitem[{{Trilling}(2000)}]{2000ApJ...537L..61T}
745: {Trilling}, D.~E. 2000, \apjl, 537, L61
746:
747: \bibitem[{{Winn} {et~al.}(2007){Winn}, {Holman}, {Henry}, {Roussanova}, {Enya},
748: {Yoshii}, {Shporer}, {Mazeh}, {Johnson}, {Narita}, \&
749: {Suto}}]{2007AJ....133.1828W}
750: {Winn}, J.~N., {Holman}, M.~J., {Henry}, G.~W., {Roussanova}, A., {Enya}, K.,
751: {Yoshii}, Y., {Shporer}, A., {Mazeh}, T., {Johnson}, J.~A., {Narita}, N., \&
752: {Suto}, Y. 2007, \aj, 133, 1828
753:
754: \end{thebibliography}
755:
756:
757: \newpage
758:
759:
760:
761:
762:
763:
764:
765:
766: \end{document}
767: