1: %\documentclass[12pt,preprint]{aastex}
2: \documentclass[12pt,preprint]{emulateapj}
3: \usepackage{graphicx}
4:
5: \shorttitle{Constraining Orbital Parameters}
6: \shortauthors{Stephen R. Kane \& Kaspar von Braun}
7: \slugcomment{Submitted for publication in the Astrophysical Journal}
8: %\received{2007 November 09}
9:
10: \begin{document}
11:
12: \title{Constraining Orbital Parameters Through Planetary Transit
13: Monitoring}
14: \author{Stephen R. Kane, Kaspar von Braun}
15: \affil{NASA Exoplanet Science Institute, Caltech, MS 100-22, 770
16: South Wilson Avenue Pasadena, CA 91125, USA}
17: \email{skane@ipac.caltech.edu}
18:
19: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
20:
21: \begin{abstract}
22:
23: The orbital parameters of extra-solar planets have a significant
24: impact on the probability that the planet will transit the host
25: star. This was recently demonstrated by the transit detection of HD
26: 17156b whose favourable eccentricity and argument of periastron
27: dramatically increased its transit likelihood. We present a study
28: which provides a quantitative analysis of how these two orbital
29: parameters affect the geometric transit probability as a function of
30: period. Further, we apply these results to known radial velocity
31: planets and show that there are unexpectedly high transit
32: probabilities for planets at relatively long periods. For a
33: photometric monitoring campaign which aims to determine if the planet
34: indeed transits, we calculate the expected transiting planet yield and
35: the significance of a potential null result, as well as the subsequent
36: constraints that may be applied to orbital parameters.
37:
38: \end{abstract}
39:
40: \keywords{planetary systems -- techniques: photometric}
41:
42: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
43:
44: \section{Introduction}\label{introduction}
45:
46: With the number of known extra-solar planets exceeding 300,
47: statistical interpretations of the distribution of orbital parameters
48: are becoming increasingly significant. These parameter distributions
49: help us unlock the mysteries surrounding the planet formation process
50: to which many challanges have been presented, not the least of which
51: contains the mechanisms that drive planetary migration
52: \citep{arm07}. \citet*{for08b} showed that transit light curves in
53: particular can be used to characterize orbital eccentricities and
54: hence give further insight into the global eccentricity distribution.
55:
56: In terms of the sheer number of transit light curves, the major
57: contributors have been the shallow wide-field surveys such as the
58: Transatlantic Exoplanet Survey (TrES) \citep{man07}, the XO project
59: \citep{joh08}, the Hungarian Automated Telescope Network (HATNet)
60: \citep{pal08}, and SuperWASP \citep{and08}. In addition, there have
61: been at least five cases in which planetary transits were detected
62: through photometric follow-up of planets already known via their
63: radial velocity (RV) discoveries. These five planets are HD 209458b
64: \citep{cha00,hen00}, HD 149026b \citep{sat05}, HD 189733b
65: \citep{bou05}, GJ 436b \citep{gil07}, and HD 17156b
66: \citep{bar07a}. The case of HD 17156b is of particular interest since
67: it is a 21.2 day period planet which happens to have a large
68: eccentricity ($e = 0.67$) and an argument of periastron which places
69: the periapsis of its orbit in the direction toward the observer and
70: close to parallel to the line of sight, resulting in an increased
71: transit probability.
72:
73: Conversely, the dominant sources of RV planet discoveries have been
74: the California \& Carnegie Planet Search \citep{mar97} and the High
75: Accuracy Radial velocity Planet Searcher (HARPS) \citep{pep04}
76: teams. However, in the near future we can expect to see larger-scale
77: surveys \citep*{kan07b} and new instruments \citep{li08} which will
78: increase both the number and diversity of known planets. There have
79: been suggestions regarding the strategy for photometric follow-up of
80: these radial velocity planets at predicted transit times
81: \citep{kan07a} and the instruments that could be used for such surveys
82: \citep{lop06a}. Some attempts have been made to detect these possible
83: transits \citep{lop06b,sha06} which have thus far been unsuccessful.
84:
85: This paper discusses the effect of orbital parameters on the geometric
86: transit probability of planets. We calculate orbital constraints that
87: may be applied, particularly in the absence of transit signatures in
88: photometric follow-up observations. Section 2 describes how the
89: eccentricity and argument of periastron of known planetary orbits
90: affect transit probability. It further presents applications of this
91: effect to known RV planets and discusses how uncertainties in the
92: orbital parameter values affect the reliability of the ephemeris
93: calculations. In Section 3, we show how orbital constraints can be
94: applied in the absence of a photometrically detected transit signal,
95: and we discuss the potential transit yield and statistical
96: significance of a scenario in which no transits are found in a large
97: sample of RV planets. We summarize and conclude in Section 4.
98:
99: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
100:
101: \section{Transit Probability}\label{transit_probability}
102:
103: Recent work by \citet{bar07b} and \citet{bur08a} showed that higher
104: eccentricities of planetary orbits will increase their transit
105: probabilities and, consequently, expected yield for transit
106: surveys. In this Section, we demonstrate the combined effect of the
107: eccentricity and argument of periastron on transit probability. For
108: explanations of the orbital parameters, including the argument of
109: periastron $\omega$, we refer the reader to \citet{kan07a} and
110: \citet{bar07b}. We first explicitly derive the dependence of transit
111: probability $P_t$ as a function of eccentricity $e$, argument of
112: periastron $\omega$, and orbital semi-major axis $a$ (i.e.,
113: period). We discuss this dependence of $P_t$ specifically with respect
114: to $\omega$ and period, apply the results to a sample of 203
115: exoplanets compiled in \citet{but06}, and briefly discuss how
116: ephemeris calculations (and thus planning of photometric follow-up
117: observations) are affected by uncertainties in $e$ and $\omega$.
118:
119: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
120:
121: \subsection{Orbital Configuration}\label{orbital_configuration}
122:
123: For a circular orbit the geometric transit probability is proportional
124: to the inverse of the semi-major axis, $a$, such that the inclination
125: of the planet's orbital plane $i$ must satisfy
126: \begin{equation}
127: a \cos i \leq R_p + R_\star
128: \end{equation}
129: where $R_p$ and $R_\star$ are the radii of the planet and star
130: respectively \citep{bor84}. For an eccentric orbit, the transit
131: probability, $P_t$, may be expressed as
132: \begin{equation}
133: P_t = \frac{R_p + R_\star}{a (1 - e \cos E)}
134: \label{transit_prob1}
135: \end{equation}
136: where $e$ is the eccentricity of the orbit and $E$ is the eccentric
137: anomaly. The eccentric anomaly and the true anomaly, $f$, are related
138: to each other by
139: \begin{equation}
140: cos E = \frac{e + \cos f}{1 + e \cos f}
141: \label{cose}
142: \end{equation}
143: where the true anomaly is defined as the angle between the direction
144: of periapsis and the current position of the planet in the orbit.
145: Equation \ref{transit_prob1} can then be evaluated at each point in
146: the planetary orbit. The transit probability can also be described in
147: terms of the geometry of an ellipse. For an elliptical orbit, the
148: separation of the planet and star is
149: \begin{equation}
150: r = \frac{a (1 - e^2)}{1 + e \cos f}.
151: \label{separation}
152: \end{equation}
153: As shown by \citet{kan07a}, the place in a planetary orbit where it is
154: possible for a transit to occur (where the planet passes the star-observer
155: plane that is perpendicular to the plane of the planetary orbit) is when
156: $\omega + f = \pi / 2$. The transit probability may then be re-expressed as
157: \begin{equation}
158: P_t = \frac{(R_p + R_\star)(1 + e \cos (\pi/2 - \omega))}{a (1 - e^2)}
159: \label{transit_prob2}
160: \end{equation}
161: consistent with the findings of \citet{bar07b}. Equations
162: \ref{transit_prob1} and \ref{transit_prob2} both yield the same result
163: based upon the orbital configuration, but Equation \ref{transit_prob2}
164: clearly shows the major role played by the the values of $e$ and
165: $\omega$ in determining the likelihood of a planet transiting the
166: parent star.
167:
168: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
169:
170: \subsection{Argument of Periastron Dependence}\label{aop_dependence}
171:
172: \begin{figure}
173: \includegraphics[angle=270,width=8.2cm]{f1.eps}
174: \caption{Dependence of geometric transit probability on the argument
175: of periastron, $\omega$, for eccentricities of 0.0, 0.3, and 0.6,
176: plotted for periods of 4.0 days (left ordinate) and 50.0 days
177: (right ordinate). Stellar and planetary radii are assumed to be a
178: Jupiter and solar radius, respectively. For details, see \S
179: \ref{orbital_configuration} and, in particular, Equation
180: \ref{transit_prob2}.}
181: \label{fig1}
182: \end{figure}
183:
184: Equation \ref{transit_prob2} states the dependence of transit
185: probability on the argument of periastron. As we rotate the semi-major
186: axis of the orbit around the star we can observe how the transit
187: probability varies. This dependence is shown in Figure \ref{fig1} for
188: eccentricities of 0.3 (dashed line) and 0.6 (dotted line) in
189: comparison with the constant transit probability for a circular orbit
190: (solid line). Since the shape of this variation is independent of
191: period, $P$, the y-axes are scaled for both a 4.0 day and 50.0 day
192: period orbits. Figure \ref{fig1} assumes a Jupiter radius and a solar
193: radius for the values of $R_p$ and $R_\star$ respectively. Note that
194: $P_t$ scales linearly with the sum of these values (Equation
195: \ref{transit_prob2}).
196:
197: The peak transit probability occurs at $\omega = \pi / 2$, and the
198: corresponding increase in $P_t$ as compared to a circular orbit can be
199: significant: a factor of 1.5 for $e = 0.3$ and a factor of 2.5 for $e
200: = 0.6$. Moreover, the fraction of the orbital path which produces a
201: higher value of $P_t$ than the circular orbit with the same period
202: (corresponding to the fraction of range in $\omega$ for which the
203: dotted or dashed line is above the solid line in Figure \ref{fig1})
204: increases with increasing eccentricity.
205:
206: The fraction of orbital orientations with $e \neq 0$ producing {\it
207: lower} transit probabilities than the corresponding circular orbits
208: is made clear in Figure \ref{fig2} in which a view from above the
209: orbit pole of two planetary orbits is depicted. The range of $\omega$
210: in Figure \ref{fig1} that produces lower values of $P_t$ than a
211: circular orbit corresponds to the angle between the intersection
212: points shown in Figure \ref{fig2} for which the planet is located
213: outside the circular orbit. For an eccentricity of 0.6 this angle is
214: $\theta = 105\degr$ and decreases with increasing
215: eccentricity. However, the Keplerian nature of the orbit is such that,
216: although the larger fraction of the orbital path is spent close to the
217: star, the larger fraction of time is spent farther away from the star
218: \citep{bar07b}. This is a crucial aspect in designing a photometric
219: follow-up campaign to monitor RV planets in eccentric orbits for
220: possible transits.
221:
222: \begin{figure}
223: \includegraphics[width=8.2cm]{f2.eps}
224: \caption{A view from above the orbit pole of a circular (solid line)
225: and eccentric ($e = 0.6$; dotted line) planetary orbit for $\omega
226: = 3 \pi / 2$. The angle $\theta$ corresponds to the range of
227: orbital orientations for which an elliptical orbit has a lower
228: transit probability than a circular orbit with the same period
229: (see \S \ref{orbital_configuration} and Figure
230: \ref{fig1}).}
231: \label{fig2}
232: \end{figure}
233:
234: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
235:
236: \subsection{Period Dependence}\label{period_dependence}
237:
238: As demonstrated in Figure \ref{fig1}, the peak transit probability
239: increases with eccentricity. Although the shape of $P_t = f(\omega)$
240: is independent of period, the magnitude of $P_t$ changes as a function
241: of period (Figure \ref{fig1}). Consequently, the fractional increase
242: in $P_t$ for eccentric orbits can be substantial, as shown in \S
243: \ref{aop_dependence} and argued by \citet{bar07b},
244:
245: The current distribution of eccentricities for the known extra-solar
246: planets indicates that orbits within 0.1~AU tend to be forced into
247: nearly circular orbits through tidal circularization, whereas longer
248: period orbits can possess a great range of eccentricities
249: \citep{for08a}. Indeed most of the planets beyond 0.1~AU have
250: eccentricities in excess of 0.3. Thus, it is the longer-period planets
251: whose transit probabilities are more likely to be affected by
252: eccentricities than the short-period ones.
253:
254: In Figure \ref{fig3} we show mean transit probability as a function of
255: period after averaging over $0 \leq \omega \leq 2 \pi$, for the period
256: range $1 \leq P \leq 50$ days. Eccentricities of 0.0, 0.3, and 0.6 are
257: shown with solid, dashed, and dotted lines, respectively. As expected,
258: we see that doubling the eccentricity from 0.3 to 0.6 creates a
259: significant increase in the mean transit probability. Most affected
260: are the longer period planets whose eccentric orbits can raise their
261: likelihood of transit from a negligible value to a statistically
262: viable number for photometric follow-up.
263:
264: \begin{figure}
265: \includegraphics[angle=270,width=8.2cm]{f3.eps}
266: \caption{The mean transit probability on a logarithmic scale,
267: averaged over all values of $\omega$ (cf. Figure \ref{fig1}), as a
268: function of period, for eccentricities of 0.0, 0.3, and
269: 0.6.}
270: \label{fig3}
271: \end{figure}
272:
273: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
274:
275: \subsection{Application to Known Exoplanets}\label{application}
276:
277: If we assume circular orbits for each of the known exoplanets, the
278: transit probability at intermediate to long period orbits makes
279: photometric searches for planets in those regimes
280: impractical. However, applying the orbital parameters of $e$ and $\omega$
281: should in general lead to an overall more favourable
282: situation for transit detection. Depending on the brightness of the
283: host star and the cadence of the RV observations, a reasonable
284: estimate of these two parameters is normally extracted from the RV
285: fitting.
286:
287: Figure \ref{fig4} shows the transit probability calculated from
288: orbital parameters provided by \citet{but06} for planets with
289: estimates of $e$ and $\omega$ (203 planets in total). The transit
290: probabilities are plotted against period, but are calculated from the
291: semi-major axis, $a$, using Equation \ref{transit_prob2}. For the
292: purposes of providing an approximate comparison of the relative
293: transit probabilities, we assume a Jupiter and Solar radius for the
294: values of $R_p$ and $R_\star$, respectively. Hence, we can include the
295: transit probability for a circular orbit, shown in the figure as a
296: solid line. In addition, the sub-panel in the plot shows the
297: difference in $P_t$ between the actual orbit and a hypothetical
298: circular one of the same period (residuals). The mean value of the
299: residuals for all 203 planets is positive but relatively small ($4.13
300: \times 10^{-5}$), and is dominated by the low transit probability of
301: the long period planets. The mean residual of planets with $P < 100.0$
302: days, however, yields an overall increase of $\sim 0.5$\% in $P_t$.
303:
304: \begin{figure}
305: \includegraphics[angle=270,width=8.2cm]{f4.eps}
306: \caption{The geometric transit probability for a circular orbit
307: (solid curve) along with the transit probability for 203 RV
308: planets from \citet{but06} calculated from their orbital
309: parameters (open circles). HD 17156b is indicated by a 5-pointed
310: star. The sub-panel plots the difference in $P_t$ between the the
311: actual orbit and a hypothetical circular orbit for each of the
312: planets.}
313: \label{fig4}
314: \end{figure}
315:
316: HD~17156b, a transiting planet with 21.2 day period \citep{bar07a}, is
317: shown as a 5-pointed star. Its transit probability is greatly
318: increased by its orbital parameters. Note that the actual $P_t$ of
319: HD~17156b is larger than the 5\% shown in Figure \ref{fig4} since the
320: radius of the host star is 1.47~$R_{\sun}$. At longer periods, the
321: planets with the largest residuals are HD~156846b, HD~4113b, and
322: HD~20782b, which have periods of 359.51, 526.62, and 585.86 days,
323: respectively. The probability residuals for these three planets are
324: 0.024, 0.032, and 0.025 respectively, the effect of which is to raise
325: their transit probabilities to the same level as HD~17156b if it were
326: in a circular orbit. It is worth noting that these three planets all
327: have eccentricities close to 0.9 which is undoubtedly the primary
328: cause of the increased transit probability.
329:
330: The increased transit probabilities of eccentric planets motivate
331: photometric follow-up programs of RV planets. Compared to transit
332: surveys, such programs require much less telescope time since the time
333: of transit is, in principle, known. However, it was shown by
334: \citet{kan07a} that reliable constraints on $e$ and $\omega$ are
335: needed to avoid significant offsets in predicted transit times. This
336: is particularly true of long period planets. In the case of planets
337: HD 156846b, HD 4113b, and HD 20782b, the uncertainties cited in
338: \citet{but06} indicate that the values of $e$ are all constrained to
339: $\pm 0.03$ and the values of $\omega$ are constrained to $\pm
340: 3.0$\degr (compared to the mean and median values for all 203 planets
341: of $\delta \omega = 20\degr$ and 10\degr, respectively). Thus,
342: ephemerides for these planets can relatively reliably be determined
343: from RV fit parameters alone. The transit duration is on the order of
344: 12 hours for these planets, ensuring that one will practically never
345: observe both the ingress and the egress of the transit during a single
346: orbit from the ground. However, the large transit duration and
347: relative low uncertainties in $e$ and $\omega$ will increase the
348: chances of observing at least a partial transit during the predicted
349: observing window.
350:
351: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
352:
353: \section{Constraining Orbital Parameters}
354:
355: In \S 2, we discussed transit probability as a function of various
356: system parameters as well as aspects of potential photometry follow-up
357: campaigns. Here we focus on what can be learned from the presence and
358: absence of a planetary transit in follow-up observations.
359:
360: For a transiting planet the physical properties (such as the mass, radius, and
361: density) can be calculated, leading to determination of (as opposed to
362: constraints on) system parameters of the planet. Furthermore, the orbital
363: inclination can be compared with the plane of stellar rotation \citep{win07}
364: and used to test planetary models regarding co-planar orbits.
365:
366: However, even the absence of a planetary transit signature in photometric data
367: can lead to interesting constraints on the orbital parameters. Below, we
368: elaborate on these constraints and apply the results to the aforementioned
369: \citet{but06} sample of RV planets.
370:
371: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
372:
373: \subsection{Orbital Radius versus Stellar Radius}
374: \label{radii}
375:
376: One implicit assumption in the derivation of the transit probability by
377: \citet{bor84} is that the planet remains well outside the star in order to
378: produce the solid angle of the planet's shadow \citep[see also][]{bar07b}. As
379: a result, the calculation of $P_t$ in Equation \ref{transit_prob2} becomes
380: invalid for extreme orbits with a small semi-major axis and a high value of
381: eccentricity (see \S \ref{orbital_configuration}).
382:
383: To quantify this assumption, we use Equation \ref{separation} to
384: calculate the maximum eccentricity, $e_{\mathrm{max}}$, allowed as a
385: function of the planet-star separation in units of $a/R$ where $R
386: \equiv R_p + R_\star$. Applying the constraint $r > R$ when $f = 0$
387: (i.e., the planet is outside the star at periapsis) to Equation
388: \ref{separation} results in
389: \begin{equation}
390: R = \frac{a (1 - e_{\mathrm{max}}^2)}{1 + e_{\mathrm{max}}}
391: \end{equation}
392: and thus,
393: \begin{equation}
394: e_{\mathrm{max}} = 1 - \frac{R}{a}.
395: \label{emax}
396: \end{equation}
397: Equation \ref{emax} is plotted in Figure \ref{fig5} for values of
398: $a/R$ ranging from 1 to 30. Also shown are dot-dashed lines which
399: indicate the $a/R$ values for OGLE-TR-56b \citep{kon03}, XO-5b
400: \citep{bur08b}, and HD 17156b.
401:
402: The restrictions upon the maximum eccentricity begin to become
403: significant for $a/R < 10$, which encompasses most of the known
404: transiting exoplanets. This restriction is purely based upon orbital
405: dynamics and there are undoubtedly additional limitations on the
406: eccentricity due to tidal effects in this region.
407:
408: \begin{figure}
409: \includegraphics[angle=270,width=8.2cm]{f5.eps}
410: \caption{The maximum orbital eccentricity, $e_{\mathrm{max}}$,
411: plotted as a function of the planet-star separation in units of
412: $a/R$ (see Equation \ref{emax}) in order for a planet to remain
413: outside the surface of its parent star. This requirement is purely
414: based on system geometry and does not take into account tidal
415: effects or planet-planet interactions, but requires that $a > R$
416: when $f = 0$. Dot-dashed lines indicate values for $a/R$ for
417: various transiting planets.}
418: \label{fig5}
419: \end{figure}
420:
421: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
422:
423: \subsection{Orbital Inclination and Argument of Periastron}
424: \label{i_and_w}
425:
426: One of the primary advantages of observing an exoplanet transiting the
427: host star is that it eliminates the ambiguity in the planetary mass
428: created by the unknown orbital inclination angle, $i$. The precise value of
429: the inclination can be derived from the impact parameter of the
430: transit across the stellar disk, defined by
431: \begin{equation}
432: b \equiv \frac{a \cos i}{R_\star}
433: \end{equation}
434: and measurable from the shape of the lightcurve and the planet-star radius
435: ratio \citep{sea03}. Due to the constraint placed on $i$ by the presence of
436: transits, the true planetary mass will be within a few percent of the value
437: as determined from RV measurements alone.
438:
439: The data available for transiting planets from the Extra-solar Planets
440: Encyclopaedia\footnote{http://exoplanet.eu/} and from \citet*{tor08}
441: show that the current distribution of inclination angles extends from
442: 90\degr to almost 78\degr. The transiting planets whose orbits feature
443: numerically lower values of $i$ (i.e., more ``face-on'') are dominated
444: by the very hot Jupiters, such as OGLE-TR-56b which has an inclination
445: of $78.8\degr \pm 0.5\degr$ \citep{pon07}.
446:
447: \begin{figure}
448: \includegraphics[angle=270,width=8.2cm]{f6.eps}
449: \caption{The maximum orbital inclination for a non-transiting planet
450: as a function of the argument of periastron, $\omega$, for
451: eccentricities of 0.0 (solid line), 0.3 (dashed line), and 0.6
452: (dotted line), plotted for periods of both 4.0 days and 50.0
453: days.}
454: \label{fig6}
455: \end{figure}
456:
457: If, however, a planet is determined not to transit, then limits may be
458: placed upon the orbital inclination if the eccentricity and argument
459: of periastron are known from RV measurements. This results from
460: re-expressing Equation \ref{transit_prob1} as follows:
461: \begin{equation}
462: \cos i = \frac{R_p + R_\star}{a (1 - e \cos E)}.
463: \label{cos_i}
464: \end{equation}
465: Figure \ref{fig6} shows the maximum inclination for various values of
466: period, $e$, and $\omega$. These are calculated by holding period and
467: $e$ fixed whilst varying $\omega$ using Equations \ref{cose} and
468: \ref{cos_i}. We further assume a Jupiter radius and a solar radius for
469: the values of $R_p$ and $R_\star$ respectively. For non-transiting
470: planets on orbits with $e \neq 0$ whose periastron is aligned towards
471: the observer (i.e., $\omega \sim \pi/2$), Figure \ref{fig6} shows that
472: the constraint on the inclination can be as high as $i \leq 75\degr$,
473: depending upon the orbital period. This is particularly useful for
474: those planets whose mass estimate places them close to the brown dwarf
475: regime. Note that Equation \ref{cos_i} reduces to Equation \ref{emax}
476: when $i = 90\degr$ and $\omega = \pi/2$, consistent with the
477: requirement that the planet remain outside the star.
478:
479: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
480:
481: \subsection{Orbital Inclination and Eccentricity}
482: \label{i_and_e}
483:
484: As we show in \S \ref{i_and_w}, the fact that a planet is found to not
485: transit limits the possible combinations of $e$, $\omega$, and $i$. We
486: now consider what constraints may be placed upon the orbital
487: inclination for a non-transiting RV planet as a function of
488: eccentricity for the specific examples of when the periapsis is
489: aligned towards ($\omega \sim \pi/2$) and away from ($\omega \sim
490: 3\pi/2$) the observer.
491:
492: Figure \ref{fig7} illustrates the range of orbital inclinations that
493: are excluded for two orbits (shown edge-on) of non-transiting planets.
494: Both of these orbits have the same semi-major axes but different
495: eccentricities and are aligned such that $\omega = 3\pi/2$ (i.e.,
496: periapsis occurs behind the star as seen from the observer). In this
497: case, the range of possible values of $i$ increases with decreasing
498: orbital eccentricity ($\phi_1 > \phi_2$). The opposite is true when
499: $\omega = \pi/2$. In fact, the inclination in that case is only
500: constrained by the requirement that the planet remain outside the star
501: during periapsis (Equation \ref{emax}).
502:
503: \begin{figure}
504: \includegraphics[width=8.2cm]{f7.eps}
505: \caption{An edge-on view of two planetary orbits with the same
506: values of semi-major axis, showing the range of excluded
507: inclinations for an orbit with low eccentricity (dotted lines;
508: $\phi_1$) and high eccentricity (dashed lines; $\phi_2$) for which
509: the planet does not transit the parent star. For $\omega =
510: 3\pi/2$, the range of possible inclinations decreases with
511: increasing eccentricity.} \label{fig7}
512: \end{figure}
513:
514: Figure \ref{fig8} graphically demonstrates these constraints by plotting
515: Equation \ref{cos_i}, except now we fix the period and $\omega$ and
516: vary $e$. A Jupiter radius and a solar radius are assumed for the
517: values of $R_p$ and $R_\star$ respectively. The lines in these plots
518: represent the maximum values for $i$ for a non-transiting planet as a
519: function of $e$. These calculations are performed for four different
520: periods and the two aforementioned orientations of $\omega$: $\omega =
521: \pi/2$; left panel; case (a), and $\omega = 3\pi/2$; right panel; case
522: (b). It is worth noting that case (a) in Figure \ref{fig8} represents
523: the physical constraint described in Figure \ref{fig5}, since Equation
524: \ref{cos_i} reduces to Equation \ref{emax} for $\omega = \pi/2$, as
525: stated above.
526:
527: \begin{figure*}
528: \begin{center}
529: \begin{tabular}{cc}
530: \includegraphics[angle=270,width=8.2cm]{f8a.eps} &
531: \includegraphics[angle=270,width=8.2cm]{f8b.eps} \\
532: \end{tabular}
533: \end{center}
534: \caption{Maximum orbital inclination as a function of $e$ for
535: non-transiting planets, plotted for four different periods (see
536: Equation \ref{cos_i}). The left panel, case (a), is for $\omega =
537: \pi/2$ and is based on the requirement that the planet remain
538: outside the star (\S \ref{radii}). The right panel, case (b),
539: shows the situation for $\omega = 3\pi/2$, and is based on the
540: geometrical arguments outlined in \S \ref{i_and_e} and shown in
541: Figure \ref{fig7}. The location of the maximum eccentricities of
542: the known transiting planets TrES-3 and TrES-4, given their
543: parameters in Table \ref{tab1} and assuming $\omega = 3\pi/2$, are
544: indicated by 5-pointed stars.} \label{fig8}
545: \end{figure*}
546:
547: In case (a), for example, a non-transiting planet in a 4-day orbit
548: with $e=0.4$ has a range of possible inclination angles of $i \leq
549: 80^{\degr}$. For larger values of $e$ (at any period), the periastron
550: distance of the planet will become so small that almost all values of
551: $i$ are possible, the maximum value of $e$ at each period being
552: defined by Equation \ref{emax}. The dependence of $i$ upon $e$ is
553: weaker for case (b) (note different scale for the left panel in Figure
554: \ref{fig8}) since the periastron passage now happens behind the star
555: (Figure \ref{fig2}) as seen from the observer, and thus, the range of
556: possible $i$-values is not very constrained by $e$.
557:
558: We now consider a planet discovered using the transit method with
559: known $i$, $a$, $R_p$, and $R_\star$, but unknown values for $e$ and
560: $\omega$. Is it possible to constrain $e$ in this scenario? Using
561: Equation \ref{cos_i}, the eccentricity can be expressed as follows
562: \begin{equation}
563: e = \frac{1}{cos E} \left( 1 - \frac{R_p + R_\star}{a \cos i}
564: \right).
565: \label{eccentricity}
566: \end{equation}
567: However, there exists a degeneracy between $e$ and $\omega = f(E)$
568: such that one cannot place constraints on one parameter without
569: knowledge of the other. Additionally, as shown in \S \ref{i_and_w},
570: the constraint upon $i$ is only limited by the orbital boundary
571: defined by Equation \ref{emax} when $\omega = \pi/2$.
572:
573: Therefore, a meaningful constraint may only be placed upon $e$ for
574: values of $\omega$ for which the orbital inclination is greater than
575: the maximum predicted for a circular orbit (i.e., the region above the
576: solid line shown in Figure \ref{fig6}). For case (b) ($\omega =
577: 3\pi/2$), Equation \ref{eccentricity} reduces to
578: \begin{equation}
579: e = \frac{R_p + R_\star}{a \cos i} - 1.
580: \end{equation}
581: As an example, consider the two known transiting planets TrES-3 and
582: TrES-4. The fit parameters shown in Table \ref{tab1} for the values of
583: $R_\star$, $R_p$, $i$, and $a$ are those reported by the discovery
584: papers for TrES-3 \citep{odo07} and TrES-4 \citep{man07}. Also shown
585: in Table \ref{tab1} are the maximum eccentricities for both case (a)
586: and case (b). For case (a), the maximum eccentricity is $\sim 0.8$ for
587: both planets. For case (b), the maximum eccentricity for these two
588: planets is 0.4--0.5 and are plotted in the right panel of Figure
589: \ref{fig8}. In each case, the maximum eccentricities are remarkably
590: similar because the longer period of TrES-4 is compensated by the
591: relatively large radii of the star and planet.
592:
593: \begin{table}
594: \begin{center}
595: \caption{Fit parameters for TrES-3 and TrES-4 from \citet{odo07} and
596: \citet{man07} respectively, along with the calculated maximum
597: eccentricities.}
598: \label{tab1}
599: \begin{tabular}{@{}ccc}
600: \hline
601: Parameter & TrES-3 & TrES-4 \\
602: \hline
603: $R_\star$ & $0.802 \pm 0.046$ & $1.738 \pm 0.092$ \\
604: $R_p$ & $1.295 \pm 0.081$ & $1.674 \pm 0.094$ \\
605: $i$ & $82.15 \pm 0.21$ & $82.81 \pm 0.33$ \\
606: $a$ & $0.0226 \pm 0.0013$ & $0.0488 \pm 0.0022$ \\
607: $e(\omega=\pi/2)$ & $0.808 \pm 0.083$ & $0.818 \pm 0.073$ \\
608: $e(\omega=3\pi/2)$ & $0.404 \pm 0.041$ & $0.451 \pm 0.040$ \\
609: \hline
610: \end{tabular}
611: \end{center}
612: \end{table}
613:
614: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
615:
616: \subsection{Global Statistics}
617:
618: The total number of transiting planets discovered thus far via radial
619: velocity surveys does not necessarily reflect the true number of
620: transiting planets in this sample. At the time of writing, most of the
621: known radial velocity planets have not been adequately monitored
622: photometrically in order to rule out transits. We can estimate the
623: number of planets that should be transiting, and determine the
624: significance of a hypothetical null result from a photometric
625: follow-up campaign by applying the results of this paper to the
626: \citet{but06} RV planet sample.
627:
628: The host star properties and planetary orbital parameters provided by
629: \citet{but06} form the foundation of a Monte-Carlo simulation of the
630: transit probabilities calculated from Equation
631: \ref{transit_prob2}. The planetary radii $R_p$ are assumed to be one
632: Jupiter radius as used in previous sections. However, the stellar
633: radii $R_\star$ are estimated individually from the values of $B - V$
634: provided by \citet{but06}, assuming the host stars are dwarf stars
635: \citep{cox00}. The orbital elements $a$, $e$ and $\omega$ are directly
636: extracted from \citet{but06}. Using these values, we calculate $P_t$
637: for each of the 203 stars in the sample and randomly determine if the
638: planet transits. This yields an integer number of projected transits
639: from the sample. By performing these calculations $\sim 100000$ times,
640: we produce a probability distribution for the number of transiting
641: planets expected from this sample, shown in Figure \ref{fig9}.
642:
643: \begin{figure}
644: \includegraphics[angle=270,width=8.2cm]{f9.eps}
645: \caption{The probability distribution (solid line) for the 203
646: planets in the \citet{but06} sample, predicting the number of
647: transiting planets based on their estimated orbital
648: parameters. Over-plotted is a gaussian distribution (dashed line)
649: using the mean and standard deviation of the simulation
650: results.}
651: \label{fig9}
652: \end{figure}
653:
654: The simulated probability distribution has a mean value of $\sim 4.5$
655: transits peaking at $P_t \sim 0.2$ with a standard deviation of $\sim
656: 2.0$. For comparison, we also generated a gaussian distribution
657: profile using this mean and standard deviation. The a priori
658: probability that none of the planets in this sample transit their host
659: star is $\sim 1$\%. In fact, three of the planets in this sample are
660: known to transit, specifically HD 17156b, GJ 436b, and HD
661: 147506b. Hence the current number of transiting planets from this
662: sample is almost 1$\sigma$ below the expectation.
663:
664: We further note that the sample of RV planets is biased toward
665: numerically higher values of $i$ since detection efficiency will
666: increase with higher $i$ for a given RV precision. As such, the
667: expected number of transiting planets in the sample should be regarded
668: as a lower limit. Though the discrepancy between known and expected
669: transiting planets is not significant in this low-number regime, it is
670: nevertheless quantifiable, and we conclude that further transit
671: discoveries in this sample are possible or even likely. Any such
672: additional detections would, in turn, lead to further understanding of
673: the respective observational biases of the RV and transit methods. For
674: example, the observational bias leads to an observed difference
675: between the period distributions of planets discovered by the transit
676: method and the radial velocity method, as discussed in detail by
677: \citet*{gau05}.
678:
679: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
680:
681: \section{Conclusions}
682:
683: It is still uncertain at this stage how many of the known radial
684: velocity planets transit their parent stars. What is clear is that the
685: eccentricity distribution of the known exoplanets will increase the
686: transit likelihood, making detections for long-period planets, such as
687: HD 17156b, feasible. We have shown in this paper that there is enough
688: potential amongst longer period planets for transit detections to
689: motivate a photometric monitoring campaign at the predicted times of
690: transit for these targets. \citet{fle08} have shown that long-period
691: transiting planets may yet be discovered through ground-based transit
692: surveys, particularly if data sets from different surveys are
693: combined.
694:
695: As pointed out by \citet{bar07b}, eccentric planets that have a
696: periastron oriented away from the observer are far more likely to
697: exhibit a secondary than a primary eclipse. The detection of such a
698: secondary eclipse is considerably more challenging than for a primary
699: eclipse since it relies on a minimum level of planetary flux and is
700: best pursued at infrared wavelengths. The discussion in \S
701: \ref{i_and_e} shows that even an assumption of $\omega = 3\pi/2$ can
702: place constraints on the orbital inclination. A prime candidate for
703: such a study is HD 80606b \citep{nae01} which has a period of 111.87
704: days and an eccentricity of 0.927. Scaling Figure \ref{fig1} to this
705: period and eccentricity yields a secondary transit probability of
706: $\sim 15$\%.
707:
708: Many of the results presented in this paper can easily be applied to
709: any system since the results generally scale linearly with the sum of
710: the stellar and planetary radii. Through applying these results to
711: current and future radial velocity planet discoveries, one can choose
712: targets for an efficient observing campaign which may help to discover
713: long-period transiting planets and hence add invaluable information to
714: planetary structure and formation theories.
715:
716: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
717:
718: \section*{Acknowledgements}
719:
720: The authors would like to thank David Ciardi, Scott Fleming, and Alan
721: Payne for several useful discussions. We would especially like to
722: thank the referee Jason W. Barnes, who provided a fast and insightful
723: report which greatly improved the quality of the paper.
724:
725: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
726:
727: \begin{thebibliography}{}
728:
729: \bibitem[\protect\citeauthoryear{Anderson et al.}{2008}]{and08}
730: Anderson, D.R., et al., 2008, MNRAS, 387, L4
731: \bibitem[\protect\citeauthoryear{Armitage}{2007}]{arm07} Armitage,
732: P.J., 2007, ApJ, 665, 1381
733: \bibitem[\protect\citeauthoryear{Barbieri et al.}{2007}]{bar07a}
734: Barbieri, M., et al., 2007, A\&A, 476, L13
735: \bibitem[\protect\citeauthoryear{Barnes}{2007}]{bar07b} Barnes, J.W.,
736: 2007, PASP, 119, 986
737: \bibitem[\protect\citeauthoryear{Bouchy et al.}{2005}]{bou05} Bouchy,
738: F., et al., 2005, A\&A, 444, L15
739: \bibitem[\protect\citeauthoryear{Borucki \& Summers}{1984}]{bor84}
740: Borucki, W.J., Summers, A.L., 1984, Icarus, 58, 121
741: \bibitem[\protect\citeauthoryear{Burke}{2008}]{bur08a} Burke, C.J.,
742: 2008, ApJ, 679, 1566
743: \bibitem[\protect\citeauthoryear{Burke}{2008}]{bur08b} Burke, C.J.,
744: et al., 2008, ApJ, submitted (arXiv:0805.2399)
745: \bibitem[\protect\citeauthoryear{Butler et al.}{2006}]{but06} Butler,
746: R.P., et al., 2006, ApJ, 646, 505
747: \bibitem[\protect\citeauthoryear{Charbonneau et al.}{2000}]{cha00}
748: Charbonneau, D., Brown, T.M., Latham, D.W., Mayor, M., 2000, ApJ,
749: 529, L45
750: \bibitem[\protect\citeauthoryear{Cox}{2000}]{cox00} Cox, A.N., 2000,
751: Allen’s Astrophysical Quantities (4th ed.; New York: AIP)
752: \bibitem[\protect\citeauthoryear{Fleming et al.}{2008}]{fle08}
753: Fleming, S.W., Kane, S.R., McCullough, P.R., Chromey, F.R., 2008,
754: MNRAS, 386, 1503
755: \bibitem[\protect\citeauthoryear{Ford \& Rasio}{2008}]{for08a} Ford,
756: E.B., Rasio, F.A., 2008, ApJ, in press (astro-ph/0703163)
757: \bibitem[\protect\citeauthoryear{Ford, Quinn, \& Veras}{Ford et
758: al.}{2008}]{for08b} Ford, E.B., Quinn, S.N., Veras, D., 2008,
759: ApJ, 678, 1407
760: \bibitem[\protect\citeauthoryear{Gaudi, Seager, \&
761: Mallen-Ornelas}{Gaudi et al.}{2005}]{gau05} Gaudi, B.S., Seager,
762: S., Mallen-Ornelas, G., 2005, ApJ, 623, 472
763: \bibitem[\protect\citeauthoryear{Gillon et al.}{2007}]{gil07} Gillon,
764: M., et al., 2007, A\&A, 472, L13
765: \bibitem[\protect\citeauthoryear{Henry et al.}{2000}]{hen00} Henry,
766: G.W., Marcy, G.W., Butler, R.P., Vogt, S.S., 2000, ApJ, 529, L41
767: \bibitem[\protect\citeauthoryear{Johns-Krull et al.}{2008}]{joh08}
768: Johns-Krull, C.M., et al., 2008, ApJ, 677, 657
769: \bibitem[\protect\citeauthoryear{Kane}{2007}]{kan07a} Kane, S.R.,
770: 2007, MNRAS, 380, 1488
771: \bibitem[\protect\citeauthoryear{Kane, Schneider, \& Ge}{Kane et
772: al.}{2007}]{kan07b} Kane, S.R., Schneider, D.P., Ge, J., 2007,
773: MNRAS, 377, 1610
774: \bibitem[\protect\citeauthoryear{Konacki et al.}{2003}]{kon03}
775: Konacki, M., Torres, G., Jha, S., Sasselov, D.D., 2003, Nature, 421,
776: 507
777: \bibitem[\protect\citeauthoryear{Li et al.}{2008}]{li08} Li,
778: C.-H., et al., 2008, Nature, 452, 610
779: \bibitem[\protect\citeauthoryear{L\'opez-Morales}{2006}]{lop06a}
780: L\'opez-Morales, M., 2006, PASP, 118, 716
781: \bibitem[\protect\citeauthoryear{L\'opez-Morales et al.}{2006}]{lop06b}
782: L\'opez-Morales, M., Morrell, N.I., Butler, R.P., Seager, S., 2006,
783: PASP, 118, 1506
784: \bibitem[\protect\citeauthoryear{Mandushev et al.}{2007}]{man07}
785: Mandushev, G., et al., 2007, ApJ, 667, L195
786: \bibitem[\protect\citeauthoryear{Marcy et al.}{1997}]{mar97} Marcy,
787: G.W., Butler, R.P., Williams, E., Bildsten, L., Graham, J.R., Ghez,
788: A.M., Jernigan, G., 1997, ApJ, 481, 926
789: \bibitem[\protect\citeauthoryear{Naef et al.}{2001}]{nae01} Naef, D.,
790: et al., 2001, A\&A, 375, L27
791: \bibitem[\protect\citeauthoryear{O'Donovan et al.}{2007}]{odo07}
792: O'Donovan, F.T., et al., 2007, ApJ, 663, L37
793: \bibitem[\protect\citeauthoryear{P\'al et al.}{2008}]{pal08}
794: P\'al, A., et al., 2008, ApJ, in press (arXiv:0803.0746)
795: \bibitem[\protect\citeauthoryear{Pepe et al.}{2004}]{pep04} Pepe, F.,
796: et al., 2004, A\&A, 423, 385
797: \bibitem[\protect\citeauthoryear{Pont et al.}{2007}]{pon07}
798: Pont, F., et al., 2007, A\&A, 465, 1069
799: \bibitem[\protect\citeauthoryear{Sato et al.}{2005}]{sat05} Sato, B.,
800: et al., 2005, ApJ, 633, 465
801: \bibitem[\protect\citeauthoryear{Seager \&
802: Mall\'en-Ornelas}{2003}]{sea03} Seager, S., Mall\'en-Ornelas, G.,
803: 2003, ApJ, 585, 1038
804: \bibitem[\protect\citeauthoryear{Shankland et al.}{2006}]{sha06}
805: Shankland, P.D., et al., 2006, ApJ, 653, 700
806: \bibitem[\protect\citeauthoryear{Torres, Winn, \& Holman}{Torres et
807: al.}{2008}]{tor08} Torres, G., Winn, J.N., Holman, M.J., 2008,
808: ApJ, 677, 1324
809: \bibitem[\protect\citeauthoryear{Winn et al.}{2007}]{win07} Winn,
810: J.N., et al., 2007, ApJ, 665, L167
811: \end{thebibliography}
812:
813: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
814:
815: \end{document}
816: