1: \documentclass[12pt,preprint]{aastex}
2: \usepackage{graphicx}
3: \usepackage{amssymb}
4: \footnotesize
5: \newdimen\digitwidth %define! a one digit width for tables
6: \setbox0=\hbox{\rm0}
7: \digitwidth=\wd0
8: \catcode`!=\active
9: \def!{\kern\digitwidth}
10: \normalsize
11: \def\Msolar{{\rm M}_{\odot}}
12: \providecommand{\bm}[1]{\mbox{\boldmath$#1$\unboldmath}}
13:
14: \begin{document}
15: \title{An improved solar wind electron-density model for pulsar
16: timing}
17: \author{X. P. You\altaffilmark{1,2,3},
18: G. B. Hobbs\altaffilmark{2}, W. A. Coles\altaffilmark{4},
19: R. N. Manchester \altaffilmark{2}, J. L. Han\altaffilmark{1}}
20: \altaffiltext{1}{National Astronomical Observatories, Chinese Academy
21: of Sciences, Beijing 100012, China. Email:xpyou@ns.bao.ac.cn}
22: \altaffiltext{2}{Australia
23: Telescope National Facility, CSIRO, PO~Box~76, Epping, NSW~1710,
24: Australia}
25: \altaffiltext{3}{Present address: School of Physics, Southwest University,
26: Chongqing 400175, China}
27: \altaffiltext{4}{Electrical and Computer
28: Engineering, University of California at San Diego, La Jolla,
29: California 92093, USA}
30:
31:
32: \begin{abstract}
33: Variations in the solar wind density introduce variable delays into
34: pulsar timing observations. Current pulsar timing analysis programs
35: only implement simple models of the solar wind, which not only limit
36: the timing accuracy, but can also affect measurements of pulsar
37: rotational, astrometric and orbital parameters. We describe a new
38: model of the solar wind electron density content which uses
39: observations from the Wilcox Solar Observatory of the solar magnetic
40: field. We have implemented this model into the \textsc{tempo2} pulsar
41: timing package. We show that this model is more accurate than
42: previous models and that these corrections are necessary for high
43: precision pulsar timing applications.
44: \end{abstract}
45:
46: \keywords{pulsars: timing --- sun: solar wind}
47:
48: \section{Introduction}
49: It is now possible to make timing observations of millisecond pulsars
50: to a precision of $\sim$100\,ns. One of the most exciting
51: applications of such data-sets is to search for the signatures of
52: gravitational waves passing over the Earth. This is a major goal of the
53: Parkes Pulsar Timing Array (PPTA) project \citep{hob05,man06}, which
54: aims to observe 20 millisecond pulsars with a timing precision close
55: to 100\,ns over more than five years. Many phenomena can affect the
56: pulse arrival times at this level of timing precision. A major
57: contributor at our primary observing frequency of $\sim$1400\,MHz is
58: the interstellar and interplanetary medium \citep{yhc+07}. A small
59: change in a pulsar's dispersion measure (DM; the integrated electron
60: density along the line of sight to the pulsar) can cause significant
61: time delays in the pulse arrival times. For example, at an observing
62: frequency of 1400\,MHz, a time delay of 100\,ns is caused by a DM
63: variation of only $\sim 5\times10^{-5}$\,cm$^{-3}$\,pc. At this level,
64: the solar wind effect is significant when the line of sight to the
65: pulsar passes within $\sim$60$^\circ$ of the Sun.
66:
67: The standard pulsar timing programs
68: (\textsc{tempo1}\footnote{http://www.atnf.csiro.au/research/pulsar/tempo}
69: and \textsc{tempo2}; see Hobbs, Edwards \& Manchester
70: 2006\nocite{hem06}, Edwards, Hobbs \& Manchester 2006\nocite{ehm06})
71: calculate the solar contribution, ${\rm DM}_\odot$, from a spherically
72: symmetric model of the solar wind density which assumes a quadratic
73: decrease with solar distance and ignores temporal variation:
74: \begin{equation}\label{eqn:model}
75: {\rm DM}_\odot = 4.85\times10^{-6}\,n_0\,\frac{\theta}{\sin\theta}\;{\rm cm^{-3}pc},
76: \end{equation}
77: where $n_0$ is the electron density at $1$~AU from the Sun (in
78: cm$^{-3}$) and $\theta$ is the pulsar-Sun-observatory angle. By
79: default, \textsc{tempo2} chooses $n_0 = 4$~cm$^{-3}$ whereas
80: \textsc{tempo1} uses $n_0 =10$~cm$^{-3}$. However, the true electron
81: density of the solar wind can change with longitude, latitude and time
82: by a factor of at least four \citep{mbf+00}. \citet{yhc+07}
83: demonstrated that this simple model is inadequate for PSR~J1022+1001,
84: a pulsar that lies close to the ecliptic plane.
85:
86: There have been several previous analyses of the timing delays or DM
87: variations due to the solar wind that occur in pulsar timing
88: observations. For instance, the ecliptic latitude ($\beta$) of the
89: Crab pulsar is only $-1.29^{\circ}$. \citet{lps88} showed, using a few
90: observations within $5^{\circ}$ of the Sun, that the maximum time
91: delay due to the solar wind was about 500\,$\mu$s at 610
92: MHz. Similarly, \citet{pw91} showed that the DM changed by $\sim
93: 0.002\,{\rm pc\,cm}^{-3}$ when the line of sight to PSR~B0950+08
94: ($\beta = -4.62^\circ$) is close to the Sun. \citet{cbl+96} observed
95: PSR~B1821$-$24 between the years 1989 and 1993 and showed that between
96: December and January each year their timing residuals were
97: significantly affected by the solar corona. More recently,
98: \citet{sns+05} and \citet{lkn+06} analysed data of PSRs~J1713+0747 and
99: J0030+0451 using the \textsc{tempo1} model, but, instead of holding
100: the electron density at 1\,AU fixed, they fitted for this scaling
101: factor. They obtained that $n_0 = 5\pm4$\,cm$^{-3}$ and $n_0 =
102: 7\pm2$\,cm$^{-3}$ respectively.
103:
104: \citet{sfal97} argued that the planetary companions to PSR~B1257+12
105: \citep{wol94} were artefacts of incorrectly modelling the solar wind.
106: The closest planet to the pulsar produces a 25.3\,d periodicity in the
107: timing residuals which is remarkably close to periodicities seen in
108: Pioneer 10 spacecraft data which are thought to be due to patterns in
109: the solar wind caused by the Sun's rotation. Even though Wolszczan et
110: al. (2000)\nocite{whk+00} proved that the periodicity was due to
111: planetary companions (based on the use of the original \textsc{tempo1}
112: solar-wind model and multi-frequency observations) it is of interest
113: to understand the effect of an unmodelled (or poorly modelled) solar wind
114: on the measured pulsar parameters.
115:
116: Recently, \citet{ojs07} observed PSRs~J1801$-$2304, J1757$-$2421,
117: J1757$-$2223 and J1822$-$2256 when their lines of sight were close to
118: the Sun. Their work has some overlap with ours as they also used
119: observations from the Wilcox Solar Observatory and a model for the
120: solar electron density. However, they concentrated on variations in
121: pulsar rotation measures due to the solar magnetic field. In our work
122: we model variations in pulsar dispersion measures and describe their
123: implications for high precision pulsar timing.
124:
125: In this paper we first describe a two-state solar wind model
126: (\S\ref{sec:swm}) and our analysis technique (\S\ref{sec:data}) before
127: considering the implications for high precision pulsar timing
128: (\S\ref{sec:discussion}).
129:
130: \section{The two-state solar wind model}\label{sec:swm}
131: The solar wind is a complex system and important features are still
132: poorly understood. A summary of the relevent physics can be found in
133: \cite{sch06}. In brief, the solar wind can be thought of as having a
134: quasi-static component which is bimodal and co-rotates with the Sun,
135: and a transient component which has a time scale of hours to days. The
136: best known of the transient events are coronal mass ejections, which
137: typically cross any given line of sight about 5\% of the time
138: \citep{sch96b}. It is currently not feasible to model the complex
139: transient events and we will concentrate on modeling the co-rotating
140: wind structure, which has ``fast'' and ``slow'' components.
141:
142: The slow wind has a relatively high density and apparently originates
143: in or around active regions of closed magnetic geometry at low or
144: middle latitudes. The fast wind has lower density and originates in
145: regions with open magnetic field geometry called coronal holes. Large
146: coronal holes are located over the solar poles during the years of
147: minimum solar activity. Smaller and shorter-lived coronal holes occur
148: at middle and low latitudes when solar activity is higher.
149:
150: We note that original \textsc{tempo1} model can be thought of assuming
151: that the entire wind is a spherically symmetric slow wind, whereas the
152: default \textsc{tempo2} model assumes that the wind is entirely fast.
153:
154: The electron density in the fast wind can be estimated from
155: \emph{Ulysses} and \emph{SPARTAN} observations to give:
156: \begin{eqnarray}\label{eqn:nefast}
157: n_e = 1.155\times10^{11}R_\odot^{-2} + 32.3\times10^{11}R_\odot^{-4.39} + \\ \nonumber
158: 3254\times10^{11}R_\odot^{-16.25}\;\;{\rm m}^{-3}
159: \end{eqnarray}
160: at a distance of $R_\odot$ solar radii \citep{gf95a,gf98}.
161:
162: We can approximate the electron density in the ``slow wind'' using a
163: combination of the Muhleman \& Anderson (1981)\nocite{ma81} model fit
164: to their own observations far from the Sun and the ``Baumbach-Allen''
165: model near to the Sun (Allen 1947)\nocite{all47},
166: \begin{eqnarray}\label{eqn:neslow}
167: n_e = 2.99\times10^{14}R_\odot^{-16} + 1.5\times 10^{14}R_\odot^{-6}
168: + \\ \nonumber
169: 4.1\times 10^{11}\left(R_\odot^{-2} + 5.74R_\odot^{-2.7}\right)\;\;{\rm m}^{-3}.
170: \end{eqnarray}
171:
172: In order to determine DM$_\odot$ the electron density must be
173: integrated along the line of sight to the pulsar. Information on
174: whether a given position along the line of sight will be within the
175: slow or the fast wind can be obtained from the Wilcox Solar
176: Observatory\footnote{
177: \url{http://soi.stanford.edu/\~{}wso/forms/prsyn.html}. To obtain
178: data-sets suitable for \textsc{tempo2} the \textsc{ClassicSS} map
179: should be selected with a \textsc{latitude} projection. Full details
180: can be obtained from the \textsc{tempo2} on-line documentation
181: (http://www.atnf.csiro.au/research/pulsar/tempo2). } which provides
182: daily maps of the solar magnetic field since May 1976. Following
183: \citet{mbf+00}, we assume that the slow wind occupies the zone within
184: 20$^\circ$ of the magnetic neutral line and outside this is dominated
185: by the fast wind and that both winds flow radially. To demonstrate
186: our technique we show, in Figure~\ref{fg:synoptic}, a synoptic chart
187: showing the projection of the line of sight on to the Sun for
188: PSR~J1744$-$1134 on the 20th December 2004. As expected, this figure
189: shows that some parts of the line of sight lie within the slow wind
190: and some within the fast wind.
191:
192: \begin{figure}
193: \includegraphics[width=6cm,angle=-90]{f1.eps}
194: \caption{Projection on to the solar surface along wind streamlines of
195: the line of sight to PSR~J1744$-$1134 on 2004, December 20. The
196: triangle shows the point of closest approach to the Sun and the
197: open circle is the projected position of the Earth. Points are at
198: 5$^\circ$ intervals in angle subtended at the Sun. The solid line
199: indicates the position of the magnetic neutral line; the dashed
200: lines on either side are plotted 20$^\circ$ away from the neutron
201: line and delimit the region assumed to be dominated by the slow
202: wind.}
203: \label{fg:synoptic}
204: \end{figure}
205:
206: \section{Data analysis and method}\label{sec:data}
207: We use observations obtained for the Parkes Pulsar Timing Array (PPTA)
208: project \citep{man06} to test our new model. A sample of 20
209: millisecond pulsars has been observed since February 2004 at intervals
210: of 2-3 weeks at frequencies around 700\,MHz, 1400\,MHz and
211: 3100\,MHz. Details of the observations and the methods used to
212: determine the DM variations are given by \citet{yhc+07}.
213:
214: For this paper we use data for four pulsars which have measurable DM
215: variations due to the solar wind. PSRs~J1022+1001 and J1730$-$2304
216: have ecliptic latitudes of $-0.064^\circ$ and $0.19^\circ$
217: respectively and hence are eclipsed by the Sun each
218: year. PSRs~J1744$-$1134 and J1909$-$3744 have higher ecliptic
219: latitudes ($11^\circ$ and $-15^\circ$ respectively) but can be timed
220: with very high precision.
221:
222: We have implemented algorithms into \textsc{tempo2} to integrate the
223: electron density along a given line of sight assuming the fast- and
224: slow-wind electron densities as given in Equations~\ref{eqn:nefast}
225: and \ref{eqn:neslow}. For every observation, \textsc{tempo2}
226: calculates the projection of points along the line of sight to the
227: pulsar on to the surface of the Sun, assuming the Carrington rotation
228: rate and a mean wind velocity of 400\,km\,s$^{-1}$. These parameters
229: are characteristic of the slow wind and are chosen since this
230: component dominates both the wind dynamics and the dispersion
231: contribution. Using data from the Wilcox Solar Observatory,
232: \textsc{tempo2} determines the position of the magnetic neutral line
233: and, hence, the regions along the line of sight that are within the
234: slow and fast winds. A numerical integration is then carried out to
235: obtain the total electron column density of the solar wind along the
236: line of sight and hence ${\rm DM}_\odot$. The derived values are not
237: significantly dependent upon our assumptions about the wind rotation
238: and velocity.
239:
240: \section{Results and Discussion}\label{sec:discussion}
241: Tools are available within the \textsc{tempo2} software to obtain such
242: synopotic charts for any pulsar on any day since the start of the
243: Wilcox Solar Observatory data in 1976. Figure~\ref{fg:DMs} shows the
244: DM variations according to our solar wind model for PSR~J1744$-$1134
245: between the years 2004 and 2006. The predictions according to the
246: earlier \textsc{tempo1} and \textsc{tempo2} models are also indicated
247: in the Figure. We notice that the new model generally predicts DM
248: values that are higher than the \textsc{tempo2} model, but lower than
249: the \textsc{tempo1} model as expected from our two-state model. The
250: new model is also not smooth. Variations of up to DM$_\odot \sim
251: 10^{-4}$\,cm$^{-3}$pc occur on a daily basis.
252:
253: \begin{figure}
254: % \includegraphics[width=7cm,angle=-90]{J1744solar.ps}
255: \includegraphics[width=7cm,angle=-90]{f2.eps}
256: \caption{Solar-wind DM variations for PSR~J1744$-$1134 from 2004 to
257: 2006. The right-hand axis gives the corresponding time delay for an
258: observing frequency of 1400\,MHz. In the upper panel the solid line
259: gives the DM variations from our new model. The dashed and dot-dashed
260: lines indicate predictions of the original \textsc{tempo2} model and
261: the \textsc{tempo1} models respectively. In the lower panel we plot
262: the difference between the new model and the original \textsc{tempo2}
263: model.}
264: \label{fg:DMs}
265: \end{figure}
266:
267:
268: \begin{figure*}
269: \includegraphics[width=8cm]{f3a.eps}
270: \includegraphics[width=8cm]{f3b.eps}
271: \caption{Left column: comparison of the measured DM values with the
272: model predictions for PSRs~J1022+1002 (triangle symbols),
273: J1730$-$2304 (open stars), J1744$-$1134 (open squares) and
274: J1909$-$3744 (open circles). Right column: comparison of the
275: measured and predicted DMs for the Cognard et al. (1996)
276: observations of PSR~B1821$-$21. Triangle, star, square and circle
277: symbols represent data starting in December 1989, 1990, 1991 and
278: 1992, respectively. In both columns the upper panels gives the
279: measured DM values without any solar wind correction. The middle
280: panels give the difference between the actual values and those
281: predicted using the original \textsc{tempo2} model. The lowest
282: panels show the difference between the measurements and the
283: prediction using the improved solar wind model.}\label{fg:compare}
284: \end{figure*}
285:
286:
287:
288: Differences between measured DM$_\odot$ values \citep{yhc+07} and the
289: predictions using the original and new \textsc{tempo2} models are
290: shown in Figure~\ref{fg:compare} for the four pulsar datasets
291: discussed in this paper. This figure shows that, for lines of sight
292: that pass close to the Sun, the original \textsc{tempo2} does not
293: correctly predict DM$_\odot$. However, the improved model predicts
294: DM$_\odot$ within experimental uncertainties for all observations.
295:
296: Our data-sets are currently poorly sampled for lines of sight that
297: pass close to the Sun. We have, therefore, compared our improved
298: model predictions with observations of PSR~B1821$-$24 using the Nan{\c
299: c}ay radio telescope \citep{cbl+96}. DM values were measured from
300: their Figure~7. There appear to be significant non-solar variations in
301: their measured DMs and we have removed a straight line fitted to the
302: values more than $\sim 40 {\rm R}_\odot$ from the Sun. Comparisons
303: with the original \textsc{tempo2} model and our improved model are
304: shown in the right hand column of Figure~\ref{fg:compare}. It is clear
305: that the new model is a significant improvement over the previous
306: \textsc{tempo2} model. However, even with our improved model there are
307: some observations that are not consistent with our predictions. These
308: inconsistencies occur at the closest approach of the line of sight to
309: the Sun. At such close approaches our simple assumptions of the
310: two-component wind model where the slow wind lies within 20$^\circ$ of
311: the magnetic neutral line and radial wind flow with projection along
312: mean flow streamlines may not be valid. We will be able to further
313: test our model with future PPTA datasets having more precise and more
314: closely sampled DM measurements at close angular distances to the
315: Sun. Such results should help us to further improve the model.
316:
317: \subsection{Implications for high precision pulsar timing}
318: Modern pulsar timing experiments are aiming to achieve rms timing
319: precisions close to 100~ns over many years. At an observing frequency
320: of 1.4~GHz the solar wind causes time delays of this magnitude for
321: pulsars up to 60$^\circ$ from the Sun and significant deviations
322: between the original and improved \textsc{tempo2} models occur at
323: $\lesssim 20^\circ$ from the Sun.
324:
325: In order to study the effect of an unmodelled, or poorly modelled,
326: solar wind on pulsar timing parameters we used \textsc{tempo2} to
327: create simulated data-sets spanning three years for
328: PSR~J1744$-$1134. For these simulations we applied the improved
329: solar-wind model and a specified amount of uncorrelated pulsar timing
330: noise. We then either switched off all solar wind models or used the
331: original \textsc{tempo2} model before fitting for the pulsar's
332: parameters. Deviations from the true values for various astrometric
333: parameters are listed in Table~\ref{tb:diffParams}. Clearly, the
334: solar-wind model has a large effect on the values of the fitted
335: parameters. For instance, for 100\,ns rms timing residuals, the
336: derived values for parallax and declination when using the standard
337: \textsc{tempo2} model deviate by $\sim 2.5\sigma$ and $\sim
338: 11.0\sigma$ respectively from their true values. For any given
339: pulsar, the error in each parameter will depend upon the rms timing
340: residual, the data-span and the ecliptic latitude of the pulsar.
341:
342: \begin{table}\caption{Effect on timing parameters for different rms timing
343: residuals when comparing the new solar wind model to 1) no model and
344: 2) the original \textsc{tempo2} model.}
345: \label{tb:diffParams}
346: \begin{center}\begin{tabular}{clrr}\hline
347: Rms resid. & \multicolumn{1}{c}{Parameter} & No model & Orig.T2
348: model\\ ($\mu s$) & & \multicolumn{1}{c}{($\sigma$)}
349: &\multicolumn{1}{c}{($\sigma$)} \\ \hline 0.0 & Right ascension & 11.8
350: & 9.0 \\ & Declination & 52.1 & 22.2 \\ & Parallax & 26.6 & 4.1 \\ 0.1
351: & Right ascension & 5.9 & 2.7 \\ & Declination & 33.4 & 11.0 \\ &
352: Parallax & 17.6 & 2.5 \\ 1.0 & Right ascension & 1.3 & 1.7 \\ &
353: Declination & 3.8 & 0.8 \\ & Parallax & 2.7 & 0.7 \\ \\ \hline
354: \end{tabular}\end{center}
355: \end{table}
356:
357: \begin{figure}
358: \includegraphics[width=6cm,angle=-90]{f4a.eps}\\
359: \includegraphics[width=6cm,angle=-90]{f4b.eps}\\
360: \includegraphics[width=6cm,angle=-90]{f4c.eps}
361: \caption{The upper panel shows the power spectrum of timing
362: residuals contributed by the improved model of the solar wind for a
363: simulated three-year data-span for the PSR~B1257+12 system. In the
364: middle panel we have reduced the power in the annual term and its
365: harmonics by fitting the spherical wind model to the residuals and
366: subtracting it. The bottom panel shows the power spectrum of
367: residuals due to both the solar-wind model and the inner planet of
368: the PSR~B1257+12 system.}\label{fg:pspectrum}
369: \end{figure}
370:
371: In order to test whether the solar wind can mimic planetary companions
372: we have simulated a data-set for PSR~B1257$+$12 with the same span and
373: observing frequency as the Wolszczan (1994) observations. As we have
374: no access to the original Wolszczan (1994) data-set, our simulated
375: observations are uniformly sampled. A power spectrum of the solar wind
376: contribution was computed with a rectangular window using the
377: Lomb-Scargle algorithm. Since the data are uniformly sampled this is
378: the same as the normal Fourier power spectrum. This spectrum, shown in
379: the top panel of Figure~\ref{fg:pspectrum}, is dominated by harmonics
380: of the annual modulation. To reduce the leakage of annual harmonics
381: into the higher frequencies we removed the annual feature by
382: subtracting a fit of a spherically symmetric model
383: (Equation~\ref{eqn:model}) to the residuals. The resulting spectrum is
384: shown in the middle panel. There is no significant feature in either
385: spectrum corresponding to the narrow 25.3-d peak seen in solar wind
386: observations by Scherer et al. (1997). In fact we do not expect to see
387: a sharp feature in the spectrum because the line of sight to the
388: pulsar changes significantly during a solar rotation and the solar
389: wind density itself evolves on that time scale. We have used
390: \textsc{tempo2} to introduce the expected signal from the planet and
391: applied the same spectral analysis to the combined solar wind plus
392: planet simulation. The resulting spectrum is plotted in the bottom
393: panel. The contribution of the planet exceeds the solar wind noise by
394: a factor of more than 100. Clearly Scherer et al. (1997) seriously
395: overestimated the importance of solar noise in the detection of a
396: planet around PSR~B1257$+$12.
397:
398:
399: \section{Conclusions}
400: We have developed a new solar-wind model for pulsar timing experiments
401: and shown that it gives a more accurate correction for delays due to
402: the solar wind than earlier models. Use of the older solar-wind models
403: (or no correction) leads to systematic errors in measured pulsar
404: parameters. We have also shown that the solar wind cannot mimic the
405: signal from inner-most planetary companion of PSR B1257+12 as
406: suggested by Scherer et al. (1997). With the improved pulsar timing
407: data expected in the future from projects such as the Parkes Pulsar
408: Timing Array, use of the new model will make an important contribution
409: to achieving the goals of these projects. The improved model has been
410: implemented in the \textsc{tempo2} software package and we recommend
411: that it be used for all high-precision timing applications.
412:
413: \section*{Acknowledgments}
414:
415: X.P.Y. and J.L.H. are supported by the National Natural Science
416: Foundation of China (No.10473015 \& 10521001). W.A.C. was partially
417: supported by the National Science Foundation of the U.S.A. under grant
418: AST0507713. The Wilcox Solar Observatory data used in this study were
419: obtained via the web site {\url{http://soi.stanford.edu/\~{}wso}} at
420: \verb|2007:06:07_21:32:20 PDT| courtesy of J.T. Hoeksema. The Wilcox
421: Solar Observatory is supported by NASA. The data presented in this
422: paper were obtained as part of the Parkes Pulsar Timing Array project
423: that is a collaboration between the ATNF, Swinburne University and the
424: University of Texas, Brownsville, and we thank our collaborators on
425: this project. The Parkes radio telescope is part of the Australia
426: Telescope which is funded by the Commonwealth of Australia for
427: operation as a National Facility managed by CSIRO.
428:
429: %\bibliographystyle{apj}
430: %\bibliography{journals,modrefs,psrrefs,crossrefs}
431:
432: \begin{thebibliography}{21}
433: \expandafter\ifx\csname natexlab\endcsname\relax\def\natexlab#1{#1}\fi
434:
435: \bibitem[{{Allen}(1947)}]{all47}
436: {Allen}, C.~W. 1947, MNRAS, 107, 426
437:
438: \bibitem[{{Cognard} {et~al.}(1996){Cognard}, {Bourgois}, {Lestrade}, {Biraud},
439: {Aubry}, {Darchy}, \& {Drouhin}}]{cbl+96}
440: {Cognard}, I., {Bourgois}, G., {Lestrade}, J.-F., {Biraud}, F., {Aubry}, D.,
441: {Darchy}, B., \& {Drouhin}, J.-P. 1996, A\&A, 311, 179
442:
443: \bibitem[{{Edwards} {et~al.}(2006){Edwards}, {Hobbs}, \& {Manchester}}]{ehm06}
444: {Edwards}, R.~T., {Hobbs}, G.~B., \& {Manchester}, R.~N. 2006, MNRAS, 372, 1549
445:
446: \bibitem[{{Guhathakurta} \& {Fisher}(1998)}]{gf98}
447: {Guhathakurta}, M. \& {Fisher}, R. 1998, ApJ, 499, L215
448:
449: \bibitem[{{Guhathakurta} \& {Fisher}(1995)}]{gf95a}
450: {Guhathakurta}, M. \& {Fisher}, R.~R. 1995, Geophys. Res. Lett., 22, 1841
451:
452: \bibitem[{Hobbs(2005)}]{hob05}
453: Hobbs, G. 2005, PASA, 22, 179
454:
455: \bibitem[{{Hobbs} {et~al.}(2006){Hobbs}, {Edwards}, \& {Manchester}}]{hem06}
456: {Hobbs}, G.~B., {Edwards}, R.~T., \& {Manchester}, R.~N. 2006, MNRAS, 369, 655
457:
458: \bibitem[{{Lommen} {et~al.}(2006){Lommen}, {Kipphorn}, {Nice}, {Splaver},
459: {Stairs}, \& {Backer}}]{lkn+06}
460: {Lommen}, A.~N., {Kipphorn}, R.~A., {Nice}, D.~J., {Splaver}, E.~M., {Stairs},
461: I.~H., \& {Backer}, D.~C. 2006, ApJ, 642, 1012
462:
463: \bibitem[{Lyne {et~al.}(1988)Lyne, Pritchard, \& Smith}]{lps88}
464: Lyne, A.~G., Pritchard, R.~S., \& Smith, F.~G. 1988, MNRAS, 233, 667
465:
466: \bibitem[{{Manchester}(2006)}]{man06}
467: {Manchester}, R.~N. 2006, Chin. J. Astron. Astrophys., Suppl. 2, 6, 139
468:
469: \bibitem[{{McComas} {et~al.}(2000){McComas}, {Barraclough}, {Funsten},
470: {Gosling}, {Santiago-Mu{\~n}oz}, {Skoug}, {Goldstein}, {Neugebauer}, {Riley},
471: \& {Balogh}}]{mbf+00}
472: {McComas}, D.~J., {Barraclough}, B.~L., {Funsten}, {Gosling}, J.~T.,
473: {Santiago-Mu{\~n}oz}, E., {Skoug}, R.~M., {Goldstein}, B.~E., {Neugebauer},
474: M., {Riley}, P., \& {Balogh}, A. 2000, J. Geophys. Res., 105, 10419
475:
476: \bibitem[{{Muhleman} \& {Anderson}(1981)}]{ma81}
477: {Muhleman}, D.~O. \& {Anderson}, J.~D. 1981, ApJ, 247, 1093
478:
479: \bibitem[{{Ord} {et~al.}(2007){Ord}, {Johnston}, \& {Sarkissian}}]{ojs07}
480: {Ord}, S.~M., {Johnston}, S., \& {Sarkissian}, J. 2007, ArXiv e-prints, 705
481:
482: \bibitem[{Phillips \& Wolszczan(1991)}]{pw91}
483: Phillips, J.~A. \& Wolszczan, A. 1991, ApJ, 382, L27
484:
485: \bibitem[{{Scherer} {et~al.}(1997){Scherer}, {Fichtner}, {Anderson}, \&
486: {Lau}}]{sfal97}
487: {Scherer}, K., {Fichtner}, H., {Anderson}, J.~D., \& {Lau}, E.~L. 1997,
488: Science, 278, 1919
489:
490: \bibitem[{{Schwenn}(1996)}]{sch96b}
491: {Schwenn}, R. 1996, Ap\&SS, 243, 187
492:
493: \bibitem[{{Schwenn}(2006)}]{sch06}
494: ---. 2006, Living Reviews in Solar Physics, 3, 2
495:
496: \bibitem[{Splaver {et~al.}(2005)Splaver, Nice, Stairs, Lommen, \&
497: Backer}]{sns+05}
498: Splaver, E.~M., Nice, D.~J., Stairs, I.~H., Lommen, A.~N., \& Backer, D.~C.
499: 2005, ApJ, 620, 405
500:
501: \bibitem[{Wolszczan(1994)}]{wol94}
502: Wolszczan, A. 1994, Science, 264, 538
503:
504: \bibitem[{{Wolszczan} {et~al.}(2000){Wolszczan}, {Hoffman}, {Konacki},
505: {Anderson}, \& {Xilouris}}]{whk+00}
506: {Wolszczan}, A., {Hoffman}, I.~M., {Konacki}, M., {Anderson}, S.~B., \&
507: {Xilouris}, K.~M. 2000, ApJ, 540, L41
508:
509: \bibitem[{{You} {et~al.}(2007){You}, {Hobbs}, {Coles}, {Manchester}, {Edwards},
510: {Bailes}, {Sarkissian}, {Verbiest}, {van Straten}, {Hotan}, {Ord}, {Jenet},
511: {Bhat}, \& {Teoh}}]{yhc+07}
512: {You}, X.~P., {Hobbs}, G., {Coles}, W.~A., {Manchester}, R.~N., {Edwards}, R.,
513: {Bailes}, M., {Sarkissian}, J., {Verbiest}, J.~P.~W., {van Straten}, W.,
514: {Hotan}, A., {Ord}, S., {Jenet}, F., {Bhat}, N.~D.~R., \& {Teoh}, A. 2007,
515: MNRAS, 378, 493
516:
517: \end{thebibliography}
518:
519:
520: \end{document}
521:
522: