1: \documentclass[12pt,preprint]{aastex}
2: %\documentclass[manuscript]{aastex}
3: % \documentclass[preprint2]{aastex}
4:
5: \shorttitle{Meridional flux transport speed in the photosphere}
6: \shortauthors{\v{S}vanda et al.}
7:
8: \begin{document}
9:
10: \title{Speed of Meridional Flows and Magnetic Flux Transport on the Sun}
11:
12: %% Use \author, \affil, and the \and command to format
13: %% author and affiliation information.
14: %% Note that \email has replaced the old \authoremail command
15: %% from AASTeX v4.0. You can use \email to mark an email address
16: %% anywhere in the paper, not just in the front matter.
17: %% As in the title, use \\ to force line breaks.
18:
19: \author{Michal \v{S}vanda, \altaffilmark{1}$^{,}$\altaffilmark{2},
20: Alexander G. Kosovichev\altaffilmark{3}, and Junwei Zhao\altaffilmark{3}}
21:
22: \altaffiltext{1}{Astronomical Institute (v.v.i.), Academy of Sciences, Ond\v{r}ejov Observatory, CZ-25165, Czech Republic}
23: \altaffiltext{2}{Astronomical Institute, Charles University, Prague, CZ-18200, Czech Republic}
24: \altaffiltext{3}{W. W. Hansen Experimental Physics Laboratory, Stanford University, Stanford, CA~94305-4085, USA}
25: \altaffiltext{*}{Emails: michal@astronomie.cz, sasha@sun.stanford.edu, junwei@sun.stanford.edu}
26:
27: \begin{abstract}
28: We use the magnetic butterfly diagram to determine the speed of the magnetic flux
29: transport on the solar surface towards the poles. The manifestation of the flux transport is clearly visible as
30: elongated structures extended from the sunspot belt to the polar
31: regions. The slopes of these structures are measured and
32: interpreted as meridional magnetic flux transport speed. Comparison with the
33: time-distance helioseismology measurements of the mean speed of the meridional flows at the depth of 3.5--12~Mm shows a
34: generally good agreement, but the speeds of the flux transport and the meridional flow
35: are significantly different in areas occupied by the magnetic field.
36: The local circulation flows around active regions, especially the strong equatorward flows
37: on the equatorial side of active regions affect the mean velocity profile derived
38: by helioseismology, but do not influence the magnetic flux transport.
39: The results show that the mean longitudinally averaged meridional flow measurements by helioseismology
40: may not be used directly in solar dynamo models for describing the magnetic flux transport, and that
41: it is necessary to take into account the longitudinal structure of these flows.
42:
43: \end{abstract}
44:
45: \keywords{Sun: atmospheric motions -- Sun: magnetic fields -- Sun: activity}
46:
47:
48: \section{Introduction}
49:
50: The largest-scale velocity fields on the Sun consist of differential
51: rotation and meridional circulation. The differential rotation is
52: defined as an integral of the zonal (East--West) component of the velocity field
53: depending on the solar latitude, $b$, and radius. The meridional flow
54: is calculated as an integral of the North--South component of
55: the velocity field, generally depending again on the latitude and radius.
56: Both the differential rotation and meridional circulation are the key
57: ingredients of the solar dynamo. The differential rotation plays an
58: important role in generating and strengthening of toroidal magnetic
59: field inside the Sun, while the meridional flow
60: transports the magnetic flux towards the solar poles resulting in
61: cyclic polar field reversals \citep[for a recent review,
62: see][]{Brandenburg2005}.
63:
64: The meridional flux transport seems to be an essential agent
65: influencing the length, strength and other properties of solar
66: magnetic cycles. Generally, the slower the meridional flows are, the
67: longer the next magnetic cycle is expected. Dynamo models showed
68: that the turn-around time of the meridional cell is between 17 and 21
69: years, and that the global dynamo may have some kind of memory lasting
70: longer than one cycle \citep{2006GeoRL..3305102D}.
71:
72: The speed of the meridional flow and its variation with the solar cycle measured by local helioseismology
73: in the subsurface layers of the Sun were used
74: as an input in the recent flux-transport models \citep{2006ApJ...649..498D}. In local helioseismology measurements
75: (e.g. \citeauthor{2004ApJ...603..776Z}
76: \citeyear{2004ApJ...603..776Z}, \citeauthor{2006ApJ...638..576G} \citeyear{2006ApJ...638..576G}), the meridional flow was
77: derived from a general subsurface flow field by averaging the
78: North-South component of the plasma velocity over longitude for a Carrington
79: rotation period. The studies revealed that the mean meridional flow
80: varied with the solar activity cycle. These variations may
81: significantly affect solar-cycle predictions based on the solar dynamo models, which assume that the
82: magnetic flux is transported with the mean meridional flow speed
83: \citep{2006ApJ...649..498D}.
84:
85: Our goal is to verify this assumption and to investigate the
86: relationship between the subsurface meridional flows and the flux
87: transport. In this study, we show that the mean meridional flows
88: derived from the time-distance helioseismology subsurface flow maps
89: are affected
90: by strong local flows around active regions in the activity belts.
91: These local flows have much less significant effect on the magnetic flux
92: transport.
93:
94: \section{Method of measurements}
95:
96: The magnetic field data were obtained from Kitt Peak synoptic maps of longitudinal
97: magnetic field. The magnetic butterfly diagram is continuously
98: constructed from synoptic magnetic maps measured at National Solar
99: Observatory by averaging the magnetic flux in longitude at each
100: latitude for each solar rotation since 1976 \citep{hat03}.
101:
102: At mid-latitudes of the magnetic butterfly diagram (Fig.~\ref{maps}a), between the active region zone and polar regions,
103: we clearly see elongated structures corresponding to the
104: poleward magnetic flux transport. The aim of our method is to measure the slopes of these structures
105: and to derive the speed of the meridional magnetic flux transport.
106:
107: In addition to the large-scale structures, the original diagram
108: contains small-scale relatively short-lived local
109: magnetic field structures, which appear as a `noise' in the diagram.
110: To improve the signal-to-noise ratio for
111: the magnetic flux structures we applied a frequency band-pass filter for the
112: frequencies between $1.06\times 10^{-8}$~s$^{-1}$ (period of
113: 1093~days) and $3.17\times 10^{-7}$~s$^{-1}$ (period of 36.5 days).
114: The filtering procedure is performed separately for each individual
115: latitudinal cut on the diagram. We tried also other methods of enhancement of structures
116: and found that they all provided comparable results. The difference between the original butterfly diagram and the filtered
117: one can be seen in Fig.~\ref{maps} (\emph{a} and \emph{b}). The flux-transport elongated structures
118: are more easily visible after the filtering, and therefore more
119: suitable for analysis.
120:
121: The meridional flux transport speed is measured on the basis of
122: cross-correlation of two rows, which are assumed to be similar in shape,
123: but positions of structures are different due to
124: their meridional transport. We cross-correlate pairs of rows separated
125: by heliographic latitude $\Delta b$ in a sliding
126: window with the size of 55 Carrington rotations. The edges of
127: the correlation window are apodized by a smooth function to avoid
128: the boxcar effects. The extremal position is calculated as a maximum
129: of the parabolic fit of the set of correlation coefficients of
130: correlated windows in five discrete displacements. If the
131: distribution of the correlation coefficients does not have a maximum,
132: or if the $R^2$ is too low (under 0.8), the meridional velocity in this
133: pixel is not evaluated. We have chosen $\Delta b=5^\circ$ as the
134: best tradeoff between the spatial resolution and precision.
135:
136: To make the procedure more robust, we average the calculated
137: meridional velocity for five consecutive frames separated by
138: 0.5\degr{} and centered at $\Delta b$ from the studied row. If any
139: of the speeds in averaged five rows is far out of the expected range
140: ($-$60 to $+$60 m\,s$^{-1}$), then it is not used in the averaging.
141: From the fit, the accuracy of the measured flow speed is evaluated
142: and the maximum value of the set of five independent measurements at
143: different rows is taken. The same procedure is done with the processed map rotated by
144: 180\,$^\circ$ to avoid any possible preferences in the direction
145: determination, and both results are averaged. The measured errors
146: were taken as the maximum value of both independent measurements.
147:
148: For the ongoing analysis, only the speeds that were measured with
149: the error lower than 3~m\,s$^{-1}$ were taken into account. This criterion and some
150: failures of the slope measurement introduce gaps in the data, which
151: we need to fill. For this purpose we need to determine the best
152: continuous differentiable field that approximates the data. The
153: determination of such a field can be done in various ways, but we
154: wish to avoid possible artifacts. For filling the gaps we used
155: the MultiResolution Analysis. It is based on wavelet analysis, and we have chosen the
156: Daubechies wavelet due to its compact support. This property is
157: important since it minimizes edge effects. Moreover, using
158: these wavelets also preserves the location of zero-crossing and
159: maxima of the signal during the analysis. Daubechies are claimed to
160: be very stable in the noisy environments. For details see
161: \cite{rieutord07}.
162:
163: The reconstructed meridional flux transport speed map can be seen in
164: Fig.~\ref{maps}\emph{c}. The flux transport remains poleward during the whole
165: studied period.
166:
167: For the comparison between the meridional flow obtained from
168: time-distance helioseismology \citep{2004ApJ...603..776Z} and the
169: magnetic flux transport from our method, we have calculated the
170: averaged values of both quantities in bins of 10 heliographic degrees.
171: For 1996--2006, only eleven Carrington rotations have been evaluated
172: (one per year) by time-distance helioseismology using the full-disc Dynamics
173: data from the MDI instrument on SoHO spacecraft. These data are available
174: only for approximately two months per year. We compared the measurements of the magnetic flux
175: transport and the meridional flows in those
176: particular non-consecutive Carrington rotations. The plots are
177: displayed in Fig.~\ref{rotations}.
178:
179: \section{Results}
180:
181: The magnetic flux transport speed and the mean meridional flow speed obtained from helioseismology
182: are very similar (correlation coefficients are in
183: a range of 0.7--0.9). We have to keep in mind that while the time-distance meridional circulation profiles
184: represent the behaviour of the plasma during particular Carrington rotations, the magnetic butterfly diagram tracking
185: profiles represent the flux transport smoothed over 10 Carrington rotations. Therefore the agreement
186: cannot be perfect in principle. To make our results more accurate, the continuous helioseismic data are needed.
187:
188: The speeds of the meridional flux transport in the near-equatorial region are less reliable, since the elongated structures in the magnetic butterfly diagram extend from the activity belts towards poles. In the equatorial region, significant parts of the measurements were excluded from the analysis due to their large measured error. The gaps were filled using MultiResolution Analysis from well-measured points. Although the results seem reasonable here, their lower reliability has to be kept in mind. The original data are constructed from the images obtained with low resolution with the orthographical projection to disc. Therefore values above the latitude of 50\,$^\circ$ are impacted by the projection effect, which reduces the spatial resolution and may cause an apparent increase of the measured meridional flux transport speed. This effect should be reduced if the higher-resolution data would be used.
189:
190: During the minimum of solar activity (such as CR 1911, 1923, or 2032), the profile of the meridional flux transport speed is very consistent with the mean longitudinally averaged profile of the meridional flow from helioseismology. The best agreement is found for depth of 9--12~Mm. This suggests that the flux transport may be influenced by flows in the deeper layers.
191:
192: With increasing magnetic activity in the photosphere of the Sun, the gradient of the mean meridional circulation profile derived from time-distance helioseismology becomes steeper. This is consistent with the results obtained
193: by numerical simulations by \cite{2004SoPh..220..333B}. The simulations show that with increasing magnetic activity, the
194: Maxwell stresses oppose the Reynolds stresses, causing an acceleration of the meridional circulation and deceleration of
195: the rotation in low latitudes. Our measurements show that the variations of the slope of the mean meridional flux transport speed in latitude are lower with the progression of the solar cycle.
196:
197: When the large-active regions emerge in the activity belt, the flow towards equator is formed on the equatorial side of the magnetic regions (see example of the subsurface flow map in Fig.~\ref{casestudy}\emph{a}). This equatorward flow acts as a counter-cell of the meridional flow (present at the same longitudes as the corresponding magnetic region) and causes a decrease of the mean meridional flow amplitude in the activity belt. This behaviour is noticed in all studied cases recorded during eleven non-consecutive solar rotations, for which the Dynamics data useful for helioseismic inversion exist. Therefore the formation of the apparent counter-cell seems to be a common property of all large active regions in depths 3--12~Mm.
198:
199: Flows in this counter-cell do not influence the magnetic flux transport, which can be demonstrated when the magnetic region is excluded from the synoptic map (Fig.~\ref{casestudy}\emph{b}). The calculated meridional circulation profile is then closer to the profile of the meridional flux transport speed derived from the magnetic butterfly diagram.
200:
201: \section{Conclusions}
202:
203: We have compared the measurements of the meridional speed derived from two
204: different techniques: by time-distance local helioseismology and by
205: measuring the flux transport speed using the magnetic butterfly
206: diagram. We have found that the results agree quite well in
207: general, but they differ in regions occupied by local
208: magnetic fields. The detailed flow maps from helioseismology show that this is partly due to the presence of meridional counter-cells at the
209: equatorial side of magnetic regions, which influence the time-distance derived meridional
210: flow profile, but does not influence the magnetic flux transport.
211:
212: We have studied eleven non-consecutive Carrington rotations covering one solar cycle. The effect of the
213: local flows around active regions and especially on their equatorial side is noticed in all the studied
214: cases. Therefore, this behaviour seems to be a common property of the subsurface dynamics around active regions
215: located in the activity belt. However, we have to keep in mind that
216: both datasets are not directly comparable, since the time-distance flow maps represent the behaviour of
217: flows during one Carrington rotation, while the butterfly diagram tracking procedure provide results
218: averaged over few Carrington rotations. The more homogeneous data for local helioseismology are needed
219: to study this effect in more detail.
220:
221: The results show that the speed of the magnetic flux transport
222: towards the solar poles may significantly deviate from the
223: longitudinally averaged meridional flow speed derived from local
224: helioseismology measurements, which are affected by local
225: circulation flows around active regions in the activity belt.
226: Therefore, using the longitudinally averaged meridional flow profile
227: from helioseismology in the solar cycle models for description the
228: flux transport is not justified. The longitudinal structure of these
229: flows should be taken into account.
230:
231:
232: \acknowledgments \sloppypar M\v{S} would like to gratefully acknowledge
233: ESA-PECS for the support under grant No.~8030, the Ministry of
234: Education of the Czech Republic under Research Program
235: MSM0021620860, and Stanford Solar Physics Group for support and hospitality.
236:
237:
238: \begin{thebibliography}{}
239: \bibitem[Brandenburg \& Subramanian(2005)]{Brandenburg2005}
240: Brandenburg, A., \& Subramanian, K.\ 2005, \physrep, 417, 1
241: \bibitem[Brun(2004)]{2004SoPh..220..333B} Brun, A.~S.\ 2004, \solphys, 220,
242: 333
243: \bibitem[Dikpati et al.(2006)]{2006GeoRL..3305102D} Dikpati, M., de Toma,
244: G., \& Gilman, P.~A.\ 2006, \grl, 33, 5102
245: \bibitem[Dikpati \& Gilman(2006)]{2006ApJ...649..498D} Dikpati, M., \&
246: Gilman, P.~A.\ 2006, \apj, 649, 498
247: \bibitem[Gonz{\'a}lez Hern{\'a}ndez et al.(2006)]{2006ApJ...638..576G}
248: Gonz{\'a}lez Hern{\'a}ndez, I., Komm, R., Hill, F., Howe, R., Corbard, T.,
249: \& Haber, D.~A.\ 2006, \apj, 638, 576
250: \bibitem[Hathaway(2003)]{hat03} Hathaway, D. H. 2003, in Proceedings of
251: SOHO 12/ GONG+ 2002 Local and Global Helioseismology: the Present and
252: Future, ESA SP-517, Noordwijk, Netherlands, 87
253: \bibitem[Rieutord et al.(2007)]{rieutord07} Rieutord, M., Roudier,
254: T., Roques, S., \& Ducottet, C.\ 2007, \aap, 471, 687
255: \bibitem[Zhao \& Kosovichev(2004)]{2004ApJ...603..776Z} Zhao, J., \&
256: Kosovichev, A.~G.\ 2004, \apj, 603, 776
257:
258: \end{thebibliography}
259:
260: \clearpage
261:
262: \begin{figure}
263: \epsscale{0.45} \plotone{f1.eps} \caption{a) The magnetic butterfly
264: diagram for cycle 23. b) The filtered magnetic butterfly diagram
265: showing enhancements of the flux transport elongated structures. c)
266: The measured meridional flux transportation speed (in the
267: South--North direction) for Carrington rotations
268: 1900--2048.\label{maps}}
269: \end{figure}
270:
271: \begin{figure}
272: \epsscale{0.8}
273: \plotone{f2.eps} \caption{The longitudinally averaged meridional
274: flow speed measured for a set of Carrington rotations by
275: time-distance helioseismology. Solid line plots the time-distance
276: mean meridional flow at 3--4.5~Mm depth, dotted at 6--9~Mm, and
277: dashed at 9--12~Mm. The dots with error-bars represent the
278: 10-degree-bin-averaged values of the flux transport speed derived
279: from the magnetic butterfly diagram (Fig~\ref{maps}a). Error bars of
280: time-distance measurements are included for reference in CR~1911
281: \citep[for details, see][]{2004ApJ...603..776Z}.
282: \label{rotations}}
283: \end{figure}
284:
285: \begin{figure}
286: \epsscale{1} \plotone{f3.eps} \caption{a) A large-scale flow map at
287: depth 3--4.5~Mm for Carrington rotation No. 1975 (April 2001) and the
288: corresponding MDI magnetogram in the grey-scale background.
289: Large-scale flows towards the equator in the magnetic regions are
290: visible around the large active region. b) The longitudinally
291: averaged meridional circulation profile for the same Carrington
292: rotation. The southern hemisphere depicts almost no magnetic
293: activity, so the meridional circulation profile obtained by
294: averaging the time-distance flow map (solid line) almost fits the
295: magnetic flux transport profile (points with error-bars) there,
296: while on the northern hemisphere they differ. After masking the
297: magnetic regions on the northern hemisphere, the recalculated
298: profile (dotted line) tends to fit the butterfly tracking one also
299: on the northern hemisphere. \label{casestudy}}
300: \end{figure}
301:
302:
303: \clearpage
304:
305: \end{document}
306: