1: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%
2: %%
3: %% This is a sample manuscript marked up using the
4: %% AASTeX v5.x LaTeX 2e macros.
5:
6: %% The command below calls the preprint style
7: %% which will produce a one-column, single-spaced document.
8: %% Examples of commands for other substyles follow. Use
9: %% whichever is most appropriate for your purposes.
10:
11: %Note: comment out \epsscale commands in figures
12: \documentclass[12pt,preprint]{aastex}
13:
14: %% manuscript produces a one-column, double-spaced document:
15:
16: % \documentclass[manuscript]{aastex}
17:
18: %% preprint2 produces a double-column, single-spaced document:
19:
20: %Note: uncomment \epsscale commands in figures
21: % \documentclass[10pt,preprint2]{aastex}
22:
23: \usepackage{graphicx}
24: \usepackage{amsmath}
25: %\usepackage{natbib}
26:
27: %
28: % hector's
29:
30: \def \etal {et al.}
31: \def \apj {ApJ}
32: \def \apjs {ApJS}
33: \def \apjl {ApJ}
34: \def \solphys {Solar Phys.}
35: \def \pasj {Pub. Astron. Soc. Japan}
36: \def \aap {A\&A}
37:
38: % for astrobib
39: %\newcommand{\citeauthor}[1]{\def\citeauthoryear##1##2##3{\rm ##1}\cite{#1}}
40: %\newcommand{\citeyear}[1]{\def\citeauthoryear##1##2##3{\rm ##3}\cite{#1}}
41:
42: \newcommand{\citeN}[1]{\citeauthor{#1} (\citeyear{#1})}
43: \newcommand{\citeNP}[1]{\citeauthor{#1} \citeyear{#1}}
44: \newcommand{\citeyearNP}[1]{\citeyear{#1}}
45: \newcommand{\citeANP}[1]{\citeauthor{#1}}
46:
47: %% --- Some useful spectral definitions
48:
49: \newcommand{\CaII}{\ion{Ca}{2}}
50: \newcommand{\NaI}{\ion{Na}{1}}
51: \newcommand{\mxcm}{Mx~cm$^{-2}$}
52:
53: %% --- Setting to make eqnarrays look better. Unfortunately, AASTex
54: %% does not link to AMS styles which have better mechanisms to
55: %% deal with tabbed equations.
56:
57: \setlength{\arraycolsep}{1pt}
58:
59:
60: \shortauthors{Socas-Navarro & Lites}
61: \shorttitle{Small-Scale Mixture of Field Strengths in the Quiet Sun}
62:
63:
64: %---------main text
65: %
66: %\received{2004 May 21}
67: \begin{document}
68:
69: %
70: \title{Observational Evidence for Small-Scale Mixture of Weak and Strong
71: Fields in the Quiet Sun}
72:
73: \author{H. Socas-Navarro}
74: \affil{High Altitude Observatory, NCAR\thanks{The National Center
75: for Atmospheric Research (NCAR) is sponsored by the National Science
76: Foundation.}, 3450 Mitchell Lane, Boulder, CO 80307-3000, USA}
77: \email{navarro@ucar.edu}
78:
79: \author{B.W. Lites}
80: \affil{High Altitude Observatory, NCAR, 3450 Mitchell Lane, Boulder,
81: CO 80307-3000, USA}
82: \email{lites@ucar.edu}
83:
84: %\date{}%
85:
86: \begin{abstract}
87: Three different maps of the quiet Sun, observed with the Advanced Stokes
88: Polarimeter (ASP) and the Diffraction-Limited Stokes Polarimeter (DLSP), show
89: evidence
90: of strong ($\simeq$1700~G) and weak ($<$500~G) fields coexisting within the
91: resolution element both at network and internetwork locations. The angular
92: resolution of the observations is of 1\arcsec
93: (ASP) and 0.\arcsec6 (DLSP), respectively. Even at the higher DLSP
94: resolution, a significant fraction of the network magnetic patches harbor a
95: mixture of strong and weak fields. Internetwork elements that exhibit kG
96: fields when analyzed with a single-component atmosphere are also shown to
97: harbor considerable amounts of weak fields. Only those patches for which
98: a single-component analysis yields weak fields do not show this mixture of
99: field strengths. Finally, there is a larger fractional area of weak fields in
100: the convective upflows than in the downflows.
101: \end{abstract}
102:
103: \keywords{line: profiles -- Sun: atmosphere -- Sun: magnetic fields --
104: Sun: photosphere}
105:
106:
107: \section{Introduction}
108: \label{sec:intro}
109:
110: Our understanding of quiet Sun magnetic fields is evolving at a very
111: rapid pace. The classical picture of the quiet Sun is based on a
112: % bunch of
113: sprinkling of kilo-Gauss (kG)
114: fluxtube-like structures of sub-arcsecond sizes (unresolved in the
115: observations) forming a magnetic network at the
116: boundaries of the
117: supergranular cells. The internetwork (cell interiors) would be almost devoid
118: of flux, exhibiting only sparse weak turbulent flux concentrations. As new
119: observations became available
120: with improved sensitivity and spatial resolution, the
121: internetwork region started to gain importance. Strong kG fields were also
122: found in many internetwork locations and the amount of magnetic flux and
123: energy detected has increased as the instrumentation improved
124: (\citeNP{KDE+94}; \citeNP{L95}; \citeNP{GDKS96}; \citeNP{LR99};
125: \citeNP{SAL00}; \citeNP{L02};
126: \citeNP{SNSA02}; \citeNP{KCS+03}; \citeNP{DCKSA03}; \citeNP{DCSAK03};
127: \citeNP{SNMPL04}; \citeNP{LSN04}).
128:
129: At the same time, the concept of organized magnetic structures (e.g.,
130: fluxtubes) as the building blocks for quiet Sun fields seems to be losing
131: some ground, at least outside of the network. Recent numerical simulations
132: (\citeNP{C99}; \citeNP{EC01}; \citeNP{SAEC03}; \citeNP{S03}; \citeNP{SN02})
133: reveal a more disorganized,
134: almost chaotic, scenario with the field being dragged around by turbulent
135: convective motions. Another important element in this picture is the
136: discrepancy in the distribution of internetwork fields as seen in visible and
137: infrared observations. Infrared data (\citeNP{L95}; \citeNP{KCS+03}) show a
138: predominance of sub-kG fields, with a distribution that peaks around 350~G. On
139: the other hand, authors working with visible observations (e.g.,
140: \citeNP{SNSA02}) obtain that most of these fields are of kG
141: strength. \citeN{SNSA03} proposed that this discrepancy is a natural
142: consequence of unresolved small-scale inhomogeneities of the field. If one
143: has a mixture of weak and strong fields coexisting in the resolution element,
144: visible and infrared observations tend to emphasize different parts of the
145: distribution.
146:
147: In a recent paper, \citeN{SN04b} (see also \citeNP{SN04a}) showed that the
148: visible \ion{Fe}{1} lines at 6302 \AA \, exhibit some sensitivity to
149: unresolved field strength inhomogeneities. In the present work we make use of
150: this property to seek evidence that mixed strengths indeed occur in
151: the quiet Sun. As we discuss below, we are able to detect a rather large
152: number of such mixed strengths. Our results support the ``disorganized''
153: picture of the quiet Sun fields discussed above, not only in the internetwork
154: but also to some extent in the magnetic network.
155:
156:
157: \section{Observations and analysis}
158: \label{sec:obs}
159:
160: The datasets analyzed in this paper come from two different instruments. The
161: Advanced Stokes Polarimeter (ASP) is a spectro-polarimeter for the Dunn Solar
162: Telescope (DST) at the Sacramento Peak observatory (Sunspot, NM, USA),
163: operated by
164: the National Solar Observatory. We used the map observed by \citeN{L96} on
165: Sep 29, 1994 (hereafter referred to as Map~1). This map has a spatial
166: resolution of $\simeq$1\arcsec \, and a spectral resolution of
167: $\simeq$30~m\AA . The spectral lines observed are the pair of \ion{Fe}{1} lines
168: in the 6302 \AA \, region.
169:
170: The other instrument employed is the Diffraction-Limited Stokes Polarimeter
171: (DLSP), which is operated at the same telescope. The DLSP is a new
172: instrument that has been designed specifically to take advantage of the new
173: adaptive optics (AO) system at the DST in order to achieve very high angular
174: resolution. It has been optimized for the routine observation
175: of the 6302 \AA \, spectral region.
176: %??? (remove this?) Presently, the DLSP
177: %uses the old ASP cameras and control system, as well as the phase~I low-order
178: %AO system. Its full potential will be realized in the near future when the
179: %new high-order AO and new detectors become available.
180:
181: %Even with the present DLSP configuration,
182: \citeN{LSN04} obtained what can
183: be considered very high-resolution Stokes observations of the quiet Sun,
184: of approximately 0.\arcsec6. This figure, alongside with the 0.\arcsec5
185: reached by \citeN{DCKSA03}, represent the highest resolution
186: spectro-polarimetric observations of quiet Sun fields made thus far. In our
187: analysis here we consider the best two maps observed by \citeN{LSN04}, namely
188: the ones observed on Sep 14 and Sep 16 2003 (hereafter Map~2 and Map~3,
189: respectively).
190:
191: %We carried out two different analyses on each one of the three data sets,
192: %which are described below in some detail.
193:
194: \subsection{One-component analysis}
195: \label{sec:1c}
196:
197: We first conducted a one-component (hereafter 1C) analysis, assuming that we
198: have one field strength ($B$) that occupies a certain area filling factor
199: ($\alpha$) of the spatial pixel. Instead of performing an iterative
200: least-squares fit to the observations, we
201: chose to do forward modeling from a large number of models. In this manner we
202: make sure that the entire model space is probed and that the
203: absolute minimum of the $\chi^2$ merit function is found. We start with two
204: models for the thermodynamic parameters (temperature, gas
205: pressure, micro- and macroturbulence and line-of-sight velocity) of the quiet
206: Sun, representing a granule and an intergranular lane. The models were
207: obtained from the inversion of average Stokes~I profiles.
208: The Stokes~V profiles were then synthesized for many values of the
209: magnetic field strength (ranging from $B=300$ to $B=2000$~G) and a global
210: velocity offset (from $v=-2.5$ to $v=2.5$ km~s$^{-1}$). The
211: calculations were performed with the code LILIA (\citeNP{SN01a}), assuming
212: LTE and hydrostatic equilibrium.
213:
214: Depending on the continuum intensity of the observations, we used the granule
215: or lane models or a suitable linear combination of both. In the case of the
216: quiet Sun this is approach is valid because the thermodynamical properties
217: of the atmosphere experience only relatively small variations across the
218: map. Fig~\ref{intfits} shows the fits obtained with this method to intensity
219: profiles from four randomly-chosen locations. Reasonably good fits are
220: obtained for the entire dataset analyzed, which justifies the approximation
221: used for the thermodynamics of the atmosphere.
222:
223: \begin{figure}
224: \plotone{f4.eps}
225: \caption{
226: \label{intfits}
227: Fits (dashed) to observed (solid) intensity profiles corresponding
228: to (x,y) coordinates (20,60),(50,90),(80,120) and (110,150) of Map~2.
229: }
230: \end{figure}
231:
232: The synthetic Stokes~V profiles ($V^{syn}$)
233: were multiplied by a filling factor $\alpha=A(V^{obs})/A(V^{syn})$ (where $A$
234: denotes the amplitude of a given profile) and then compared one by one to the
235: observed profile ($V^{obs}$) at each spatial position of the maps. The sign
236: of the synthetic profile is chosen to match the polarity of $V^{obs}$.
237: %, which
238: %is determined as the integral in wavlength of the red lobe of the lines
239: %minus the blue one.
240: %The merit
241: %function $\chi^2$ was computed for each model as the sum of squared
242: %differences between $\alpha V^{syn}$ and $V^{obs}$.
243: The values of $B$, $v$
244: and $\alpha$ that yield the best fit to $V^{obs}$ are taken as representative
245: of the conditions at the spatial location under consideration. This process
246: is repeated for every pixel in every map. The spatial distributions of $\alpha$
247: and $B$ thus obtained are shown if Figs~\ref{fig:map1} and~\ref{fig:map23}
248: (upper panels).
249:
250: It is important to note that the 6302 \AA \, lines are in the weak field
251: regime (i.e., the regime in which the Zeeman splitting is much smaller than
252: the Doppler width) for field strengths lower than $\simeq$500~G. In the
253: weak field regime the shape of the Stokes~V profiles is independent of the
254: magnetic field. The profiles are simply scaled with the value of the
255: field. Thus, it is not possible to disentangle the effects of
256: the filling factor from the field strength. In other words, any value of the
257: field weaker than 500~G will result in exactly the same $\chi^2$.
258: %This is the
259: %reason why we took this value as a lower limit for the field
260: %strength.
261: Therefore, the reader must
262: keep in mind that any fields below 500~G depicted in the figures might be
263: actually weaker (with correspondingly larger filling factors).
264:
265: \subsection{Two-component analysis}
266: \label{sec:2c}
267:
268: \citeN{SN04b} suggested that the \ion{Fe}{1} lines at 6302 \AA
269: \, exhibit some sensitivity to the presence of two magnetic strength
270: components. When the two field strengths are at least as far apart as 500~G
271: and 1700~G, their response functions are sufficiently decoupled that the
272: {\it relative} filling factors of these two components can be inferred with
273: an uncertainty of 0.10 or less.
274: %As noted in that paper, the lower value of
275: %20~G was chosen arbitrarily and any other value within the weak field regime
276: %could have been used. This is because, within this regime, the response
277: %functions of the lines are the same regardless of the value of the field.
278:
279: As a further step in our study, we carried out a two-component (2C) analysis
280: of the observations.
281: Considering the arbitrariness in the strength of the weak component we
282: chose a value of 300~G, which is close to the peak of the distribution
283: obtained from 1C inversions of infrared internetwork observations
284: (\citeNP{KCS+03}). Thus, we (arbitrarily) fixed the strengths of the weak and
285: strong components to 300~G and 1700~G, respectively. The problem now is to
286: find the filling factors $\alpha_w$ and $\alpha_s$. Let us define these filling
287: factors relative to the magnetic element, so that $\alpha_w +
288: \alpha_s = 1$. The total magnetic filling factor in the observed pixel is
289: still $\alpha$. The filling factors of the two components relative to the
290: resolution element are then $\alpha \alpha_w$ and $\alpha \alpha_s$. This
291: convention may seem somewhat confusing at first, but it is useful for the
292: discussion in \S\ref{sec:results} below.
293:
294: The synthetic profiles $V^{syn}$ for the 2C case were calculated in the
295: following manner. As before, we start with the a model for the atmospheric
296: thermodynamics which depends on the observed continuum intensity. We then
297: synthesized the profiles $V^w$ and $V^s$ for 300
298: and 1700~G, respectively. At each spatial location we compared the profiles
299: $V^{obs}$ and $\alpha V^{syn}=\alpha (\alpha_w V^w + \alpha_s V^s)$ (where,
300: again, $\alpha=A(V^{obs})/A(V^{syn})$). The combinations of
301: $\alpha$, $\alpha_w$, $\alpha_s$ and $v$ that
302: lead to the best fit of the observations are
303: selected. Notice that only two of these
304: parameters, $v$ and either $\alpha_s$ or $\alpha_w$, are independent.
305:
306: In order to ensure that the detection of mixed field strengths
307: has significance, we rewarded solutions with either $\alpha_w=0$ or
308: $\alpha_s=0$. The $\chi^2$ corresponding to these solutions is reduced by
309: 10\%. This conservative
310: approach makes the procedure ``prefer'' one-component solutions. It is also
311: important to point out that, while we are not adding any degrees of freedom
312: when going from 1C (above) to 2C, the $\chi^2$ is smaller in
313: the 2C analysis virtually everywhere. These arguments (alongside with those
314: in \S\ref{sec:reliability}) give us confidence in
315: the results reported below.
316:
317: \section{Results}
318: \label{sec:results}
319:
320: Figs~\ref{fig:map1} and~\ref{fig:map23} show the spatial distribution of
321: $\alpha$ (upper left), $B$ (upper right), $\alpha_s$ (lower left) and
322: $\alpha_w$ (lower right). The 1C analysis results in strong kG fields in most
323: spatial pixels inverted (i.e., those exhibiting significant polarization
324: signal), with the exceptions of very few weak field patches in the
325: network and some weak field elements in the internetwork. This is consistent
326: with previous observations published in the literature based on visible
327: lines, in which the distribution of fields peaks around 1.5~kG (see
328: references in \S\ref{sec:intro}).
329:
330: The lower panels of the figures clearly show the presence of mixed-strength
331: pixels both in network and internetwork locations. Let us start by discussing
332: the results for the internetwork. In this region, visible and infrared
333: observations (always using 1C analyses) have led to disparate conclusions,
334: with the infrared lines showing a much larger fraction of weaker
335: fields. According to \citeN{SNSA03}, this can be explained by a small-scale
336: mixture of weak and strong fields beyond the spatial resolution. They showed
337: that, when such mixture exists, a 1C inversion of the visible lines is biased
338: towards the stronger fields. Therefore, we would expect to have mixed field
339: strengths in those spatial pixels where the 1C analysis results in strong
340: fields. Those with weak fields, on the other hand, are probably rather
341: homogeneous. This is exactly what we find in the analysis of our internetwork
342: profiles, as seen in Figs~\ref{fig:map1} and~\ref{fig:map23}, as well as in
343: Table~\ref{table:maps}. We did not find any mixed strengths in pixels with
344: $\alpha < 0.1$ and $B < 500$~G.
345:
346: Let us now turn to the network. The relatively strong flux concentrations in
347: network patches have been traditionally associated with strong kG fields (see
348: references in \S\ref{sec:intro}). While our 1C analysis agrees with this
349: assessment, the 2C analysis reveals the presence of a significant amount of
350: weak fields mixed at small scales. The mixed strengths occur all over the
351: network patches and not only
352: around their perimeter. Table~\ref{table:maps} lists some properties of the
353: network elements, including the percentage of pixels showing mixed
354: strengths. Finally, we find (not shown in the table) that the fractional area
355: occupied by weak fields is larger in the convective upflows than in the
356: downflows (as one would expect).
357:
358: \clearpage
359:
360: \begin{figure*}
361: \vspace{-0.5in}
362: \epsscale{0.9}
363: \plotone{f1.eps}
364: \caption{Map 1 (ASP). Upper left: Total magnetic filling factor, saturated at
365: 0.10. The color scale in this panel ranges from 0 to 0.10. Upper right: Field
366: strength inferred from the 1C inversion (G). Lower
367: left: Relative filling factor of the strong field component. Lower right:
368: Relative filling factor of the weak field component. In the lower left
369: (right) image the color scale represents the percentage of the magnetic
370: area occupied by strong (weak) fields. Spatial units are
371: arc-seconds. Pixels appearing black in both
372: images do not exhibit polarization signal above the noise and have been
373: excluded from the analysis.
374: \label{fig:map1}
375: }
376: \end{figure*}
377:
378: \begin{figure*}
379: %\epsscale{2.2}
380: \plotone{f2.eps}
381: \caption{Maps 2 and 3 (DLSP). Four panels represent the following quantities
382: for each map. Upper left: Total magnetic filling factor,
383: saturated at
384: 0.10. The color scale in this panel ranges from 0 to 0.10. Upper right:
385: Field strength inferred from the 1C inversion (G). Lower
386: left: Relative filling factor of the strong field component. Lower right:
387: Relative filling factor of the weak field component. In the lower left
388: (right) image the color scale represents the percentage of the magnetic
389: area occupied by strong (weak) fields. Spatial units are
390: arc-seconds. Pixels appearing black in both
391: images do not exhibit polarization signal above the noise and have been
392: excluded from the analysis.
393: \label{fig:map23}
394: }
395: \end{figure*}
396: %\begin{figure*}
397: %\epsscale{1.}
398: %\plotone{f3.eps}
399: %\caption{Map 3 (DLSP). Upper left: Total magnetic filling factor, saturated at
400: % 0.10. The color scale in this panel goes from 0 to 0.10. Upper right: Field
401: % strength inferred from the 1C inversion (G). Lower
402: % left: Relative filling factor of the strong field component. Lower right:
403: % Relative filling factor of the weak field component
404: %\label{fig:map3}
405: %}
406: %\end{figure*}
407:
408: \clearpage
409:
410: \begin{deluxetable}{lccc}
411: \tablewidth{0pt}
412: \tablecaption{Flux and mixed-strengths statistics
413: \label{table:maps}}
414: \tablehead{ & Map 1 & Map 2 & Map 3 }
415: \startdata
416: Mean 1C flux (\mxcm) & 8.7 & 5.2 & 6.2 \\
417: Mean 2C flux (\mxcm) & 9.2 & 5.4 & 6.1 \\
418: Mean 1C field (G) & 1529 & 999 & 1064 \\
419: Mean 2C field (G) & 1192 & 928 & 924 \\
420: Pixels analyzed & 25.9\% & 24.5\% & 24.0\% \\
421: Pixels harboring & & & \\
422: mixed strengths & 14.8\% & 7.6\% & 2.8\% \\
423: Mixed strength in pixels & & & \\
424: with $\alpha < 0.1$, $B < 500$~G & 0\% & 0\% & 0\% \\
425: Mean flux in & & & \\
426: strong fields (\mxcm) & 6.3 & 3.3 & 2.6 \\
427: Mean flux in & & & \\
428: weak fields (\mxcm) & 2.9 & 2.1 & 3.5 \\
429: Fraction of upflowing & & & \\
430: area with strong fields & 12.9\% & 1.6\% & 9.5\% \\
431: Fraction of downflowing & & & \\
432: area with strong fields & 28.2\% & 24.4\% & 13.2\% \\
433: \enddata
434: \end{deluxetable}
435:
436:
437: \clearpage
438:
439: \section{Reliability of the results}
440: \label{sec:reliability}
441:
442: We have not been concerned thus far with problems such as line asymmetries or
443: details of the sub-pixel field distribution. This would be a concern if we
444: were interested in obtaining a detailed description of the solar atmosphere,
445: including velocity and magnetic field gradients. Our goal, however, is not
446: that, but rather to
447: demonstrate that weak and strong fields generally coexist at sub-arcsecond
448: spatial scales. How is this overall conclusion affected by the simplifying
449: assumptions employed in our work? This important question is explored in the
450: present section by means of numerical simulations. Before going into the
451: details, though, it is probably a good idea to consider the broader context
452: of the subject in order to put our work into perspective.
453:
454: In doing analysis of solar magnetic fields based on spectro-polarimetric
455: observations there
456: are various levels of complexity that one can consider, depending on the
457: sophistication of the physical models employed. The most simplistic analyses
458: are the ``traditional'' techniques, such as the line ratio, the
459: separation of Stokes~V extrema, or fits by Gaussian profiles. These techniques
460: do not involve radiative transfer calculations and do not require a detailed
461: knowledge of the atmospheric model. They typically provide only a rough
462: approximation to the magnetic field, but are unable to deal with line
463: asymmetries due to unresolved inhomogeneities, atmospheric gradients, etc.
464: The most sophisticated techniques
465: make use of full radiative transfer calculations in detailed model
466: atmospheres to fit the observed profiles in a least-squares sense (inversion
467: codes).
468:
469: Most of the published work (see references in
470: \S\ref{sec:intro}) makes use of the simplistic techniques to infer ``typical''
471: values of the photospheric field in the quiet Sun. These simplified analyses
472: led to the inference that network fields are typically strong
473: ($\simeq$1.5~kG), or that internetwork fields appear to be mostly kG when
474: observed in the visible and sub-kG ($\simeq$300~G) when observed in the
475: infrared. Very few authors (e.g., the works of \citeNP{SAL00};
476: \citeNP{SNSA02}) have used inversions that consider detailed models with
477: atmospheric gradients. The overall picture (i.e., whether the fields are weak
478: or strong) does not change signifincantly, although the inversion codes
479: obviously provide
480: more information. Detailed inversions have not been carried out in the
481: infrared so far. The analyses published in the literature are derived from
482: fits of Gaussian functions to the observed profiles.
483:
484: Our analysis in this paper lies somewhere between the two
485: ends of the range of sophistication. We calculate actual Stokes profiles from
486: a model atmosphere, but do not consider line-of-sight variations of the
487: quantities. The thermodynamics of our models is probably
488: reasonably good, as suggested by the fits shown in
489: Fig~\ref{intfits}. However, there are two important simplifications. First,
490: we do not consider line-of-sight gradients of the field. Second, we
491: parameterize the unresolved field distribution by two discrete field values
492: of 300~G and 1700~G. It is important to note that none of these assumptions
493: are worse than those of the traditional methods (save for the exceptions
494: noted above) which are based on single-valued fields without height
495: variations.
496:
497: We have carried out some numerical simulations in order to gain insights
498: into the adequateness of our approximations. First consider how our two-point
499: approach is representative of the sub-pixel field distribution.
500: Suppose that we have an unresolved probability
501: distribution function (PDF) of field strengths $f(B)$. The filling factor of
502: fields with a strength between $B$ and $B+\Delta B$ is:
503:
504: \begin{equation}
505: \alpha(B,B+\Delta B)=\int_B^{B+\Delta B} f(B) dB \, .
506: \end{equation}
507:
508: The emerging profile $P(\lambda)$ from such distribution is:
509:
510: \begin{equation}
511: P(\lambda)=\int_0^\infty f(B) P_B(\lambda) dB \, ,
512: \end{equation}
513: where $P_B(\lambda)$ denotes the profile produced by a field of strength
514: $B$. For the moment we are only interested in horizontal
515: inhomogeneities. The effects of gradients of $B$ along the line of sight will
516: be discussed later. If we consider $P(\lambda)$ as a simulated observation
517: and apply to it the procedure introduced in \S\ref{sec:2c} above, we can
518: compare the inferred values with the actual PDF, $f(B)$,
519: employed. Table~\ref{table:pdf} shows the results of various tests with
520: different shapes of the PDF. As proposed by \citeN{EC01},
521: and by \citeN{KCS+03}, \citeN{SNSA03}, and \citeN{TBSAR04}, we adopted an
522: exponential dependence\footnote{
523: It should be noted that the PDF obtained by \citeN{EC01} from the
524: simulations is actually somewhat more complicated. Their PDF is a
525: stretched exponential in the absence of net unsigned flux, and has a
526: shoulder when such flux is non-zero.
527: } for $f(B)$ with a
528: normalization factor $N^{-1}=\int_0^\infty f(B) dB$. The filling factor of
529: fields stronger than 2~kG has been set to zero. We also considered a
530: Dirac-delta PDF ($\delta(B-B_0)$), which simply means that the field is
531: homogeneous over the resolution element, with a strength $B_0$.
532:
533: \begin{deluxetable}{ccccc}
534: \tablewidth{0pt}
535: \tablecaption{Results from simulations. Inferrences using values of 300~G
536: and 1700~G.
537: \label{table:pdf}}
538: \tablehead{ PDF & Actual & Inferred & Actual & Inferred \\
539: ($B$ in G) & $\alpha(B<$1~kG) &
540: $\alpha(B=300$~G) & $\alpha(B>$1~kG) &
541: $\alpha(B=1.7$~kG) }
542: \startdata
543: $f=\delta(B-100)$ & 1.00 & 1.00 & 0.00 & 0.00 \\
544: $f=\delta(B-2000)$ & 0.00 & 0.00 & 1.00 & 1.00 \\
545: $f=N\exp(-B/100)$ & 1.00 & 1.00 & 0.00 & 0.00 \\
546: $f=N\exp(-B/300)$ & 0.96 & 0.93 & 0.04 & 0.07 \\
547: $f=N\exp(-B/600)$ & 0.84 & 0.82 & 0.16 & 0.18 \\
548: $f=N\exp(-B/1000)$ & 0.73 & 0.75 & 0.27 & 0.25 \\
549: $f=N\exp(-B/1500)$ & 0.66 & 0.70 & 0.34 & 0.30 \\
550: $f=N\exp(-B/2000)$ & 0.62 & 0.67 & 0.38 & 0.33 \\
551: \enddata
552: \end{deluxetable}
553:
554: The results in Table~\ref{table:pdf} suggest that the filling factors that we
555: obtained for $B=300$~G and $B=1700$~G are more or less representative of the
556: weak and strong fields present in more complex distributions expected to be
557: present in the actual quiet Sun. We also tested the sensitivity of our
558: results to the values that we used to represent weak and
559: strong fields. To this aim we repeated the calculations above but this time
560: using
561: 500 and 1500~G. The results obtained, listed in Table~\ref{table:pdf2}, are
562: very similar to those of Table~\ref{table:pdf}.
563:
564: \begin{deluxetable}{ccccc}
565: \tablewidth{0pt}
566: \tablecaption{Results from simulations. Inferrences using values of 500~G
567: and 1500~G.
568: \label{table:pdf2}}
569: \tablehead{ PDF & Actual & Inferred & Actual &
570: Inferred \\
571: ($B$ in G) & $\alpha(B<$1~kG) &
572: $\alpha(B=500$~G) & $\alpha(B>$1~kG) &
573: $\alpha(B=1.5$~kG) }
574: \startdata
575: $f=\delta(B-100)$ & 1.00 & 1.00 & 0.00 & 0.00 \\
576: $f=\delta(B-2000)$ & 0.00 & 0.00 & 1.00 & 1.00 \\
577: $f=N\exp(-B/100)$ & 1.00 & 1.00 & 0.00 & 0.00 \\
578: $f=N\exp(-B/300)$ & 0.96 & 0.97 & 0.04 & 0.03 \\
579: $f=N\exp(-B/600)$ & 0.84 & 0.79 & 0.16 & 0.21 \\
580: $f=N\exp(-B/1000)$ & 0.73 & 0.77 & 0.27 & 0.33 \\
581: $f=N\exp(-B/1500)$ & 0.66 & 0.60 & 0.34 & 0.40 \\
582: $f=N\exp(-B/2000)$ & 0.62 & 0.57 & 0.38 & 0.43 \\
583: \enddata
584: \end{deluxetable}
585:
586: Let us now consider the issue of velocity and magnetic field gradients along
587: the line of sight. Such gradients give rise to asymmetries in the line
588: profiles. We introduced gradients in our simulated observations and tested
589: their effects on the filling factors inferred by our analysis. The
590: line-of-sight velocity that we introduced varies linearly
591: between $\tau_{500}=10$ and $\tau_{500}=10^{-3}$ (with $\tau_{500}$ denoting
592: the optical depth at 500~nm). The amplitude of the
593: variation is of 2~km~s$^{-1}$. The field strength varies linearly as well,
594: from $B$ at $\tau_{500}=10^{-1}$ (with $B$ ranging from 0 to 2~kG) to zero at
595: $\tau_{500}=10^{-3}$. The asymmetric profiles calculated from the PDFs above
596: are again taken as simulated observations and applied our analysis (which
597: neglects gradients). The results from this experiment are summarized in
598: Table~\ref{table:pdf3} (the numbers within parentheses are obtained when
599: the velocity gradient is doubled). These values indicate that the presence of
600: asymmetries does not invalidate our conclusions on the coexistence of field
601: strengths in the resolution element.
602:
603: \begin{deluxetable}{ccccc}
604: \tablewidth{0pt}
605: \tablecaption{Results from simulations with asymmetric profiles.
606: \label{table:pdf3}}
607: \tablehead{ PDF & Actual & Inferred & Actual &
608: Inferred \\
609: ($B$ in G) & $\alpha(B<$1~kG) &
610: $\alpha(B=500$~G) & $\alpha(B>$1~kG) &
611: $\alpha(B=1.5$~kG) }
612: \startdata
613: $f=\delta(B-100)$ & 1.00 & 1.00 (0.97) & 0.00 & 0.00 (0.03) \\
614: $f=\delta(B-2000)$ & 0.00 & 0.01 (0.00) & 1.00 & 0.99 (1.00) \\
615: $f=N\exp(-B/100)$ & 1.00 & 0.99 (0.95) & 0.00 & 0.01 (0.05) \\
616: $f=N\exp(-B/300)$ & 0.96 & 0.91 (0.87) & 0.04 & 0.09 (0.13) \\
617: $f=N\exp(-B/600)$ & 0.84 & 0.80 (0.75) & 0.16 & 0.20 (0.25) \\
618: $f=N\exp(-B/1000)$ & 0.73 & 0.73 (0.75) & 0.27 & 0.27 (0.25) \\
619: $f=N\exp(-B/1500)$ & 0.66 & 0.69 (0.63) & 0.34 & 0.31 (0.37) \\
620: $f=N\exp(-B/2000)$ & 0.62 & 0.66 (0.60) & 0.38 & 0.34 (0.40) \\
621: \enddata
622: \end{deluxetable}
623:
624: \section{Conclusions}
625: \label{sec:conc}
626:
627: This paper reports on observational evidence for mixed
628: field strengths in the quiet Sun ($\sim$300 and $\sim$1700~G) on spatial
629: scales smaller than 0.\arcsec6. Mixed strengths are found in network and
630: internetwork magnetic elements. The results for the internetwork are not
631: entirely unexpected. \citeN{SNSA03} proposed that such mixture is the
632: most natural explanation for the discrepancy between visible and
633: infrared observations. Our work strongly supports their
634: conjecture and starts to bridge the gap between the two types of
635: observations. The presence of weak fields inside network patches is more
636: surprising and had not been anticipated before.
637:
638: The reliability of our results is backed by several facts. Our 1C analysis
639: reproduces what had been obtained in the past from visible
640: spectro-polarimetric observations. When we extend the analysis to 2C we find
641: that the merit function $\chi^2$ is systematically smaller, even though we
642: did not add free parameters in the model. Therefore, our 2C analysis is at
643: least as reliable as comparable 1C studies which have been used thus
644: far. In order to ensure a conservative criterion for the occurrence of 2C,
645: our procedure was implemented with a ``preference'' for 1C solutions when
646: possible.
647: The spatial distributions of the filling factors $\alpha_w$ and $\alpha_s$ are
648: smooth, exhibiting spatial coherence even though each pixel has been inverted
649: separately. These
650: distributions, as well as the other results obtained from our study, are
651: consistent in the three maps analyzed (which have been obtained from
652: two different instruments). Several numerical tests (\S\ref{sec:reliability})
653: show that our method is able to distinguish between the filling
654: factors of weak and strong fields in the presence of unresolved PDFs and/or
655: line asymmetries. Finally, our results are sensible from a physical
656: point of view. For example, we find mixed strengths in internetwork locations
657: showing strong 1C fields but not in those showing weak 1C fields. Moreover,
658: the relative fractional area occupied by weak and strong fields is different
659: for upflows and downflows, with the downflows having a larger filling factor
660: of kG fields (in agreement with existing simulations).
661:
662: It is important to keep in mind that the mixed strengths that we have
663: detected may be just the tip of the iceberg. It is very likely that we only
664: see the most conspicuous ones. We have chosen to parameterize the sub-pixel
665: field distribution by two discrete values at 300 and 1700~G. However, we
666: know from both observations and simulations that quiet Sun fields obey a
667: continuous probability distribution, with the weaker fields covering a larger
668: fraction of the resolution element. Unfortunately, according to
669: \citeN{SN04b}, we can only infer (at most) two points of such distribution
670: using visible observations. Simultaneous visible and infrared observations
671: will offer a much more detailed picture. It might then be possible to infer
672: more than two points of the distribution.
673:
674: %\bibliographystyle{../bib/apj}
675: %\bibliography{../bib/aamnem99,../bib/articulos}
676:
677: \begin{thebibliography}{22}
678: \expandafter\ifx\csname natexlab\endcsname\relax\def\natexlab#1{#1}\fi
679:
680: \bibitem[{{Cattaneo}(1999)}]{C99}
681: {Cattaneo}, F. 1999, \apjl, 515, L39
682:
683: \bibitem[{{Dom{\'{\i}}nguez Cerde{\~ n}a}
684: {et~al.}(2003{\natexlab{a}}){Dom{\'{\i}}nguez Cerde{\~ n}a}, {Kneer}, \&
685: {S{\' a}nchez Almeida}}]{DCKSA03}
686: {Dom{\'{\i}}nguez Cerde{\~ n}a}, I., {Kneer}, F., \& {S{\' a}nchez Almeida}, J.
687: 2003{\natexlab{a}}, \apjl, 582, L55
688:
689: \bibitem[{{Dom{\'{\i}}nguez Cerde{\~ n}a}
690: {et~al.}(2003{\natexlab{b}}){Dom{\'{\i}}nguez Cerde{\~ n}a}, {S{\' a}nchez
691: Almeida}, \& {Kneer}}]{DCSAK03}
692: {Dom{\'{\i}}nguez Cerde{\~ n}a}, I., {S{\' a}nchez Almeida}, J., \& {Kneer}, F.
693: 2003{\natexlab{b}}, \aap, 407, 741
694:
695: \bibitem[{{Emonet} \& {Cattaneo}(2001)}]{EC01}
696: {Emonet}, T., \& {Cattaneo}, F. 2001, \apjl, 560, L197
697:
698: \bibitem[{{Grossmann-Doerth} {et~al.}(1996){Grossmann-Doerth}, {Keller}, \&
699: {Sch\" ussler}}]{GDKS96}
700: {Grossmann-Doerth}, U., {Keller}, C.~U., \& {Sch\" ussler}, M. 1996, \aap, 315,
701: 610
702:
703: \bibitem[{{Keller} {et~al.}(1994){Keller}, {Deubner}, {Egger}, {Fleck}, \&
704: {Povel}}]{KDE+94}
705: {Keller}, C.~U., {Deubner}, F.-L., {Egger}, U., {Fleck}, B., \& {Povel}, H.~P.
706: 1994, \aap, 286, 626
707:
708: \bibitem[{{Khomenko} {et~al.}(2003){Khomenko}, {Collados}, {Solanki}, {Lagg},
709: \& {Trujillo Bueno}}]{KCS+03}
710: {Khomenko}, E.~V., {Collados}, M., {Solanki}, S.~K., {Lagg}, A., \& {Trujillo
711: Bueno}, J. 2003, \aap, 408, 1115
712:
713: \bibitem[{{Lin}(1995)}]{L95}
714: {Lin}, H. 1995, \apj, 446, 421
715:
716: \bibitem[{{Lin} \& {Rimmele}(1999)}]{LR99}
717: {Lin}, H., \& {Rimmele}, T. 1999, \apj, 514, 448
718:
719: \bibitem[{{Lites}(1996)}]{L96}
720: {Lites}, B.~W. 1996, \solphys, 163, 223
721:
722: \bibitem[{{Lites}(2002)}]{L02}
723: ---. 2002, \apj, 573, 431
724:
725: \bibitem[{{Lites} \& {Socas-Navarro}(2004)}]{LSN04}
726: {Lites}, B.~W., \& {Socas-Navarro}, H. 2004, \apj, {\it in press}
727:
728: \bibitem[{{S\'anchez Almeida} {et~al.}(2003){S\'anchez Almeida}, {Emonet}, \&
729: {Cattaneo}}]{SAEC03}
730: {S\'anchez Almeida}, J., {Emonet}, T., \& {Cattaneo}, F. 2003, \apj, 585, 536
731:
732: \bibitem[{{S{\'a}nchez Almeida} \& {Lites}(2000)}]{SAL00}
733: {S{\'a}nchez Almeida}, J., \& {Lites}, B.~W. 2000, \apj, 532, 1215
734:
735: \bibitem[{{Sch\"ussler}(2003)}]{S03}
736: {Sch\"ussler}, M. 2003, in ASP Conf. Ser. (San Francisco: ASP), Vol. 307, Solar
737: Polarization Workshop 3, ed. J.~{Trujillo-Bueno} \& J.~{S\'anchez Almeida},
738: 601
739:
740: \bibitem[{{Socas-Navarro}(2001)}]{SN01a}
741: {Socas-Navarro}, H. 2001, in ASP Conf. Ser. 236: Advanced Solar Polarimetry --
742: Theory, Observation, and Instrumentation, 487
743:
744: \bibitem[{{Socas-Navarro}(2004{\natexlab{a}})}]{SN04a}
745: {Socas-Navarro}, H. 2004{\natexlab{a}}, \apj, {\it in press}
746:
747: \bibitem[{{Socas-Navarro}(2004{\natexlab{b}})}]{SN04b}
748: ---. 2004{\natexlab{b}}, \apj, {\it in press}
749:
750: \bibitem[{{Socas-Navarro} {et~al.}(2004){Socas-Navarro}, {Mart\' \i nez
751: Pillet}, \& Lites}]{SNMPL04}
752: {Socas-Navarro}, H., {Mart\' \i nez Pillet}, V., \& Lites, B.~W. 2004, \apj,
753: {\it in press}
754:
755: \bibitem[{{Socas-Navarro} \& {S{\' a}nchez Almeida}(2002)}]{SNSA02}
756: {Socas-Navarro}, H., \& {S{\' a}nchez Almeida}, J. 2002, \apj, 565, 1323
757:
758: \bibitem[{{Socas-Navarro} \& {S{\' a}nchez Almeida}(2003)}]{SNSA03}
759: ---. 2003, \apj, 593, 581
760:
761: \bibitem[{{Stein} \& {Nordlund}(2002)}]{SN02}
762: {Stein}, R.~F., \& {Nordlund}, A. 2002, in SOLMAG2002
763: IAU Colloquium 188, Sawaya-Lacoste (ed.), ESA SP-505, 83
764:
765: \bibitem[{{Trujillo Bueno} {et~al.}(2004){Trujillo Bueno}, {Schukina}, \&
766: {Asensio Ramos}}]{TBSAR04}
767: {Trujillo Bueno}, J., {Schukina}, N., \& {Asensio Ramos}, A. 2004, Nature, {\it
768: in press}
769:
770: \end{thebibliography}
771:
772: \end{document}
773:
774: