1: \documentclass[12pt,preprint]{aastex}
2:
3: %% manuscript produces a one-column, double-spaced document:
4:
5: %\documentclass[manuscript]{aastex}
6: %\citestyle{aa}
7: %% preprint2 produces a double-column, single-spaced document:
8:
9: %% \documentclass[preprint2]{aastex}
10:
11: %% \documentclass[preprint2,longabstract]{aastex}
12: \usepackage{color} % For creating coloured text and background
13:
14: % COLOR %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
15: \def\note#1{{\textcolor{green}{\bf [#1]}}} % note
16: \def\resp#1{{\textcolor{blue}{\bf [#1]}}} % response
17: \def\add#1{{\textcolor{red}{#1}}} % addition
18: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
19:
20:
21: \def \etal {et al.~}
22: \def \apj {ApJ}
23: \def \apjs {ApJS}
24: \def \apjl {ApJ}
25: \def \solphys {Solar Phys.}
26: \def \pasj {Pub. Astron. Soc. Japan}
27: \def \aap {A\&A}
28:
29: \begin{document}
30:
31: %
32: \title{Spectropolarimetric inversions of the \ion{Ca}{2} 8498 \AA\ and 8542 \AA\ lines in the quiet Sun}
33:
34: \author{A. Pietarila\altaffilmark{1}, H. Socas-Navarro}
35: \affil{High Altitude Observatory, National Center for Atmospheric Research\altaffilmark{2}, 3080 Center Green, Boulder, CO 80301, USA\vbox{}}
36: \author{T. Bogdan}
37: \affil{Space Environment Center, National Oceanic and Atmospheric Administration, 325 Broadway, Boulder, CO 80305, USA\vbox{}}
38:
39:
40: \altaffiltext{1}{Institute of Theoretical Astrophysics, University of Oslo, P.O.Box 1029 Blindern, N-0315 Oslo, Norway}
41: \altaffiltext{2}{The National Center for Atmospheric Research (NCAR) is sponsored by the National Science Foundation.}
42:
43:
44:
45: \begin{abstract}
46: We study non-LTE inversions of the Ca II infrared triplet lines as a tool for inferring physical properties of
47: the quiet Sun. The inversion code is successful in recovering the
48: temperature, velocity and longitudinal magnetic flux density in the
49: photosphere and chromosphere, but the height range where the
50: inversions are sensitive is limited, especially in the
51: chromosphere. We present results of inverting spectropolarimetric observations of the lines in a quiet Sun region. We find
52: three distinct ranges in chromospheric temperature: low
53: temperatures in the internetwork, high temperatures in the enhanced magnetic
54: network and intermediate temperatures associated with low magnetic
55: flux regions in the network. The differences between these regions become more pronounced with height as the plasma-$\beta$ decreases. These inversions support the picture of the chromosphere, especially close to the magnetic network, being highly inhomogeneous both in the vertical and horizontal directions.
56:
57:
58: \end{abstract}
59: \keywords{polarization, Sun: chromosphere, magnetic fields, waves}
60:
61: \section{Introduction}
62:
63: A transition from a high plasma-$\beta$ ($\beta=p_{gas}/p_{mag}$) to a
64: low-$\beta$ regime takes place in the chromosphere. With this
65: transition the nature of the dynamics
66: changes: in the high-$\beta$ regime the plasma dominates the magnetic fields whereas in the low-$\beta$ regime the opposite is true. Waves undergo mode mixing, reflection and refraction in the layer where plasma-$\beta \approx 1$. Because of the tight connection between the magnetic fields and dynamics it is crucial
67: to understand how the two are coupled in the quiet Sun chromosphere. Several
68: recent observational works have focused on this
69: (e.g., \citealt{Judge+others2001, McIntosh+Judge2001, McIntosh+Fleck+Judge2003, Jefferies+others2006} and \citealt{Vecchio+others2007}). All find a clear correlation between the
70: magnetic field topology and waves in the chromosphere. None of the
71: above mentioned works, however, directly addresses the magnetic field
72: in the chromosphere. Instead, proxies such as potential field
73: extrapolation or the existence of fibril-like structures in the
74: emission of chromospheric lines, are used.
75:
76: In order to obtain physical parameters (instead of proxies) from spectropolarimetric
77: observations an inversion problem is
78: unavoidable. Since the radiative transfer equation for polarized light
79: is a set of coupled differential equations, and in the chromosphere
80: local thermodynamical equilibrium (LTE) is no longer valid, solving the
81: inverse problem is challenging. Furthermore, one is never certain that any
82: solution obtained is in fact the correct solution, instead we have
83: to rely on some {\it a priori} knowledge of the solar
84: atmosphere. Inversions are commonly used to to produce quantitative information of the photosphere. They are not yet as widely used for chromospheric observations. Significant progress in chromospheric inversions has been made for the
85: \ion{He}{1} 10830 \AA\ multiplet (e.g., \citealt{Lagg+others2004,
86: Centeno+others2005, TrujilloBueno+others2005, Sasso+Lagg+Solanki2006,
87: Merenda+others2006}) and \ion{Ca}{2} infrared (IR) triplet lines
88: (e.g., \citealt{Socas-Navarro+others2000b, Socas-Navarro2005a,
89: Socas-Navarro+others2006b}).
90:
91: The magnetic structure of the quiet Sun chromosphere is still a matter
92: of debate, but there is no doubt that it plays an important role. The leakage of
93: low-frequency (i.e., below acoustic cut-off frequency) waves into the
94: chromosphere and the formation of spicules have been attributed
95: to the existence of inclined magnetic fields in the network
96: (e.g., \citealt{DePontieu+others2004, Jefferies+others2006,
97: Hansteen+others2006, DePontieu+others2007}). It is also suggested that
98: these waves may be related to the heating of the chromosphere
99: (\citealt{Jefferies+others2006}). Existence of canopy-like structures such as proposed by
100: \cite{Gabriel1976} is still an open debate. For example \cite{Vecchio+others2007} observe oscillatory patterns in the
101: chromosphere that are highly consistent with the canopy-scenario, while
102: models by e.g., \cite{Schrivjer+Title2003} and
103: \cite{Jendersie+Peter2006} show that the existence of a canopy-like
104: structure is disrupted by internetwork magnetic fields. The connection between magnetic fields, dynamics and thermal properties of the quiet Sun chromosphere appears complex. In this paper we extend the use of non-LTE inversions to the quiet Sun chromosphere. With non-LTE inversions we do not have to rely on proxies, such as line intensity, to infer physical quantities.
105:
106: \nocite{Pietarila+others2006}
107: \nocite{Pietarila+others2007a}
108:
109: This paper is a continuation to Pietarila et al.\ (2006, hereafter P06) and Pietarila et al.\ (2007, hereafter P07) in which the use of \ion{Ca}{2} IR triplet lines as a diagnostic for chromospheric magnetism are studied. Here we focus on non-LTE inversions of the \ion{Ca}{2} IR
110: triplet lines, and what information can be extracted from them. In
111: section \ref{sec:code} the non-LTE inversion code used in this work is
112: described in some detail. Also the
113: sensitivity of the inversions is discussed. We demonstrate that the Ca lines
114: are sensitive only in a narrow height range in the chromosphere, and
115: that our inversion code cannot recover discontinuities in magnetic field inclination
116: or velocity (at least with the observables used). We also show that discontinuities can cause
117: complex patterns in Ca line Stokes $V$ asymmetries. Inversions of
118: quiet Sun observations are discussed in section \ref{sec:obs}. The
119: network is found to be more dynamic than the internetwork. Three different
120: regimes are found in chromospheric temperature: low temperatures
121: corresponding to the ``magnetic field free '' internetwork, high
122: temperatures in the enhanced magnetic network and an intermediate temperature
123: regime associated with low magnetic flux regions. The main
124: results are discussed (in the context of previous work) in section \ref{sec:disc}, and final remarks are made in section \ref{sec:conc}.
125:
126: \section{Inversion code}
127:
128: The non-LTE inversion code employed here has been described in detail in \cite{Socas-Navarro+RuizCobo+TrujilloBueno1998} and \cite{Socas-Navarro+TrujilloBueno+RuizCobo2000}. For completeness, we provide here a brief summary of its main features. The user needs to supply an initial model atmosphere, which consists of the run with standard optical depth ($\tau_{500}$) of temperature, microturbulence, line-of-sight velocity and magnetic field vector (strength, inclination and azimuth). In addition to these vector quantities, a model atmosphere includes the following single-valued parameters: electron pressure at the top of the atmosphere, macroturbulence and filling factor. The macroturbulence is used also to describe the spectral instrument profile. The filling factor is used to consider that the magnetic structure is smaller than the resolution element (or that there is some stray light in the spectrograph).
129:
130: With the top boundary condition for the electron pressure and the run of temperature, the hydrostatic equilibrium equation is solved to obtain the run of electron pressure, gas pressure and density and the geometrical scale height. The code then uses a preconditioning strategy (\citealt{Rybicki+Hummer1991, Socas-Navarro+TrujilloBueno1997}) to solve the statistical equilibrium equations in that atmosphere and synthesizes the Stokes profiles emerging from the initial model as well as the response functions, i.e. the derivatives of the Stokes vector at a given wavelength with respect to the model parameters.
131:
132: The derivatives enter a Levenberg-Marquardt scheme that determines the first-order correction needed to compensate for the misfit between the observed and the synthetic profiles. The correction for a given depth-dependent parameter (e.g., temperature) is obtained as a smooth function of $\tau_{500}$, depending on the number of {\it nodes} selected by the user for that parameter (for two nodes the code retrieves a correction that is linear with $\tau_{500}$, for three nodes a parabola, and for four or more nodes the correction is a cubic spline interpolation). The entire algorithm is iterated until it converges to a solution that minimizes the difference between the observed and synthetic profiles.
133:
134: \label{sec:code}
135: \subsection{Sensitivity of the inversion code}
136: \label{sec:tests}
137:
138: % and velocity nodes}
139: \label{sec:beta}
140:
141: The inversion code retrieves model atmospheres with the atmospheric
142: parameters as a function of optical depth, ranging from
143: $log(\tau)=[-6$, $0.8]$. The sensitivity of the spectral lines,
144: however, is limited to only a range of atmospheric depths. To identify
145: what the depths are we made the following numerical experiment: a set of
146: simultaneous Stokes $I$ and $V$ profiles of the 8498 \AA\ and 8542
147: \AA\ \ion{Ca}{2} lines and the two photospheric \ion{Fe}{1} lines,
148: 8497 \AA\ and 8538 \AA\ (located in the wings of the Ca lines), were
149: inverted using 100 different initializations. The initial models were
150: chosen randomly from existing inversions of observations. To make
151: the test comparable to the observations discussed later, we do not
152: include Stokes $Q$ or $U$ profiles. Additionally, we force the magnetic field to be vertical, i.e., only the longitudinal magnetic field is considered. The lines are, as expected, sensitive to
153: only part of the atmosphere (fig. \ref{fig:sensitivity}). The spread in the output models is smaller in two regions: in the photosphere, where the Fe lines and the Ca line wings are
154: formed, and in the chromosphere, where the Ca line cores are formed. For
155: temperature the heights are $log(\tau) \approx 0$ and $log(\tau) \approx
156: -5$ , for magnetic flux $log(\tau) \approx -1$ and $log(\tau) \approx -5$
157: , and for velocity $log(\tau) \approx -1$ and $log(\tau) \approx -5.5$. Since
158: neither the Ca nor the Fe lines are formed in the region between
159: $log(\tau) \approx [-4, -2]$ the output models have a large spread
160: at these heights. We conclude that the inversion results at this
161: height range are not reliable, and that the retrieved atmospheric
162: values are reliable only in the fairly narrow height ranges.
163:
164:
165: In figure \ref{fig:sensitivity} are also shown the retrieved magnetic
166: field strengths with nodes in both the field strength (3 nodes) and
167: inclination (4 nodes). The spread of field strengths obtained is unreasonably
168: large, especially in the chromosphere. Since the inversion code has no
169: information of Stokes $Q$ and $U$, some ambiguity between the magnetic
170: field strength and inclination cannot be removed. Due to this, for
171: the observations we choose to force the inclination to zero and
172: consider only the longitudinal magnetic flux density which is better
173: constrained.
174:
175: \begin{figure*}
176: \epsscale{.8}
177: \plotone{f1.eps}
178: \caption{
179: \label{fig:sensitivity}
180: Sensitivity of the lines in the inversions. Upper row: input Stokes $I$ and $V$ profiles are in red and fits to the profiles in black lines. Lower row: Temperature (left), longitudinal magnetic flux (middle) and velocity (right) as a function of optical depth. Black points show results for models with variation in magnetic field strength only, orange points for models with variation in both inclination and magnetic field strength. Blue and orange curves show the mean of the black and orange points, respectively, and the error bars show the standard deviation.}
181: \end{figure*}
182:
183:
184:
185:
186: \subsection{Velocity and inclination discontinuities}
187:
188:
189: Gradients and/or discontinuities in velocity and magnetic field cause
190: asymmetries in Stokes $V$ profiles. The high-$\beta$ simulation (P06)
191: demonstrated how velocity gradients from shocking acoustic waves cause
192: a pattern of time-varying Stokes $V$ asymmetries in the Ca lines.
193:
194: To study the effect of discontinuities we use the inversion code in
195: synthesis mode to produce a set of Stokes profiles from atmospheres
196: with velocity discontinuities of different sizes (1, 3 and 5 km/s) and
197: at different locations (between log$(\tau)=[-5.5, 0]$. We then compute Stokes $V$ asymmetries for the
198: profiles. The asymmetries are defined as in \cite{MartinezPillet+others1997}. The amplitude asymmetry of a Stokes V
199: profile is given by
200:
201: \begin{eqnarray*}
202: \sigma_{\mathrm{a}}= \frac {a_b-a_r}{a_b+a_r},
203: \end{eqnarray*}
204: where $a_b$ and $a_r$ are the unsigned extrema of the blue and red
205: lobes of the Stokes $V$ profile.
206:
207:
208: The area asymmetry of a Stokes $V$ profile is given by:
209:
210: \begin{eqnarray*}
211: \label{eqn:area}
212: \sigma_{\mathrm{A}}=s\frac{\int_{\lambda_0}^{\lambda_1} V(\lambda)d\lambda}{\int_{\lambda_0}^{\lambda_1} |V(\lambda)|d\lambda},
213: \end{eqnarray*}
214: where $s$ is the sign of the blue lobe.
215:
216:
217: \begin{figure*}
218: \epsscale{.8}
219: \plotone{f2.eps}
220: \caption{
221: \label{fig:vel-discont1}
222: Asymmetries in Stokes $V$ profiles caused by discontinuities in velocity. Upper row: Stokes V amplitude asymmetry in 8498 \AA\ (left) and 8542 \AA\ (right) as a function of the location of the discontinuity. Green, blue and red solid lines are discontinuities of 1, 3 and 5 km/s, respectively. Dotted and dashed lines are asymmetries from inversion fits for 3 and 4 velocity nodes, respectively. Color coding is the same as for the solid lines. Where no inversion value is plotted, no satisfactory fit was found. Lower row: As upper row except for Stokes V area asymmetry. }
223: \end{figure*}
224:
225:
226: In figure ~\ref{fig:vel-discont1} the results are shown for the velocity discontinuities. For the sign of the velocity we use the standard
227: astrophysical notation where positive values indicate red-shifts. The sign and amplitude of the Stokes $V$ asymmetries depend both on
228: the location and size of the discontinuity. Note that both positive
229: and negative asymmetries are present even though all velocities are
230: positive and in all cases the
231: photosphere has a smaller velocity. The overall patterns in the 8498 \AA\ and 8542 \AA\ lines
232: are not identical, or even all that similar.
233:
234:
235: To see if the inversion code can handle velocity discontinuities we
236: invert the profiles (Stokes $I$ and $V$ of the two Ca lines and the
237: two photospheric Fe lines) using either 3 or 4 nodes in
238: velocity. Asymmetries computed from the inversion fits are shown in
239: dashed and dotted lines in figure \ref{fig:vel-discont1}. In most
240: cases the fits do not reproduce the asymmetries. However, the overall
241: qualitative shape of the line profiles is in most cases
242: satisfactory. Increasing the number of velocity nodes to 10 does not significantly improve the results.
243:
244: Also discontinuities in inclination cause complex asymmetry patterns in the lines. The effects of a discontinuity on Stokes
245: $V$ asymmetries are not easily predictable since they depend strongly
246: on both the size and location of the discontinuity. Furthermore, the
247: discontinuities can cause the two Ca lines, with relatively close
248: formation heights and similar radiative transfer, to exhibit very
249: different asymmetries. Also the area and amplitude asymmetries can be
250: of opposite signs. A combination of discontinuities or strong
251: gradients in velocity and inclination will be even more
252: difficult to disentangle.
253:
254:
255:
256: %\begin{figure*}
257: %\epsscale{.8}
258: %\plotone{plots/incl_discont2.eps}
259: %\caption{
260: %\label{fig:incl-discont2}
261: %Upper row: Scatter plots of actual and inverted location (left) and size (right) the inclination discontinuities for the cases with nodes in inclination illustrated in ~\ref{fig:incl-discont1}. Key for symbols shown in the figure legends. Middle row: $\chi^2$ of the fits versus the location (left) and size (right) of the discontinuity. Key for symbols locate din figure legends. Lower row: Left panel: Location of inclination discontinuity vs. location of largest velocity gradient in inversion models. Right panel: Size of inclination discontinuity vs. size of largest velocity gradient in inversion models.}
262: %\end{figure*}
263:
264:
265:
266:
267:
268: \section{Inversions of quiet Sun observations}
269: \label{sec:obs}
270: We use the non-LTE inversion code to invert quiet Sun observations of the
271: \ion{Ca}{2} 8498 \AA\ and 8542 \AA\ lines and \ion{Fe}{1} 8497 \AA\
272: and 8538 \AA\ lines taken with the the Spectro-Polarimeter for
273: INfrared and Optical Regions ({\em SPINOR},
274: \citealt{Socas-Navarro+others2006a}) at the Dunn Solar telescope,
275: Sacramento Peak Observatory. The raster scan was obtained in a quiet Sun
276: region near disk center at S17.3 W32.1 on May 19, 2005 at 14:14
277: UT. The upper part of the slit was positioned above internetwork and lower part above an enhanced network patch consisting of a
278: single polarity. For a detailed description of the
279: observations see P07.
280:
281: The inversion is performed in 2 cycles. In the first cycle both Stokes
282: $I$ and $V$ are weighted equally and there are 4, 1 and 3 nodes in
283: temperature, magnetic field strength and velocity, respectively. In
284: the second cycle more weight (factor 500) is put on the Stokes $V$
285: profile and the number of nodes are increased to 7, 3 and 4
286: (temperature, magnetic field strength and velocity). A node for
287: macroturbulence is also added. The second cycle allows the code to fine-tune the solution
288: obtained in cycle 1. In both cycles the magnetic field inclination and
289: azimuth are forced to zero. 5 different initial models are used for
290: each inversion and the best fit is chosen based on the smallest
291: $\chi^2$ value.
292:
293:
294: A 2-component atmosphere is needed to fit most of the line profiles satisfactorily. The filling factor is a free
295: parameter and one of the atmospheric components is a non-magnetic external atmosphere which is the same for all pixels. The non-magnetic external atmosphere is obtained by inverting an average non-magnetic
296: profile using the same number of nodes in velocity and temperature as
297: in the main inversions, but with no nodes in magnetic field and a zero
298: magnetic field in the initial models. The filling factor is forced to
299: be unity. The
300: non-magnetic profile is an average of intensity profiles in all pixels
301: where the 8498 \AA\ Stokes $V$ amplitude is below
302: $7\times10^{-3}$. Because of atomic polarization, Stokes $Q$ and $U$
303: are given no weight in the inversions. Consequently, we do not have
304: enough information to resolve both atmospheric components, hence the
305: non-magnetic atmosphere is set to be constant.
306:
307: Since the telluric lines present are weak, they
308: cannot be used for a precise absolute wavelength calibration. Instead,
309: the velocities are calibrated so that the mean photospheric velocity
310: in ``non-magnetic'' regions (pixels where the photospheric magnetic field strength is
311: less than 300 G) is
312: set to zero.
313:
314: \subsection{Goodness of fits}
315:
316: $\chi^2$ was used as a proxy for the goodness of the fits produced by
317: the inversions (figure \ref{fig:chi}). In general, the $\chi^2$-values are below 0.2. Figure \ref{fig:proxy} illustrates how well the inversions retrieve
318: quantities used to describe the line profiles. The inversion code can
319: retrieve the amplitudes of both Stokes $I$ (integrated intensity in a
320: 0.75 \AA\ wide wavelength band around the line core zero wavelength
321: normalized to the band width, $I_c/$\AA) and $V$ very well, but the
322: Stokes $V$ asymmetries are not reproduced. The 8498 \AA\
323: area asymmetry is especially problematic. Since the two Ca lines have very
324: different statistics in terms of the area asymmetries (P07) and, as shown in the previous section, discontinuities in
325: velocity and/or inclination cause complex asymmetry patterns, the
326: failure is not all that surprising. The only quantity that shows a
327: correlation with $\chi^2$ is Stokes $V$ amplitude. This may be
328: explained by it being easier to fit only the intensity profiles instead of fitting both the
329: Stokes $I$ and $V$ profiles. Also it is possible, if not likely, that
330: the magnetic elements are sub-resolution and the pixel consists of a
331: magnetic component and an external non-magnetic component. It is
332: uncertain how well the average external non-magnetic atmosphere describes the
333: non-magnetic component. No clear correlation is found between $\chi^2$
334: and Stokes $V$ asymmetries.
335:
336:
337:
338: \begin{figure*}
339: \epsscale{1}
340: \plotone{f3.eps}
341: \caption{
342: \label{fig:chi}
343: $\chi^2$ of the inversion fits. right: image of the spatial distribution of $log_{10}(\chi^2)$. The black strip covers the hairline and the close-by pixels that are affected by it. Left: histogram of the $\chi^2$ values. }
344: \end{figure*}
345:
346:
347: \begin{figure*}
348: \epsscale{0.8}
349: \plotone{f4.eps}
350: \caption{
351: \label{fig:proxy}
352: Scatter plots of the observed and fitted 8498 \AA\ and 8542 \AA\ Stokes $I$ and $V$ amplitudes, and Stokes $V$ area and amplitude asymmetries. The over-plotted diamonds and asterisks show the mean for 8498 \AA\ and 8542 \AA\ lines, respectively. In the bottom right corners of the figures are cross correlation coefficients for the observed and fitted values. Darker gray is 8498 \AA\ and lighter gray 8542 \AA. }
353: \end{figure*}
354:
355: \subsection{Spatial patterns}
356:
357: Since the inversions have the filling factor as a free parameter, the
358: results need to be weighted in respect to how much of the light from
359: the pixel originates from the magnetic component and how much from the
360: external non-magnetic component. Since this atmosphere is assumed the
361: same for all pixels and only the filling factor varies, for the
362: following discussion, only pixels
363: with a filling factor larger than 15 \% are considered without further
364: restrictions. For pixels with filling factors less than 15 \%, only
365: those with 8498 \AA\ Stokes $V$ amplitude above 0.02 $I_c$ are
366: included. The Stokes $V$ amplitude threshold was chosen to be the
367: amplitude above which the standard deviation of $\chi^2$ in models producing good
368: fits in a given pixel is the same for pixels with filling factors less
369: than 15\% and pixels with filling factors greater than 15 \%. For pixels with filling factors less than
370: 15 \% and 8498 \AA\ Stokes $V$ amplitude below 0.02, the standard
371: deviation of the inverted atmospheres is significantly larger, and
372: they are omitted from the analysis.
373:
374: Note that the filling factors are largest in the network and the very quiet internetwork whereas the
375: areas surrounding the network have very small filling factors (figure
376: \ref{fig:alfas}). This illustrates two points: the network is best
377: described with a 2-component scenario. Second, since the internetwork
378: has no signal in Stokes $V$ the code can combine the non-magnetic
379: profile with a significant contribution from an additional
380: (essentially also non-magnetic) atmospheric component resulting in a
381: 2-component non-magnetic atmosphere. In the regions by the network,
382: the Stokes $V$ amplitude is in general small, but not negligible. Since the Stokes $V$ signal is weighted more by the inversion
383: code in
384: such cases the non-magnetic component is described solely by the
385: external non-magnetic atmosphere even though it is the dominant
386: component. Since there is little signal in Stokes $V$ in these pixels, consequently also the
387: filling factors will be small.
388:
389: \begin{figure*}
390: \epsscale{1}
391: \plotone{f5.eps}
392: \caption{
393: \label{fig:alfas}
394: Right: filling factors for the inversions. The contours are for 75 G photospheric longitudinal magnetic flux. Left: mask based on chromospheric temperature used to define the enhanced network, low-flux network and internetwork. }
395: \end{figure*}
396:
397: The photospheric longitudinal magnetic flux density (at log$(\tau)$=-1, hereafter referred to as the magnetic flux, panel {\it c)} in figure \ref{fig:atmos-image}) has a
398: fragmented pattern with distinct edges. The strongest flux, about 1200
399: Mx/cm$^2$, is in the centers of flux ``bundles''. There are no
400: isolated points of strong flux. This is due to both the photospheric
401: lines being weak and thus not very sensitive to magnetic fields, and
402: to the spatial resolution of the observations. Since individual flux
403: tube size is below the spatial resolution, only regions with a concentration of several flux
404: tubes appear strong. The strongest flux in the chromosphere
405: (at log$(\tau)$=-5, panel {\it f)} in figure \ref{fig:atmos-image}) does not lie directly above the photospheric
406: flux concentrations, and not all of the strong flux seen in the
407: photosphere reaches the chromosphere. In fact, the strongest flux
408: concentrations are significantly weaker higher up, and the total area
409: covered by strong flux is smaller. The magnetic network is more
410: confined in the photosphere. In the chromosphere the field
411: has expanded to fill more space causing the more diffuse appearance of the network.
412:
413: The photospheric temperature (at log$(\tau)$=0, panel {\it a)} in figure \ref{fig:atmos-image}) is significantly
414: cooler in the strong flux concentrations. To maintain a pressure
415: balance between the interior and exterior of the flux tube, the gas
416: pressure inside the tube must be smaller than outside leading to a reduced density inside the flux tube. Also the convective energy transport in the tube is suppressed by the magnetic field. If the diameter of the flux tube (or concentration of tubes) is large enough the lateral radiative flux from the hotter surroundings does not reach the tube center, and it remains cooler (and appears as a dark feature in intensity images). Hints of a granulation pattern are visible in
417: photospheric temperature: the cell centers are
418: hotter than the edges. The network is well visible in the
419: chromospheric temperature (at log$(\tau)$=-5, panel {\it d)} in figure \ref{fig:atmos-image}), and is clearly hotter than the
420: internetwork. The locations of the hottest pixels are not correlated
421: with the strongest flux. Instead the hot pixels form a fragmented
422: pattern that covers most of the region where flux is seen. The network
423: boundaries are visible as a region of intermediate temperature,
424: clearly distinct from both the network and internetwork. The
425: intermediate temperature region extends to form a network like pattern
426: in the mostly flux-free region (hence hereafter this region
427: will be referred to as low-flux network). There is no clear indication
428: of the low-flux network in photospheric temperature or
429: magnetic flux.
430:
431: Since no absolute wavelength calibration was done, defining the zero velocity is not exact and velocities close to zero
432: should not be taken as definite measures. In the photospheric
433: internetwork patches of negative and positive velocities are
434: present (at log$(\tau)$=-1, panel {\it b)} in figure \ref{fig:atmos-image}). The network is dominated
435: by negative velocities. Some positive velocities are also seen, though
436: they are located closer to the edges and are not as large in amplitude
437: as in the internetwork. There are large scale
438: patterns of positive velocities, interlaced with significantly smaller
439: patches of negative velocities, in the chromospheric internetwork (at log$(\tau)$=-5.5, panel {\it e)} in figure \ref{fig:atmos-image}). The chromospheric network has
440: mostly positive velocities, though above one of the strong flux
441: concentrations a patch with a large negative velocity is seen. The
442: overall velocity pattern in the chromosphere is more fragmented in the network, and also the velocity amplitudes vary less than in the internetwork. This may be due to the underlying magnetic field
443: ``compartmentalizing'' the plasma. There is little correlation between photospheric and chromospheric velocities.
444:
445:
446: \begin{figure*}
447: \epsscale{1}
448: \plotone{f6.eps}
449: \caption{
450: \label{fig:atmos-image}
451: {\it a)} Photospheric temperature (at log($(\tau)$)=0). {\it b)} Photospheric velocity (at log($(\tau)$)=-1). {\it c)} Photospheric longitudinal magnetic flux. (at log($(\tau)$)=-1) {\it d)-f)} as {\it a)-c)} but for the chromosphere (at log($(\tau)$)=-5, -5.5 and -5). For explanation of choice of pixels, see text. Numbers indicate locations of pixels discussed in more detail in section \ref{sec:ind}.}
452: \end{figure*}
453:
454:
455:
456: \subsection{Histograms of atmospheres}
457:
458: In figure \ref{fig:hists} are shown histograms of the
459: photospheric and chromospheric temperature, magnetic flux and
460: velocity. The photospheric temperature histogram is
461: narrower than the chromospheric one. It has a single peak at 6400
462: K, which is also the temperature in the photosphere of the external
463: non-magnetic atmosphere. The chromospheric temperature histogram has
464: 2 peaks: the first one is at 5800 K, and corresponds to the low-flux
465: network. The coolest pixels, below 5400 K, are in the
466: internetwork. The second peak, corresponding to the hot network
467: pixels, at 6400 K, is narrower but has an extended tail to high
468: temperatures. This is higher than in the external non-magnetic
469: atmosphere (6260 K).
470:
471: The photospheric velocity histogram is centered at 0 km/s,
472: which is also by definition the photospheric velocity in the external
473: non-magnetic atmosphere and in pixels with a photospheric magnetic flux less than 300 G. The chromospheric velocity histogram peaks
474: at a positive values: 0.5 km/s. It is significantly wider than the photospheric histogram.
475:
476: The longitudinal magnetic flux in both the photosphere and chromosphere has
477: exponentially decaying histograms. The flux in the photosphere decays
478: slower than in the chromosphere. Neither histogram has peaks
479: corresponding to the network flux.
480:
481: \begin{figure*}
482: \epsscale{1}
483: \plotone{f7.eps}
484: \caption{
485: \label{fig:hists}
486: Left: Histograms of photospheric (dark gray line) and chromospheric (light gray line) temperature (left), velocity (middle) and longitudinal magnetic flux (right).}
487: \end{figure*}
488:
489:
490: There are three distinct regions in the chromospheric temperature: the enhanced
491: network, low-flux network and internetwork. We define a
492: mask based on the temperature: the
493: internetwork is where the chromospheric temperature is below 5400 K, low-flux network where the temperature
494: is between 5400 K and 6200 K and network where the temperature is
495: above 6200 K. The mask clearly separates the map into three distinct
496: regions (shown in figures \ref{fig:alfas} and \ref{fig:nw-inw} where the three colors mark the three regions). Histograms
497: illustrate similarities and differences between the three regions and the line
498: profiles in them. The photospheric
499: temperature distributions are nearly identical in all three
500: regions: the network, except for the cool strong flux concentrations,
501: is not visible in photospheric temperature.
502:
503:
504:
505: The photospheric velocity histograms are centered around zero or close to
506: zero. The chromospheric histograms peak at positive values with the internetwork having the largest red-shift: it peaks at 1 km/s. The enhanced network regions have the narrowest velocity histograms both in the photosphere and in the chromosphere. This is consistent with the magnetic field suppressing convection. However, the bottoms of the histograms in the enhanced network are roughly as wide as in the internetwork histograms. There is no significant difference between the internetwork and low-flux network histograms in the photosphere. In the chromosphere, however, the low-flux network histogram is significantly wider.
507:
508: As expected, the magnetic flux is by far largest in the network. No
509: flux is seen in the internetwork. (Since the photospheric lines are
510: weak and the seeing conditions during the observing sequence were not
511: optimal, no internetwork fields are seen.) The low-flux network
512: has some, though very little when
513: compared to the network, magnetic flux.
514:
515: The network is clearly brightest
516: of the three regions when viewed in continuum intensity. No difference is seen between the internetwork and low-flux
517: network. In the Ca line intensities the low-flux network is somewhat brighter than the
518: internetwork, but still clearly darker than the network. The Stokes
519: $V$ amplitudes of the Ca lines further illustrate that the
520: low-flux network is magnetic: the amplitudes are larger in
521: the low-flux region than in the internetwork. In general, the differences
522: between the three regions, especially between the internetwork and
523: low-flux network, become more pronounced with height.
524:
525: \begin{figure*}
526: \epsscale{0.9}
527: \plotone{f8.eps}
528: \caption{
529: \label{fig:nw-inw}
530: Histograms of photospheric and chromospheric temperatures, velocities and magnetic flux, continuum intensities and Ca line amplitudes in the internetwork, low-flux network and network. Image shows the mask used to identify the pixels: dark gray is enhanced network, light gray low-flux network and white internetwork. The black regions are pixels not included in the analysis.}
531: \end{figure*}
532:
533:
534: \begin{figure*}
535: \epsscale{1}
536: \plotone{f9.eps}
537: \caption{
538: \label{fig:i-vs-flux}
539: 8498 \AA\ and 8542 \AA\ Stokes $I$ amplitudes versus chromospheric temperature. The network, low-flux network and internetwork are shown in different shades of gray: the dark is the internetwork, medium dark the low-flux network and light the enhanced network. }
540: \end{figure*}
541:
542: The dependence of the Ca lines' Stokes $I$ amplitudes on the
543: chromospheric temperature is shown in figure \ref{fig:i-vs-flux}. Both
544: lines show a saturation in intensity at high temperatures, and a
545: minimum level intensity at low temperatures. The intensities show a
546: strong dependence on the temperature only in the intermediate temperature
547: range. Implications on the use of proxies based on spectral lines formed in non-LTE are discussed in section
548: \ref{sec:disc}. Scatter plots of chromospheric temperature and
549: longitudinal magnetic flux, both photospheric and chromospheric,
550: show some signs of saturation at high fluxes, but the spread is large
551: and the number of data points with large magnetic flux is not high
552: enough for the saturation to be determined as statistically significant.
553:
554:
555:
556:
557: \subsection{Individual atmospheres}
558: \label{sec:ind}
559:
560:
561: In the following, individual atmospheres representative of different
562: regions seen in the map are discussed. Locations of the regions are marked in figure \ref{fig:atmos-image}.
563:
564: Line profiles and atmospheres for pixels belonging to area 1 are shown in
565: figure \ref{fig:a1}. Area 1 is located in a strong photospheric flux-bundle. It is also clearly visible in photospheric
566: temperature. The line profiles are all fairly symmetric and similar to
567: one another (except for one asymmetric 8498
568: \AA\ Stokes $V$ profile. It corresponds to the atmosphere that deviates most from the mean
569: in both the magnetic field and velocity). The average filling factor of the pixels
570: is 0.66. The Stokes $V$ profiles of
571: the photospheric iron lines are identical to one another indicating that the photospheric magnetic field is
572: fairly homogeneous in area 1. Photospheric temperature in the atmosphere is ~430 K cooler than in the external
573: non-magnetic photosphere. The chromosphere is significantly
574: hotter, by 1000 K. The
575: longitudinal magnetic flux
576: is of the order of 1 kG in the photosphere and decays by a factor of 5
577: before reaching the chromosphere. This is already seen in the images of magnetic flux where the strong
578: photospheric flux concentrations are not visible in the
579: chromosphere. The velocity is zero in the photosphere and positive in the chromosphere.
580:
581:
582: \begin{figure*}
583: \epsscale{0.8}
584: \plotone{f10.eps}
585: \caption{Upper row Stokes $I$ and $V$ profiles for area 1 (for location see figure \ref{fig:atmos-image}). Observed profiles and inversion fits are shown in black and red lines, respectively. Lower row: Temperature (left), longitudinal magnetic flux (middle) and velocity (right) as a function of optical depth. The dashed red lines in the temperature and velocity panels are for the external non-magnetic atmosphere. The vertical dotted lines show where the inversion code results are most reliable.
586: \label{fig:a1}
587: }
588: \end{figure*}
589:
590:
591: Area 2 (figure \ref{fig:a2}) is located deep in the internetwork. The
592: Stokes $I$ profiles are all nearly identical: deep, narrow and very
593: symmetric. There is no signal above noise in Stokes $V$, and
594: consequently, the filling factor (0.51) only describes how much of the
595: average non-magnetic profile is present in the line profiles. The temperature
596: in the photosphere is nearly identical to the external non-magnetic
597: atmosphere, but the chromosphere is significantly cooler. Because the pixels for the non-magnetic profile were chosen based on
598: magnetic signal and not on intensity, pixels in the low-flux
599: network (which are hotter than the internetwork) were included. The inversion code returns a flux that is less
600: than 20 Mx/cm$^2$, and the fitted Stokes $V$ profiles are well below
601: noise level. The velocity profile differs from the non-magnetic both
602: in the photosphere and chromosphere. This is expected since the non-magnetic profile is an average of many pixels with
603: different phases, i.e., all but net shifts should be eliminated from
604: it. In general, there is very little scatter between the model
605: atmospheres indicating that the region is fairly homogeneous, both in
606: terms of thermal and dynamic properties.
607:
608:
609:
610: \begin{figure*}
611: \epsscale{0.8}
612: \plotone{f11.eps}
613: \caption{As figure \ref{fig:a1} but for area 2.
614: \label{fig:a2}
615: }
616: \end{figure*}
617:
618: Pixels in area 3 (figure \ref{fig:a3}) are located in the network, but outside
619: of the strong flux concentrations. The Stokes $I$ and $V$ profiles are
620: fairly symmetric, but only the 8542 \AA\ line has nearly identical Stokes
621: $V$ profiles in all pixels. The other 3 lines show two sets of Stokes
622: $V$ amplitudes: one set significantly larger than the other. This is
623: especially clear in the photospheric iron lines. The 8498 \AA\ line
624: has a set of Stokes V profiles with strong signal in the line
625: wings and a set with little signal in the wings and also a
626: slightly smaller amplitude. The temperature in the photosphere is roughly the same as in the external non-magnetic
627: atmosphere: the coolest photosphere in area 3 is only 50 K cooler. This is a very small difference when
628: compared to area 1 where the Stokes $V$ profiles had similar shapes
629: (albeit somewhat larger amplitudes) as here. In contrast to area 1 where the
630: chromosphere is significantly hotter than the external non-magnetic
631: atmosphere, in area 3 the chromosphere is
632: cooler, by roughly 150 K. Above log$(\tau)=-6$ the
633: temperature decreases even further. This is
634: likely a consequence of spline interpolation when the code is adjusting the
635: temperature in the lower parts of the atmosphere. Since the
636: lines are not sensitive to conditions above ~$log(\tau)=-6$ this
637: peculiar temperature profile has no effect on the fits. As expected based on the two sets of Stokes $V$
638: amplitudes, there are two sets of magnetic fluxes in the
639: photosphere. The spread of fluxes in the chromosphere is significantly smaller. Both
640: the photospheric and chromospheric velocities are lower than in area
641: 1. The velocities
642: are slightly negative in the photosphere and positive in the chromosphere.The models are in fairly
643: good agreement with one another. Since the code cannot retrieve sizes of velocity gradients, this
644: does not rule out that the differences in the 8498 \AA\ and 8542 \AA\
645: profile shapes are not caused by large gradients or discontinuities in
646: velocity (or inclination).
647:
648: \begin{figure*}
649: \epsscale{0.8}
650: \plotone{f12.eps}
651: \caption{As figure \ref{fig:a1} but for area 3.
652: \label{fig:a3}
653: }
654: \end{figure*}
655:
656:
657: Area 4 (figure \ref{fig:a4}) is also located in the network, but even farther from the strong photospheric magnetic flux concentrations than area 3. The chromospheric Stokes $I$ profiles show
658: self-reversals which are not seen in area 3. The self-reversals are stronger on the red side in the 8542 \AA\ line and in
659: most cases also in the 8498 \AA\ line. The chromospheric Stokes $V$ profiles all
660: show varying degrees of asymmetry where the red sides tend to have
661: multiple components. The inversions reproduce the general shapes of
662: the profiles, but not the amplitudes of the self-reversals and the
663: emission features in Stokes $V$. Since
664: the external non-magnetic atmosphere is based on an average profile of
665: both the map and time series, anything leading to asymmetries in Stokes $I$
666: has to be incorporated into the magnetic component. The photospheric
667: temperature is not very different from the external non-magnetic
668: atmosphere but the chromosphere is clearly hotter.
669:
670:
671: \begin{figure*}
672: \epsscale{0.8}
673: \plotone{f13.eps}
674: \caption{As figure \ref{fig:a1} but for area 4.
675: \label{fig:a4}
676: }
677: \end{figure*}
678:
679: \section{Discussion}
680: \label{sec:disc}
681:
682: Two points need to be kept in mind when interpreting the inversion results. Firstly, the sensitivity of the inversions is limited to specific
683: heights. Since the Fe lines and Ca line wings are formed in the photosphere, and only the Ca line cores are chromospheric, the sensitivity range is larger in the
684: photosphere than in the chromosphere. Because the
685: inversion code does not include treatment of atomic polarization (and
686: consequently Stokes $Q$ and $U$ cannot be used in the weak field
687: regime), it is not possible to resolve the ambiguity between the
688: magnetic field strength and inclination, instead only the longitudinal
689: magnetic flux density can be retrieved.
690:
691: Secondly, discontinuities in velocity and magnetic field inclination
692: result in complicated Stokes $V$ asymmetry patterns. The responses to
693: discontinuities in the two Ca lines are not always similar. Also the area and amplitude asymmetries
694: can respond in different ways. The
695: inversion code cannot reproduce the discontinuities. Such discontinuities may explain the
696: observed Stokes $V$ area and amplitude asymmetry histograms (P07): the 8542 \AA\ area asymmetry histogram
697: peaks at a positive value whereas all other histograms (8542 \AA\ amplitude asymmetry, 8498 \AA\ area and amplitude asymmetry) peak at a
698: negative value. Based on the high-$\beta$ simulations it seems
699: unlikely that the difference is caused by shocking acoustic waves alone.
700: The current inversions are not a good diagnostic for discontinuities
701: or strong localized gradients, especially since both velocity and inclination discontinuities can be
702: present. Since we do not include linear polarization in the inversions, and since the asymmetries are not reproduced by the inversions, there is information in the spectra which is not used by the current inversion code. A more suitable inversion code, especifically designed for the inhomogeneities of the chromosphere, is needed to solve the problem.
703:
704:
705:
706:
707: Despite the limitations, these non-LTE inversions remain a useful tool: observed Stokes
708: $I$ and $V$ amplitudes are reproduced successfully
709: even if asymmetries in Stokes $V$ are not.
710:
711: \subsection{Magnetic field}
712:
713: The magnetic contrast of the network and surroundings is significantly reduced
714: in the chromosphere, and there are clear indications of the network expanding with height. These indications include the spatial distribution of Stokes $V$ amplitudes in the Ca and Fe lines, the appearance of the magnetograms (figure 2.\ in P07) and spatial distributions of temperature and magnetic flux in the photosphere and chromosphere. The diffusive appearance and spreading of the
715: network in the chromosphere was already noted by
716: \cite{Harvey+Hall1971} who attributed it to the magnetic field
717: becoming more horizontal with height. This is in agreement with our
718: data, and especially with the clearly more diffuse and larger appearance of the network
719: in the 8542 \AA\ magnetogram when compared to the 8498 \AA\ magnetogram. However, since we do not have
720: information of the field inclination, we cannot conclude this for
721: sure.
722:
723: Not all of the strong magnetic flux concentrations in the photosphere
724: penetrate into the chromosphere. One possible cause is the expansion of
725: the field with height. This would reduce the longitudinal flux and increase
726: the amount of horizontal field on the expense of vertical
727: field. Another explanation is that some of the flux returns back to
728: the photosphere. \cite{Schrivjer+Title2003},
729: \cite{Jendersie+Peter2006} and \cite{Aiouaz+Rast2006} have explored
730: the connection of the network magnetic field to the corona, and shown
731: that the expansion of the network is affected by magnetic fields of
732: opposite polarity in the network and internetwork. Unfortunately, our
733: observations of the photospheric magnetic field are greatly limited by
734: the weakness of the iron lines. There is no indication of opposite
735: polarity magnetic field in our data, but as pointed out by the
736: above-mentioned authors, {\it sub-resolution} magnetic fields in the
737: internetwork play a crucial role in determining what fraction of the
738: network flux survives to the corona and what fraction returns to the
739: photosphere forming loops at various heights. The low-flux network
740: is a clear indication of there being a substantial amount of magnetic
741: flux not captured by the photospheric iron lines.
742:
743:
744: Based on the inspection of the individual atmospheres the photosphere appears more homogeneous, especially in terms of
745: magnetic flux. There is more small-scale spatial variation in the
746: chromosphere. The photosphere is unlikely to have large gradients or
747: discontinuities, and the photospheric radiation is in
748: LTE. Consequently, it is easier to find a model to fit the
749: observations in the photosphere and the spread of atmospheric models
750: will be smaller there. However, since the small-scale chromospheric
751: spatial patterns are coherent over more than just one pixel, and because
752: we see spatial variations on small scales in the line profiles as
753: well, it is likely that the inhomogeneous appearance of the chromosphere is real.
754:
755: \subsection{Dynamics}
756:
757: Velocities in the chromosphere, especially in most of the network,
758: are predominantly red-shifted in relation to the photosphere. The
759: chromospheric network velocity patterns are fragmented and large
760: velocities are more often seen outside the network. The latter is also
761: true in the photosphere where the difference is even more
762: pronounced. A similar finding is presented in \cite{Rezaei+others2007} where the fraction of
763: large velocities (determined from photospheric Stokes $V$ profiles) is
764: larger in the internetwork than in the network. In the current data, there is no difference between the
765: low-flux network and internetwork in terms of photospheric
766: velocities, but in the chromosphere the distribution of velocities is
767: slightly wider in the low-flux regions. The strongest magnetic fields suppress oscillations, although
768: large asymmetries are also seen in network line profiles. In general, the largest asymmetries are located at network boundaries. Also increasing Stokes $V$ asymmetries are correlated with decreasing magnetic flux. The asymmetries are in most cases at least partly due to
769: emission features in the line profiles, i.e. they are also likely to be related
770: to the temperature structure, not just velocity and magnetic field gradients. The difference between
771: the velocity histograms in the low-flux network and internetwork
772: indicates that magnetic fields that do not play a visibly important
773: role in the photosphere, become important higher up.
774:
775: Recent work by e.g., \cite{Jefferies+others2006} and
776: \cite{Vecchio+others2007}, has tied together the
777: magnetic topology and the observed oscillatory properties in the chromosphere. The network edges being more dynamic is
778: consistent with the picture presented by \cite{Jefferies+others2006}. In it inclined magnetic fields channel low-frequency waves into the
779: chromosphere. \cite{Jefferies+others2006} also point out that the low-frequency waves are
780: intermittent owing to the continual changes caused by the motion of
781: the convective cells and by the ``magnetic carpet'' (i.e., small
782: magnetic dipoles created in the internetwork and transported to the network where they interact with existing flux). Reconnection events would also probably lead to asymmetries in line profiles. In our data the
783: network clearly expands with height, and the difference between the
784: 8498 \AA\ and 8542 \AA\ magnetograms is striking, indicating that
785: conditions described above (i.e., inclined fields and more dynamic network boundaries) are likely to exist. Furthermore, since the
786: photospheric driver has most of its power at periods of 5 min, the
787: driver of chromospheric oscillations in regions with inclined fields
788: is stronger than in the internetwork where the 5 minute period
789: waves are filtered out. This was pointed out by \cite{DePontieu+others2007}
790: in connection to the dynamic behavior of fibrils. All the above is consistent
791: with the network, and especially the boundaries, being more dynamic
792: and exhibiting more self-reversals and profile asymmetries than the
793: internetwork.
794:
795:
796:
797:
798:
799:
800: \subsection{Temperature proxies and heating of the magnetic chromosphere}
801:
802: \label{sec:heating}
803: Figure \ref{fig:i-vs-flux} shows that the intensity of the Ca II IR triplet lines
804: saturates at high chromospheric temperatures. A similar saturation
805: effect is seen if the magnetic flux is plotted instead of the
806: temperature. In \cite{Rezaei+others2007} a similar plot, but for the
807: Ca H intensity and photospheric flux, is interpreted as a
808: saturation limit for the magnetic heating (i.e., the heating as a
809: function of field strength or filling factor levels off at high
810: values). The inversions show that the leveling off in the Ca II IR triplet lines is, at least partly,
811: a radiative transfer effect, and does not necessarily imply that the
812: magnetic heating has an upper limit that depends on the field
813: strength or filling factor. The data do not conclusively show this not
814: to be the case either. (In fact, there are some signs of a saturation effect
815: in scatter plots of magnetic flux and chromospheric temperature.)
816: Also a lower limit on chromospheric emission, similar to that seen in
817: \cite{Rezaei+others2007}, is seen in our data, but no lower limit on
818: the chromospheric temperature is found. The \ion{Ca}{2} IR triplet line intensities are
819: linearly dependent on temperature only in the range
820: between roughly 5800 and 6500 K. Interpreting proxies based on lines
821: formed outside of LTE is not straight-forward, and the effects of
822: radiative transfer need to be considered in detail when inferring physical properties, such as velocity and temperature, from proxies.
823:
824: The network patch appears on large scales fairly
825: uniform in temperature, and the hottest pixels do not
826: always coincide with strongest photospheric or chromospheric flux. Nor do
827: they coincide with the largest filling factors. Based on this the
828: relationship between heating and magnetic flux does not appear to be
829: linear. However, since the low-flux network is cooler, the amount of flux
830: is clearly a factor in the heating. We see signs of a saturation effect in
831: the heating (flux as a function of chromospheric temperature) as
832: proposed by \cite{Rezaei+others2007} and references therein. The
833: number of data points is not, however, large enough for making definite
834: conclusions. Further observations and inversions, preferably
835: simultaneously with strong photospheric lines, are needed.
836:
837:
838: \section{Final remarks}
839: \label{sec:conc}
840: Because of non-LTE it is important to take into account radiative
841: transfer effects when interpreting proxies based on line
842: profiles. This is why non-LTE inversions are an invaluable tool: with them it is possible to infer macroscopic physical quantities without resorting to proxies. The inversions presented in this work
843: further highlight the physical differences between the dynamics and structure of the chromospheric
844: network and internetwork. The chromosphere is divided into three regions
845: characterized best by chromospheric temperature, but also visible in other chromospheric parameters. The fundamental differences
846: between the three regions are most likely related to magnetic fields.
847:
848: There are limitations to what can be done with current
849: non-LTE inversion codes, and forward modeling remains a crucial
850: part in understanding the dynamics of the magnetic chromosphere. We
851: plan to make simultaneous observations of the Ca II IR triplet lines and
852: the \ion{He}{1} 10830 \AA\ lines to further validate the inversions,
853: to obtain a more 3-dimensional view of the magnetic chromosphere, and
854: to study the connection between thermal properties of the chromosphere (magnetic heating) and magnetic field topology.
855:
856: \begin{thebibliography}{30}
857: \expandafter\ifx\csname natexlab\endcsname\relax\def\natexlab#1{#1}\fi
858:
859: \bibitem[{{Aiouaz} \& {Rast}(2006)}]{Aiouaz+Rast2006}
860: {Aiouaz}, T. \& {Rast}, M.~P. 2006, \apjl, 647, L183
861:
862: \bibitem[{{Centeno} {et~al.}(2005){Centeno}, {Socas-Navarro}, {Collados}, \&
863: {Trujillo Bueno}}]{Centeno+others2005}
864: {Centeno}, R., {Socas-Navarro}, H., {Collados}, M., \& {Trujillo Bueno}, J.
865: 2005, \apj, 635, 670
866:
867: \bibitem[{{De Pontieu} {et~al.}(2004){De Pontieu}, {Erd{\'e}lyi}, \&
868: {James}}]{DePontieu+others2004}
869: {De Pontieu}, B., {Erd{\'e}lyi}, R., \& {James}, S.~P. 2004, \nat, 430, 536
870:
871: \bibitem[{{De Pontieu} {et~al.}(2007){De Pontieu}, {Hansteen}, {Rouppe van der
872: Voort}, {van Noort}, \& {Carlsson}}]{DePontieu+others2007}
873: {De Pontieu}, B., {Hansteen}, V.~H., {Rouppe van der Voort}, L., {van Noort},
874: M., \& {Carlsson}, M. 2007, \apj, 655, 624
875:
876: \bibitem[{Gabriel(1976)}]{Gabriel1976}
877: Gabriel, A. 1976, Phil Trans. Royal Soc. Lond., 281, 339
878:
879: \bibitem[{{Hansteen} {et~al.}(2006){Hansteen}, {De Pontieu}, {Rouppe van der
880: Voort}, {van Noort}, \& {Carlsson}}]{Hansteen+others2006}
881: {Hansteen}, V.~H., {De Pontieu}, B., {Rouppe van der Voort}, L., {van Noort},
882: M., \& {Carlsson}, M. 2006, \apjl, 647, L73
883:
884: \bibitem[{{Harvey} \& {Hall}(1971)}]{Harvey+Hall1971}
885: {Harvey}, J. \& {Hall}, D. 1971, in IAU Symp. 43: Solar Magnetic Fields, ed.
886: R.~{Howard}, 279--+
887:
888: \bibitem[{{Jefferies} {et~al.}(2006){Jefferies}, {McIntosh}, {Armstrong},
889: {Bogdan}, {Cacciani}, \& {Fleck}}]{Jefferies+others2006}
890: {Jefferies}, S.~M., {McIntosh}, S.~W., {Armstrong}, J.~D., {Bogdan}, T.~J.,
891: {Cacciani}, A., \& {Fleck}, B. 2006, in ESA SP-624: Proceedings of SOHO
892: 18/GONG 2006/HELAS I, Beyond the spherical Sun
893:
894: \bibitem[{{Jendersie} \& {Peter}(2006)}]{Jendersie+Peter2006}
895: {Jendersie}, S. \& {Peter}, H. 2006, \aap, 460, 901
896:
897: \bibitem[{Judge {et~al.}(2001)Judge, Casini, Tomczyk, Edwards, \&
898: Francis}]{Judge+others2001}
899: Judge, P., Casini, R., Tomczyk, S., Edwards, D.~P., \& Francis, E. 2001,
900: Coronal Magnetogmetry: a feasibility study, Tech. Rep. NCAR/TN-446-STR,
901: National Center for Atmospheric Research
902:
903: \bibitem[{{Lagg} {et~al.}(2004){Lagg}, {Woch}, {Krupp}, \&
904: {Solanki}}]{Lagg+others2004}
905: {Lagg}, A., {Woch}, J., {Krupp}, N., \& {Solanki}, S.~K. 2004, \aap, 414, 1109
906:
907: \bibitem[{{Mart{\'{\i}}nez Pillet}(1997)}]{MartinezPillet+others1997}
908: {Mart{\'{\i}}nez Pillet}, V. et~al. 1997, \apj, 474, 810
909:
910: \bibitem[{{McIntosh} {et~al.}(2003){McIntosh}, {Fleck}, \&
911: {Judge}}]{McIntosh+Fleck+Judge2003}
912: {McIntosh}, S.~W., {Fleck}, B., \& {Judge}, P.~G. 2003, \aap, 405, 769
913:
914: \bibitem[{{McIntosh} \& {Judge}(2001)}]{McIntosh+Judge2001}
915: {McIntosh}, S.~W. \& {Judge}, P.~G. 2001, Astrophys.\ J., 561, 420
916:
917: \bibitem[{{Merenda} {et~al.}(2006){Merenda}, {Trujillo Bueno}, {Landi
918: Degl'Innocenti}, \& {Collados}}]{Merenda+others2006}
919: {Merenda}, L., {Trujillo Bueno}, J., {Landi Degl'Innocenti}, E., \& {Collados},
920: M. 2006, \apj, 642, 554
921:
922: \bibitem[{{Pietarila} {et~al.}(2007){Pietarila}, {Socas-Navarro}, \&
923: {Bogdan}}]{Pietarila+others2007a}
924: {Pietarila}, A., {Socas-Navarro}, H., \& {Bogdan}. 2007, in press
925:
926: \bibitem[{{Pietarila} {et~al.}(2006){Pietarila}, {Socas-Navarro}, {Bogdan},
927: {Carlsson}, \& {Stein}}]{Pietarila+others2006}
928: {Pietarila}, A., {Socas-Navarro}, H., {Bogdan}, T., {Carlsson}, M., \& {Stein},
929: R.~F. 2006, \apj, 640, 1142
930:
931: \bibitem[{{Rezaei} {et~al.}(2007){Rezaei} et~al.}]{Rezaei+others2007}
932: {Rezaei}, R. et~al. 2007, ArXiv Astrophysics e-prints
933:
934: \bibitem[{{Rybicki} \& {Hummer}(1991)}]{Rybicki+Hummer1991}
935: {Rybicki}, G.~B. \& {Hummer}, D.~G. 1991, \aap, 245, 171
936:
937: \bibitem[{{Sasso} \& {Solanki}(2006)}]{Sasso+Lagg+Solanki2006}
938: {Sasso}, C., L.~A. \& {Solanki}, S. 2006, A\&A, 456, 367
939:
940: \bibitem[{{Schrijver} \& {Title}(2003)}]{Schrivjer+Title2003}
941: {Schrijver}, C.~J. \& {Title}, A.~M. 2003, \apjl, 597, L165
942:
943: \bibitem[{{Socas-Navarro}(2005)}]{Socas-Navarro2005a}
944: {Socas-Navarro}, H. 2005, \apjl, 631, L167
945:
946: \bibitem[{{Socas-Navarro} {et~al.}(1998){Socas-Navarro}, {Ruiz Cobo}, \&
947: {Trujillo Bueno}}]{Socas-Navarro+RuizCobo+TrujilloBueno1998}
948: {Socas-Navarro}, H., {Ruiz Cobo}, B., \& {Trujillo Bueno}, J. 1998, \apj, 507,
949: 470
950:
951: \bibitem[{{Socas-Navarro} \& {Trujillo
952: Bueno}(1997)}]{Socas-Navarro+TrujilloBueno1997}
953: {Socas-Navarro}, H. \& {Trujillo Bueno}, J. 1997, \apj, 490, 383
954:
955: \bibitem[{{Socas-Navarro} {et~al.}(2000{\natexlab{a}}){Socas-Navarro},
956: {Trujillo Bueno}, \& {Ruiz Cobo}}]{Socas-Navarro+others2000b}
957: {Socas-Navarro}, H., {Trujillo Bueno}, J., \& {Ruiz Cobo}, B.
958: 2000{\natexlab{a}}, Science, 288, 1398
959:
960: \bibitem[{{Socas-Navarro} {et~al.}(2000{\natexlab{b}}){Socas-Navarro},
961: {Trujillo Bueno}, \& {Ruiz Cobo}}]{Socas-Navarro+TrujilloBueno+RuizCobo2000}
962: ---. 2000{\natexlab{b}}, \apj, 530, 977
963:
964: \bibitem[{{Socas-Navarro} {et~al.}(2006)}]{Socas-Navarro+others2006a}
965: {Socas-Navarro}, H. {et~al.} 2006, Solar Physics, 235, 55
966:
967: \bibitem[{{Socas-Navarro} {et~al.}(2006)}]{Socas-Navarro+others2006b}
968: {Socas-Navarro}, H. et~al. 2006,
969: Solar Physics, 235, 75
970:
971: \bibitem[{{Trujillo Bueno} {et~al.}(2005){Trujillo Bueno}, {Merenda},
972: {Centeno}, {Collados}, \& {Landi Degl'Innocenti}}]{TrujilloBueno+others2005}
973: {Trujillo Bueno}, J., {Merenda}, L., {Centeno}, R., {Collados}, M., \& {Landi
974: Degl'Innocenti}, E. 2005, \apjl, 619, L191
975:
976: \bibitem[{{Vecchio} {et~al.}(2007){Vecchio}, {Cauzzi}, {Reardon}, {Janssen}, \&
977: {Rimmele}}]{Vecchio+others2007}
978: {Vecchio}, A., {Cauzzi}, G., {Reardon}, K.~P., {Janssen}, K., \& {Rimmele}, T.
979: 2007, \aap, 461, L1
980:
981: \end{thebibliography}
982: \end{document}