A&A 431, 1069-1073 (2005)
DOI: 10.1051/0004-6361:20041550

Solar prominences with Na and Mg emissions and centrally reversed Balmer lines

G. Stellmacher1 - E. Wiehr2

1 - Institut d'Astrophysique (CNRS), 98bis Bd. Arago, 75014 Paris, France
2 - Universitäts-Sternwarte, Geismarlandstraße 11, 37083 Göttingen, Germany

Received 29 June 2004/ Accepted 14 October 2004

We observed bright solar limb prominences with significant emission of NaD2 and Mgb2 simultaneously with the H$\alpha $, H$\beta $, HeD3, He+4685, and the He $^{\rm singl}$ 5015 Å lines, using the THEMIS telescope on Tenerife. We find that most prominences with significant NaD2 and Mgb2 emissions show pronounced centrally reversed H$\alpha $ profiles, and occasionally even of H$\beta $; the strongest emissions reach integrated intensities $E\beta>16\times{}10^4$ [erg/(cm2 s str)]. The centrally reversed profiles are well reproduced by semi-infinite models. The source function reaches S $\alpha\le{}36~\times~{}10^4$ [erg/(cm2 s str Å)] corresponding to an excitation temperature $T_{\rm ex}^{\alpha}\approx3950$ K; here, the optically thickness of H$\alpha $ amounts $\tau^0_{\alpha}\approx{}10$. The line widths of the NaD2, Mgb2, and HeD3 profiles yield kinetic temperatures $7000\le{}T_{\rm kin}<8000$ K and non-thermal broadening $v_{\rm tu}=5$ km s-1.

Key words: Sun: prominences - radiation mechanisms: non-thermal - line: formation

1 Introduction

The simultaneous emission of resonance lines with low ionization potential, like Mgb2 and NaD2, and of hydrogen and helium with much higher excitation and ionization energies, illustrates the large deviations from LTE in atmospheres of solar prominences. However, only a few comprehensive data sets of such emissions have so far been published, e.g., Yakovkin & Zel'dina (1964), Kim (1987). High precision data of spectral prominence photometry show for faint emissions ( $E\beta\le{}1\times{}10^4$ erg/(cm2 s str), corresponding to $\tau_{\alpha}\le{}1$) a unique empirical relation between H$\alpha $ and H$\beta $ independent on the individual prominence atmospheres (Stellmacher & Wiehr 1994b). For higher emissions, this relation depends on the prominence atmospheres. The present observations are considered as an extension toward strongest emissions $E\beta>4~\times~{}10^4$ erg/(cm2 s str). Such bright prominences are known to be cool, dense, and rather unstructured (Stellmacher & Wiehr 1995). They allow a determination of upper limits of the source function of H$\alpha $ as well as a quantitative analysis of the centrally reversed H$\alpha $ profiles and their representation by models.

\par\includegraphics[width=8.2cm] {1550figA.ps}\par\mbox{\hspace*...
...lsize October 23, 2000, SE\hspace*{1.4cm} October 25, 2000, SE}
\par\end{figure} Figure 1: H$\alpha $ solar survey images from the Meudon observatory with the observed prominences at E/22N and W/24S ( upper panel) and at E/5N, E/15S, E/23S, and E/32S ( lower left panel), together with the corresponding disk filaments two days later ( lower right panel) showing the different aspect angles.
Open with DEXTER

2 Observations and data reduction

We simultaneously observed with the French-Italian solar telescope THEMIS the emission lines H$\alpha $, H$\beta $, NaD2, Mgb2, HeD3, He+4685, and He $^{\rm singlet}$5015 Å with seven CCD-cameras on October 18 and 23, 2000. Entrance slits of 0.5 arcsec and 0.75 arcsec widths for the two data sets were oriented parallel to the direction of atmospheric refraction. Exposure times of a few seconds yielded about 2000 counts for the brightest H$\alpha $, and 300 counts for the faintest Mgb2 emissions.

The CCD-images were corrected for the dark and the gain matrices; the underlying stray-light aureole was subtracted using spectra from locations adjacent to the corresponding prominence. For the calibration of the prominence emissions, we took disk-center spectra and used the absolute intensities of Labs & Neckel (1970): $I(6563~\rm\AA{}) = 2.86$, $I(5890~\rm\AA{})=3.34$, $I(5876~\rm\AA{})=3.36$, $I(5173~\rm\AA{}) = 3.93$, $I(5015~\rm\AA{})=4.06$, $I(4861~\rm\AA{})=4.16$, $I(4685~\rm\AA{}) = 4.3$ [106erg/(cm2 s str Å)].

We determine the total emission E of a line as the intensity integrated over the whole emission profile: $E~=~\int I \rm d\lambda$ [erg/(cm2 s str)]; we give I and E in units of 104 to enable an easy comparison with former data. H$\alpha $ solar survey images of the prominences analyzed (Fig. 1) show that these low solar latitude objects ( $\phi<32^\circ$) occurred under various aspect angles between "face-on'' and "end-on''. An example of simultaneously observed CCD spectra is displayed in Fig. 2.

\par\includegraphics[width=8.5cm]{1550fig2.eps}\end{figure} Figure 2: CCD spectra of the Balmer lines H$\alpha $ ( upper left), H$\beta $ ( upper right), both with central reversions, and the simultaneously observed HeD3 ( lower left), NaD2 ( lower right) in the prominence at E/22N on Oct.18; each sub-image: 4 $1'' \times 2.5$ Å.
Open with DEXTER

3 Results for the H$\alpha $ and H$\beta $ lines

3.1 The H$\alpha $ versus H$\beta $ emission relations

Our observed H$\alpha $ and H$\beta $ emissions, given in Fig. 3, reach integrated intensities of up to $E\beta=16\times{}10^4$, being four times higher than the maximum values by Stellmacher & Wiehr (1994b, their Figs. 2 and 3). Comparably high values were published by Yakovkin & Zel'dina (1963; shown in Fig. 3).

\par\includegraphics[width=8.5cm,clip]{1550fig3.eps}\end{figure} Figure 3: Observed integrated intensities of H$\alpha $ and H$\beta $in units of 104 erg/(cm2 s str); upper panel: $E\alpha $ versus $E\beta $; lower panel: first Balmer decrement $E\alpha $/$E\beta $ versus $E\beta $; prominences with conspicuous centrally reversed H$\alpha $ profiles are marked by filled circles; mean observations by Stellmacher & Wiehr (1994b) by crosses; observations by Yakovkin & Zel'dina (1963) by dags; model calculations by GHV (1993) for $T_{\rm kin}=4300$ K (full) and $T_{\rm kin}=6000$ K (dashed line).
Open with DEXTER

For faint emissions, our data perfectly match those from Stellmacher & Wiehr (1994b) who found for $E\beta\le{}1\times{}10^4$ a unique empirical relation between $E\alpha $ and $E\beta $ (crosses in Fig. 3); for strong emissions $E\beta>3\times{}10^4$ our recent data slightly fall below their curve. For the first Balmer decrement $E\alpha $/$E\beta $ we find values near 3.0 for the brightest H$\beta $ emissions, and limiting values clearly above 10.0 for the faintest emissions. This agrees with Stellmacher & Wiehr (1994b), and follows the curves calculated by Gouttebroze et al. (1993; hereafter referred to as GHV) for temperatures of $T_{\rm kin}=4300$ K and $T_{\rm kin}=6000$ K. The fact that the present observations coincide well with earlier ones of much fainter and more structured prominences, obtained under quite different conditions, indicates that possible influences from "filling'' are negligible (in agreement with Stellmacher et al. 2003). This is also seen from the independence on the aspect angle: we do not find significant differences between face-on (e.g. the prominence from Oct. 18 at W/24S) and end-on objects (e.g. the prominence from Oct. 23 at E/32S; cf., Fig. 1).

3.2 Centrally reversed H$\alpha $

We observe distinct central reversions (double peaks) of the H$\alpha $ line profile for integrated intensities $E\beta>5~\times~{}10^4$. An example of centrally reversed H$\alpha $ and H$\beta $ profiles is shown in Fig. 4 together with profiles of other simultaneously observed lines. We find the most prominent central reversions in the strongest, yet narrow emission profiles. In Fig. 5 we give the observed relations in comparison to corresponding curves deduced from the comprehensive set of H$\alpha $ emission profiles calculated by GHV for thick slabs of models with $T_{\rm kin}=6000{-}8000$ K and $v_{\rm tu}=5$ km s-1, acting as semi-infinite layers. We find that profiles with the most prominent central reversions are markedly narrow; their values (filled circles in Fig. 5) well follow the calculated relations and can, hence, be explained by pure line-saturation. Data that deviate from the calculated curves (open circles) are derived from broader line profiles; - stronger broadening (e.g., by macro velocities or superpositions) will readily lead to a deterioration of the pure saturation effect.

\includegraphics[width=4cm]{1550figG.eps}\par\end{figure} Figure 4: Profiles of simultaneously observed emissions with distinct saturation of H$\alpha $ and H$\beta $ ( upper panels) together with He D3 and Na D2 ( lower levels); ordinates = CCD counts; thick abscissa tick marks = 0.5 Å; (He-CCD miscentered).
Open with DEXTER

\par\includegraphics[width=8cm,clip]{1550fig5.eps}\end{figure} Figure 5: Observed relations between central-intensities $I^0_{\alpha }$, peak-intensities $I^{\max}_{\alpha}$, peak-to-peak distance ptpd of the two H$\alpha $ maxima, and half line width $\Delta {}\lambda {}_e^{\alpha }$ at $I^0_{\alpha }/e$ together with corresponding curves (dashed lines) deduced from calculations of slab-models by GHV; observed emissions with prominent central reversals are marked by filled circles (as in Fig. 3); filled squares denote values with corresponding H$\beta $ profiles being also centrally reversed.
Open with DEXTER

The mean Doppler widths of our simultaneously observed lines amount to $\Delta{}\lambda_{\rm D}^{\rm HeD_3} \le 160$, $\Delta{}\lambda_{\rm D}^{\rm NaD_2} \approx 95$, and $\Delta{}\lambda_{\rm D}^{\rm Mgb_2} \approx 82$ mÅ; (the latter two being obtained from plots $E_{\rm Na,Mg}~=~I^0_{\rm Na,Mg}\cdot{}\Delta\lambda_{\rm D}\cdot{}\sqrt\pi$). These widths set an upper limit for the broadening parameters of the observed prominence lines of $7000\le{}T_{\rm kin}<8000$ K with a non-thermal (Maxwellian) velocity of 5 km s-1.

The fact that the bright and rather unstructured prominences show strikingly narrow lines was already mentioned by Stellmacher & Wiehr (1994b). We consider the central reversions as a signature of emission in semi-infinite dense layers. No evident relation is found between the intensity difference of the two emission peaks and the line-center wavelength. Here, non-LTE transfer calculations should be considered, including "spatially correlated velocity fields'' as suggested by Magnan (1976).

3.3 Central Intensities and source function S$\alpha $

The double peaked H$\alpha $ emissions originate from thick layers $\tau^0_{\alpha}>>1$, for which the line center intensities become $I^0_{\alpha} = S_{\alpha}(1-{\rm e}^{-\tau_{\alpha}})\approx{}S_{\alpha}$allowing us to directly deduce the source function. For the stronger emissions, we find $\tau^0_{\alpha}>5$ (following the method described by Stellmacher & Wiehr (1994b; Sect. 4). If we express the relative level population of the H$\alpha $ transition by the Boltzmann formula: $(n_{0,3}~g_{0,3})/(n_{0,2}~g_{0,2})=\exp[-hc/(\lambda_{\alpha}~k~T^{\alpha}_{\rm ex})]$, and insert this into the general equation for the source function, we obtain the corresponding Planck formula for the excitation temperature $T_{\rm ex}^{\alpha}$:

\begin{displaymath}S_{\alpha}=(2~h~c^2/\lambda^5_{\alpha}/(\exp[-hc/(\lambda_{\alpha}~k~T^{\alpha}_{\rm ex})]-1)
= B(T^{\alpha}_{\rm ex})

(B being the Planck function).

The mean upper values $I^0_{\alpha}\approx{}36\times{}10^4$ (Fig. 6) correspond to an excitation temperature $T_{\rm ex}^{\alpha}\approx3950$ K. Fainter prominences analyzed by Stellmacher & Wiehr (1994b) gave smaller mean upper values $I^0_{\alpha}\approx{}26.5\times{}10^4$ corresponding to $T_{\rm ex}^{\alpha}\approx{}3700$ K. GHV obtain for their models with $T_{\rm kin}=6000$ K and $v_{\rm tu}=5$ km s-1 a source function $S_{\alpha}=36\times{}10^4$ at $\tau^0_{\alpha}=10$, in good agreement with the present observations (Fig. 6).

\par\includegraphics[width=8.3cm,clip]{1550fig6.eps}\end{figure} Figure 6: Observed relation between the integrated intensity $E_{\alpha }$ and the central intensity $I^0_{\alpha }$; filled circles mark H$\alpha $ profiles with prominent central reversions, filled squares denote values with centrally reversed corresponding H$\beta $ profiles.
Open with DEXTER

The intensity peaks of the centrally reversed profiles reach values near $I^{\max}_{\alpha}=43\times{}10^4$ (cf., Fig. 5) which correspond to $T_{\rm ex}^{\alpha}=4000$ K. These peaks outside the line center arise from smaller $\tau_{\alpha}$ values, their higher intensity then indicates an increase of the source function towards the prominence interior (see also Yakovkin & Zel'dina 1964). The model calculations GHV (1994; their Fig. 18) indicate a rise of $S_{\alpha}$ beyond $0.16\times{}I^{\rm phot}$, i.e. $46\times{}10^4$; this doe not seem to be observed in prominences, and it would disagree with their visibility as dark filaments on the disk. Such high $S_{\alpha}$ may be valid for spicules and eruptive objects.

4 The helium emissions

The simultaneously observed HeD3 and H$\beta $ lines show distinct branches in their intensity relation (Fig. 7): Emissions from prominence locations with prominent H$\alpha $ reversions show ratios E $_{\rm HeD_3}/E_{\beta}=0.2{-}0.3$, while less thick prominences follow branches with ratios 0.4-0.5. This confirms earlier results by Stellmacher & Wiehr (1994a, 1995) who found from analysis of He 3889 and H 3888, and of He D3 and H$\beta $, that the emission ratios of the He-triplet and the Balmer lines show for individual prominences typical mean values, in the sense that prominences with stronger Balmer emissions (known to be cooler, less structured, and denser; cf., Introduction) yield lowest Helium-to-Balmer ratios.

\par\includegraphics[width=6.8cm,clip]{1550fig7.eps}\end{figure} Figure 7: Integrated intensities of HeD3 and H$\beta $; emissions corresponding to prominent H$\alpha $ reversions are marked by filled circles; full lines trace emission ratios of 0.5, 0.4, 0.3, and 0.2.
Open with DEXTER

The ratio of the HeD3 fine-structure components is a measure of the optical thickness, as was shown by Stellmacher et al. (2003) for the analogous case of the triplet He 10830 Å. In contrast, the HeD3 triplet does not allow us to determine the ratio of the faint red and the (not separated) two blue components with similar reliability, due to the much smaller spectral separation of its components. We find ratios above 1/6, indicating that HeD3 begins to saturate, for emissions $E(\rm HeD_3) \ge 0.6\times10^4$.

Our instrumental set-up also allowed the simultaneous observation of the lines He  II 4685.7 and He  I 5015.7 (singlet). We did not find any significant He  II emission in the prominences observed, which may be too cool and dense for sufficient He  II excitation. The faint He  I singlet line was only measurable in one prominence, its width is quite similar to that of HeD3: $(\Delta{}\lambda{}_{\rm D}/\lambda_0)_{\rm He}<3\times{}10^{-5}$. The integrated intensity $E(\rm He~5016)\approx{}0.05\times{}10^4$ gives an emission ratio with HeD3 of He $^{\rm singl}/\rm He^{tripl}\approx{}0.016$. Other line combinations, including the stronger singlet line He 6678, might be useful to extend the study of singlet-to-triplet ratio by Stellmacher & Wiehr (1997).

\par\includegraphics[width=7cm,clip]{1550fig8.eps}\end{figure} Figure 8: Emission relation of the Mgb2 and the NaD2 lines.
Open with DEXTER

5 Emissions of the metal lines NaD2 and Mgb2

The integrated intensities E(Mgb2) and E(NaD2) show a linear relation (with a gradient of 0.7; Fig. 8), indicating that the emissions of both lines are closely related. Their integrated intensities strongly depend on the prominence thickness, as can be seen from the relation with H$\beta $ in Fig. 9. We observe reliable emissions of NaD2 only if $E_{\beta}>3\times{}10^4$. The strongest NaD2 emissions, up to $E(\rm NaD_2)\approx{}0.78\times{}10^4$, are observed at prominence locations where the H$\beta $ profiles are saturated or even centrally reversed (i.e., $E_\beta>10\times{}10^4$).

The narrow widths of these turbulence-sensitive metal lines ( $\Delta{}\lambda_{\rm D}^{\rm NaD_2} \approx 95$ mÅ, $\Delta{}\lambda_{\rm D}^{\rm Mgb_2} \approx 82$ mÅ) imply rather cool emission layers with line broadening parameters $T_{\rm kin}~\approx~7000$ K and $v_{\rm tu}~\approx~5$ km s-1 (see Sect. 3.2). Similarly narrow profiles of Mgb2 were reported by Landman (1985). Comparison with model calculations by Kim (1987; Fig. 7) indicates that these observations can only be reproduced with high total number densities $N \ge 10^{12}$ cm-3.

\par\includegraphics[width=7cm,clip]{1550fig9.eps}\end{figure} Figure 9: Integrated intensity of NaD2 versus that of H$\beta $; dags show the observations by Yakovkin & Zel'dina (1963).
Open with DEXTER

6 Conclusion

The present spectro-photometry extends former analysis (Stellmacher & Wiehr 1994b, 1995) to four times higher H$\beta $ emissions. Our observed bright prominences are low latitude objects, i.e. at $\phi\le32\hbox{$^\circ$ }$. We find prominent central reversions of H$\alpha $ and occasionally of H$\beta $ for integrated intensities $E_{\beta}>5\times{}10^4$ [erg/(cm2 s str)]. These centrally reversed profiles can well be modeled assuming semi-infinite layers, as is seen from a comparison with model calculations by Gouttebroze et al. (1993). The emitting layers should then consist of densely wound fibers forming massive ropes or wicks (cf., Engvold 1997).

THEMIS proved to be a powerful instrument for multi-line spectral photometry, also useful for solar prominences. Due to its low stray-light level (seen in the rather faint aureoles in ourspectra) and its low instrumental polarization, one may extend these observations to filtergram techniques (cf., Stellmacher & Wiehr 2000) for a study of the dynamics of small-scale prominence structures inclusive their magnetic field as, e.g., done by Wiehr & Bianda (2003).

We thank the THEMIS team, in particular Dr. C. Briand, for kind support. We are indebted to ENO for the grant "THEMIS project No. 42''. The THEMIS telescope on Tenerife is operated by the French "Centre National de la Recherche Scientifique'' and the Italian "Consiglio Nazionale delle Ricerche'' at the Spanish "Observatorio del Teide'' of the Instituto de Astrofísica de Canarias.



Copyright ESO 2005