4. Diagnostics on the emitting region
Figs. 1, 2 and 4 indicate a relationship between relative redshifts and intensities for the transition region lines. They also show that the brighter pixels belong to the fine structures which appear to be loops of varying sizes. To investigate these results further, we have divided each dataset into several classes according to the line intensity.
First, the fitted line intensities of O v (for the SEP16 dataset) and Si iv (for the SEP17 dataset) have been divided into 11 intensity classes. The spatial pixels included in each of these intensity classes determine a spatial "mask" into the field of view. Then, the spectrum of each dataset has been averaged according to these masks. This has produced 11 average spectra from each dataset: spectral lines have been fitted using Gaussian functions. It should be noted that the two brightest intensity classes include pixels where line profiles were heavily distorted due to plasma activity: in these two classes the averaged line profiles depart from a Gaussian shape and therefore fitted line positions and intensity ratios are not precise. The two lowest intensity classes correspond to the darker plasma outside the structures, while the remaining classes include emission from structured plasma.
4.1. Line position trends
The measured line positions are displayed in Fig. 5 for O v 2629.7 Å, N v 1238.8 Å and Si iv 1402.8 Å. In all cases line positions become more red-shifted as the intensity increases. The maximum relative motions between the bright fine structures and dark unstructured plasma is 4.0 km s-1 (N v) and 6.0 km s-1 (O v and Si iv). The O iv 1401.2 Å line, which is weaker than the other lines, shows a similar trend, with relative motions up to 7.7 km s-1. These values are somewhat lower than those given by Teriaca et al. (1999a) for the quiet Sun, particularly for N v.
Unfortunately, the lack of reliable measurements of chromospheric line positions prevented the determination of an absolute wavelength scale, so that it is not possible to say whether the darker plasma outside the structures is moving outward or the structured plasma is red-shifted relatively to the rest position. Other observations with SUMER indicate that the latter is the most likely scenario.
4.2. Line width trends
Fitted line widths are displayed as a function of line intensity in Fig. 6 for Si iv 1402.8 Å, O v 2629.7 Å and N v 1238.8 Å. Also the O iv 1401.2 Å line width has been measured. In all cases the spectral line width increases as intensity increases. It is to be noted however that the two highest intensity classes include active pixels where line profiles are severely distorted by plasma motions (see Sect. 5) and could not be reliably fitted with Gaussian functions: these pixels provide artificially high line widths and so the last two data points in Fig. 6 must be treated with caution.
Each intensity class includes a large number of pixels, whose fitted line positions are different from pixel to pixel due to relative small line shifts. When the emission of all pixels of each intensity class is averaged, these line shifts may introduce a spurious line broadening which needs to be removed. In order to correct for this, we have shifted the emission of each pixel along the wavelength direction to a fixed, arbitrarily chosen common value. The resulting spectrum has been averaged and fitted to obtain the corrected values for the line width. These are displayed in Fig. 6 as dashed lines. The resulting, corrected line widths are on the average smaller than the uncorrected values, but differences are quite small. It is important to note that the width versus intensity trend is confirmed.
4.3. Non-thermal speeds trends
The width-intensity relation has consequences on the non-thermal velocities in the emitting plasma. Signatures of a relation between line intensity and non-thermal velocities have been reported by Dere et al. (1984) for C iv and Dere & Mason (1993) for C iv and Si iv. They found that brighter regions show higher non-thermal velocities. More recently, Chae et al. (1998b) have measured non-thermal velocities for several ions spanning a temperature range from photosphere to the corona (C ii, Si iv, C iv and O vi). These authors find the same trend, that lines formed at different temperatures show different degrees of correlation between line intensity and non-thermal speed.
In the present work, the use of the `intensity classes' technique allows us to investigate this relation for Si iv, N v, O iv and O v. We determine the most probable non thermal speed following the method described by Chae et al. (1998b). The resulting vs. intensity curves are displayed in Fig. 7. It is possible to see a clear relation between the two quantities, although the gradient of the curve is different for different spectral lines. On average, values determined in the present work are somewhat higher than those reported by Chae et al. (1998b) and Dere & Mason (1993). In order to determine the gradient of the -I relation we have rejected the two brightest measurements, as line profiles in these two intensity classes were corrupted by activity. We have fitted a linear curve on each of the plots and values for the gradients are reported in Table 2.
Table 2. Gradient of the -I relation for Si iv, N v, O iv and O v observed lines.
4.4. Electron density and temperature
The O iv spectral lines have frequently been used as a density diagnostic for the transition region plasma. It should be noted that with any density diagnostic techniques, only a rough `average' density can be obtained, since the transition region is dynamic, highly structured and inhomogeneous.
O'Shea et al. (1998) investigated the 1401/1407 line ratio as a density diagnostic for the transition region, using SUMER data. However, the 1407.4 Å line is blended with the O iii second order multiplet at 703.8 Å. Although it is possible to separate the O iv and O iii lines, this leaves a greater uncertainty in the measured line intensities. Branching ratios with the isolated 1399.8 Å line show that the O iv 1407.4 Å line is stronger than expected. In this work, the 1401/1399 line ratio is preferred for density determination. These two lines are free from blending contributions but their ratio is slightly temperature dependent.
The electron density has been measured inside each averaged intensity class and results are displayed in Fig. 8 (top). Theoretical line ratios are taken from the CHIANTI database (Dere et al. 1997, Landi et al. 1999b). A consistent value of around cm-3 is found, with some trend to increase when the Si iv line intensity increases, although the uncertainties are quite large. The two lowest intensity classes provide only an upper limit to the electron density. An indication of a correlation between intensity and electron density has been reported by Griffiths et al. (1999). They found that electron density increases as intensity increases.
The line ratio O v 629.8 / N v 1238.8 is temperature dependent, but its value is sensitive to changes in the relative O/N abundance. This intensity ratio is density insensitive for cm-3. The ion fractions used to calculate the theoretical ratio come from Arnaud & Rothenflug (1985), and the relative O/N abundance is taken from Feldman (1992). Fig. 8 (bottom) gives the resulting values for the intensity ratio. It shows that the temperature is approximately constant with intensity, except for the lowest classes, corresponding to dark plasma outside the bright structures. The change in temperature is less than 10% even though the ratio has a strong temperature dependence. The measured electron temperature value is K; this value is very close to the O iv maximum abundance temperature in ionization equilibrium.
As O iv is formed at temperatures very similar to the value determined by the O v/N v ratio, we can determine the electron pressure using the measured electron temperature and electron density values. We assume that the quiet Sun plasma observed in the two different dates has on the average the same characteristics. The resulting electron pressure is between and K cm-3. This pressure range is about a factor 2 lower than the estimate by O'Shea et al. (1998) for active region conditions, and is in good agreement with the values obtained by Doscheck et al. (1998) and Griffiths et al. (1999) for several quiet Sun regions.
Using the derived electron density from O iv and temperature from the O v/N v ratio, it is possible to estimate a value for the `filling factor' from the emission line intensities from the fine structures (we have not used the two lowest intensity classes, which have large errors). We assume that the transition region at K is composed of isothermal loops with a 2 arcsec (1400 km) diameter. For N v, O v and Si iv, we find `fill factors' which range from to , correlated with intensity. The values for these three lines are nearly identical. These values are similar to those derived by Dere et al. (1987). If we attempt to derive a `fill factor' from the O iv lines, we obtain values which are two orders of magnitude smaller. This might suggest that the ionisation ratios for O iv are not at the equilibrium values. However, further work is in progress on this important subject.
Since the electron density which we derived (Fig. 8, top) stays approximately constant, we surmise that the increase in intensity is due to a greater number of unresolved filaments within the loop-like structures for the brighter network locations.
© European Southern Observatory (ESO) 2000
Online publication: June 5, 2000