Astron. Astrophys. 346, 995-1002 (1999)

3. Results

Fig. 3a shows soft X-ray fluxes over the times of our study from the 1-8 Å channel of the GOES  9 satellite, with the standard letter designations for the intensity levels given on the right-side axis. The active region was completely behind the Sun until about September 16, and its emergence is marked by an increase in the GOES flux. This shallow increase in flux is in contrast to sharp changes in the flux seen when the region was brighter earlier in its life (e.g., Hudson et al. 1998; Dryer et al. 1998), suggesting a more extended and dimmer emitting volume in its later phases. There is only a slight increase in the GOES level between the times of our background spectrum (which lacks spectral lines) from September 13 data, and the times when the active region is on the disk and emission lines are prominent (see Fig. 2a). By the end of the period, on October 7, the region has rotated to the far side of the Sun again. There is a gap in the GOES data near the time of the disappearance.

Fig. 3a-c. Time variation of X-ray properties of the late-phase active region. a  GOES (1-8 Å) flux variation with time, with the standard GOES alphabetical classification on the right-hand side axis. The diurnal modulation is a satellite artifact. b  Flux derived from BCS SXV spectra (total flux seen in the channel after removal of an appropriate background spectrum for each datum). c  Electron temperatures from the SXV spectra (asterisks and triangle), and from SXT (circles). The value represented by the triangle is plotted at 0.5-times its actual value in order to fit on the plot. The SXV and SXT properties are not calculated for a time period around September 28, since at that time there is a flux enhancement from an area of the Sun not associated with the late-phase active region of this study.

A movie constructed from full-frame SXT images shows that there are two brief time periods of substantial flaring activity. The first of these appears as a flux enhancement in the GOES data on September 28, and is due to a small spatial scale brightening that occurs in a region separated from the active region of our study. We omit data from this secondary feature in our analysis. The second activity brightening occurs on October 5 (there is a gap in the GOES flux at that time in Fig. 3a), when the active region had rotated around the west solar limb, and was associated with the eruption of a large-scale coronal mass ejection (CME) seen in SXT (Watari et al. 1997).

There is a steady oscillation in the GOES flux with period of about one day. We believe that this is due to the low solar flux level and the response of the GOES sensors to particle background. The peaks correspond to passages of the GOES satellite through the anti-solar direction. Since GOES is geostationary, it passes though the same magnetospheric tail region roughly once every twenty-four hours, since the belts are approximately stationary with respect to the Sun-Earth axis.

Fig. 3b shows the SXV flux as a function of time. There are no measurable emission line spectra attainable prior to our first datum on September 17. After a rise in intensity between September 17 and 20 as the region emerges from the East limb, the flux remains very constant. It begins to decrease near the end of the observation period, but there is a strong flux enhancement at the time of the CME-related event on October 5.

Fig. 3c shows values derived from the SXV spectra as asterisks and a triangle. We have determined uncertainties assuming two sources of errors summed in quadrature: the 1 error associated with the fit to the spectra, and the spread in the resulting temperatures assuming a % change in our primary estimate of the background level (i.e., the background multiplication factor described in Sect. 2). These uncertainties average 0.1 MK, with the largest being 0.3 MK; the uncertainty in the background level contributes most to these values. We do not plot the resulting error bars since they are generally comparable to the size of the asterisks, with the lowest-temperature values generally having the largest errors. The plotted temperatures are the averages of the three values obtained by assuming our primary background level estimate, and the temperatures obtained assuming a % change in the background estimate. One of the temperatures, represented by the triangle, is plotted at 0.5 times its actual value; it is associated with the October 5 CME event and is consequently much higher than the other temperatures. For the 38 data points between September 17 and October 4, the SXV -derived electron temperatures range from 2.26 to 3.03 MK, with an average value and 1 sample standard deviation of  MK.

SXT temperatures are plotted in Fig. 3c as filled circles, and average  MK. Statistical uncertainties in the derived SXT temperatures are less than 0.05 MK for all of the points.

SXT and BCS are sensitive to plasmas of differing temperatures, as indicated by their respective signal response functions shown in Fig. 4. The SXT functions are for the two filters used in our analysis, and are derived by convolving the instrument response functions with the theoretical X-ray emission line spectra of Mewe et al. (1985) and the continuum expression of Mewe et al. (1986), as discussed in Tsuneta et al. (1991); these curves use DN s^-1 units, where 1DN=100 electrons created by the CCD (see Hara et al. 1992). For SXV Fig. 4 shows the contribution function given by

where is the fractional ion abundance in ionization equilibrium, taken from Arnaud & Rothenflug (1985), and is the transition energy for the line formation. The two SXT response functions decrease with decreasing temperature below about 5 MK, while that of SXV peaks near 15 MK (e.g., Culhane et al. 1991) and falls more sharply with decreasing temperatures than do the SXT filters. Therefore SXV tends to pick out higher temperatures in the active region than does SXT, and SXT is more sensitive to cooler temperatures than is SXV . So the differing SXT and SXV temperatures can be explained in terms of a multithermal active-region plasma. Fig. 4 also indicates that there is little or no material hotter than the SXV values given in Fig. 3c, since SXV would have a much higher sensitivity to any such hotter material than it does to 2-3 MK material. That is, if a substantial amount of hotter material were present, SXV would yield values higher than the 2-3 MK we find.

Fig. 4. Plots of the sensitivity functions as functions of temperature for the SXT Al.1 filter (thin solid line), AlMg filter (dashed line) and the contribution function for the BCS SXV resonance line (thick solid line). The scale for SXV (right-side) covers many more decades than the scale for the SXT filters (left-side).

We calculate emission measures,

where is the electron density and V is the volume of the emitting plasma, using the spectral fit outputs for SXV and the ratio of the two SXT filter responses for SXT (e.g., Hara et al. 1992). Average log emission measures are for SXV and for SXT. These values are consistent with studies of active-region loops with SXT. Cargill & Klimchuk (1997), for example, deduce linear emission measures of 0.25-100 cm Those loops had lengths  cm, so that an area factor of imply volume emission measures of - cm Klimchuk & Gary (1995) also find emission measures of about - cm and Kano & Tsuneta (1995) find similar values. Thus our values for the volume emission measures fall near the upper end of the range of those from those studies, assuming the same area factor as above. Our values are about an order of magnitude higher than some other active region studies, e.g., Pye et al. (1978) found cm^-3 in the 2-3 MK range using Skylab data and Saba & Strong (1991) found about cm^-3 for an active region at about 3 MK using MgXI spectra from the FCS instrument on SMM. The difference is likely due to the large, diffuse volume of the region we study here.

Because of SXV 's strong bias toward detecting the highest temperatures in the non-flaring active region, and since the emission measures of the cooler material detectable by SXT are not much larger than the SXV emission measures, we expect that the SXV spectra are not strongly affected by contributions from the cooler material. Thus our single-temperature component assumption used in our fits to the spectra (see Sect. 2) should be a good one.

Non-thermal line broadenings are commonly observed in solar flare spectra (see, e.g., Khan et al. 1995; Mariska et al. 1995; and Harra-Murnion et al. 1997, for discussions of non-thermal broadening in flares obtained from Yohkoh 's BCS). There is a difficulty in measuring line widths with BCS for distributed sources, however, as the line widths are affected by the north-south distribution of the emitting source, due to the geometry of the BCS crystals and their mounting orientation on the spacecraft. In the SXV channel, this broadening due to instrument orientation occurs at the rate of 1.2 km s^-1 per each 1" in the north-south direction (Mariska 1994; Sterling 1997a), where the excess broadening is expressed in terms of an equivalent turbulent velocity. Because of the large extent of the source indicated by the SXT images, we are not able to reliably separate the physical line broadenings from the instrumental line broadenings for this region.

By observing a much more compact region than that of the study here, Sterling (1997a) was able to examine active region non-thermal line broadenings using BCS SXV data, obtaining an average value of  km s SXT images of that active region at the 50% contour level corresponded to a SXV line broadening of 15 km s^-1, which was small compared to the derived non-thermal velocity values. That region, however, was not as quiescent as the current one is, and so some short-duration, relatively low- or moderate-level (although possibly higher-level than the background GOES level in Fig. 3a) microflares may also be included in that analysis.

© European Southern Observatory (ESO) 1999

Online publication: June 17, 1999
helpdesk.link@springer.de