Our findings indicate that the X-ray nature of an old, docile active region is considerably different from that of young, "active" active regions. A key differing parameter appears to be the frequency of strong microflares, which are ubiquitous in young active regions, and virtually absent in the region studied here. Some microflares are the source of the hotter temperature component ( 5 MK in active regions (e.g., Watanabe et al. 1995, Yoshida & Tsuneta 1996, Sterling et al. 1997; see also Feldman et al. 1996), and their paucity in our old active region results in there being no hotter component observed in the SXV spectra. Our hottest temperatures in this region only reach about 3.0 MK (excluding the brief period of higher activity), and still cooler plasmas exist in the active region also as evidenced by the SXT-derived temperatures of Fig. 3c.
We are not, however, able to say whether all microflares are absent in this old active region. Shimizu (1995) and Feldman et al. (1996) indicate that some microflares are extremely weak and have MK, and some such low-level microflares may be included over the long integration times we use in our analysis. Nonetheless, Fig. 3a shows that there are very few microflares distinguishable above the GOES background level over the time range of our study here. We can therefore say that the frequency of microflares detectable in GOES , i.e. those which were found to be abundant in young, "normal" active regions and often could be linked to the MK hot component, are suppressed in this older active region.
Sterling (1997a) found average SXV values of 5.5-6.2 MK for a younger (and substantially smaller in spatial extent) active region of 1996 March (Porter & Klimchuk, 1995, find similar temperatures for active region coronal loops). That study did not find temperatures as low as those we find here, but there were many microflares occurring throughout the life of that region, and so it is possible that Sterling (1997a) integrated over both hotter and cooler component temperatures when forming spectra in that study. A typical integration time outside times of brighter microflares in that study was 1000-2000 s, and so there may have been several low- to moderate-level shorter-lived microflares occurring during the integrations. It may also be that such microflares occur at such a high frequency in young active regions that their coronal plasmas never have time to cool. Shimizu (1995) gives typical cooling times for microflares to be 40 s due to conduction, and so microflares occurring on this timescale would keep the plasma heated (the radiative cooling time scale is much longer, but the conductive energy loss is two orders of magnitude larger than the radiative loss in his analysis). Since this time scale is shorter than the typical BCS integration times used, we cannot be certain that the 1996 March active region did not have temperatures as low as those we find here.
There is much better agreement between our SXV temperatures here and those of Sterling et al. (1997), who found 5 MK for the isolated upper portions of the same active region studied in this paper when it was very young (about nine days after it became prominent in X-rays). They also found higher temperatures ( 5 MK at lower altitudes at that time for the region, consistent with the finding that microflares in young active regions are generated at low altitudes (Sterling et al. 1997; Sterling 1997b). The microflares would probably be due to interaction and reconnection between emerging flux and pre-existing coronal fields (e.g., Heyvaerts et al. 1977; Yokoyama & Shibata 1996; Canfield et al. 1996). We believe, therefore, that both Sterling et al. (1997) and our current work see the cool temperature coronal component alone, but for different reasons: Sterling et al. (1997) see it because they are looking high in the active-region corona while the microflares producing the high-temperature component are restricted to heights occulted by the solar limb, and we see only the cool component here because there are no longer enough microflares to produce (or maintain) the hot component.
By using the occultations of active regions by the solar limb, Sterling (1997b) and Sterling et al. (1997) concluded that seen in SXV decreased with height in active regions. We do not see any evidence for a decrease with height in the region studied here. In fact, in contrast to our previous studies, close inspection of Fig. 3c shows a slight downward trend in the SXV values between September 17 and 20, and perhaps a slight increasing trend between October 1 and 4. These time periods respectively correspond to when the active region is emerging from the east limb and going behind the west limb, and therefore the temperature trends would suggest an increase in temperature with height in this region. Such an increase in with height would be expected if, e.g., the region consists of simple, large-scale loops (e.g., Wragg & Priest 1981, Priest et al. 1998), without a hot source (such as microflaring loops) at low altitudes. Moreover, very weak, diffuse regions have been found to show an increase in temperature with height (Sturrock et al. 1997; Wheatland et al. 1997). We do not, however, see a trend with height of the SXT temperatures in Fig. 2c, so the trends seen in SXV could be an aspect of the slightly hotter plasma seen in SXV , the longer averaged integration times of SXV , or they could be a systematic artifact of the BCS analysis. For example, similar to the widths of the spectral lines discussed earlier, the shape of the spectra will also depend on the north-south distribution of the intensity of the emitting source, and this distribution will change between when the region is partially occulted and when it is on the disk. This change may alter the spectral shape only modestly, but perhaps enough to generate the weak trends in temperatures we see. We surmise that the trends in temperature and the possible relation to the spectral shape are too subtle for effective analysis, and we are therefore unable to draw strong conclusions about the temperature structure with height in this region. We can only say that any temperature variation with height is weak at best.
As noted in the Introduction, we could not obtain spectroscopic temperatures from BCS for individual solar structures during solar maximum since BCS is a full-Sun instrument. But we can use some of our quiet-Sun results to speculate about the thermal properties of solar features during enhanced-activity periods by comparing intensities of various structures seen in SXT. Fig. 5 shows an AlMg SXT image from near the previous solar activity maximum. Wheatland et al. (1997) analyzed the area labeled as the "Diffuse Region" of this figure. In Fig. 5 that area has an intensity level of 5 MK SXT DN s In comparison, our active region in Fig. 1b has an intensity level of 50-250 DN s and the region studied by Sterling (1997a) (Fig. 1 of that paper, also taken with the AlMg filter), has intensity of about 5000-25 000 DN s In Fig. 5, regions labeled "Diffuse Loops" have intensities similar to that of the active region in Fig. 1b, and regions labeled "Bright Loops" have intensities similar to those of the region in Sterling (1997a). Based on this, we speculate that the Bright Loops of Fig. 5 contain both hot and cool temperature coronal plasmas with SXV values around 5.5-6.2 MK (Sterling 1997a), while the Diffuse Loops consist of the cool component only with SXV values similar to those we found in Fig. 3c (-3 MK). Although we do not directly see in SXV features which have SXT intensities corresponding to the Diffuse Region of Fig. 5, we would guess that they would have SXV temperatures slightly lower than that of the Diffuse Loops, based on an intensity comparison. In this way, using the SXT-intensity SXV -temperature relationship for isolated features on the Sun allows us to estimate the SXV -temperatures for features which cannot be individually resolved by BCS.
This methodology can also be used to address what seems to be a contradiction in some of the previous spectroscopic active region studies. Sterling (1997a) found that SXV electron temperature (average values of 5.5-6.2 MK) from an isolated active region during solar cycle minimum was correlated with the SXV intensity. Similarly, Watanabe et al. (1995) found that the temperature of the corona near solar maximum was higher at a time when the overall coronal flux was higher ( MK), compared to the temperature at a time two months later ( MK) when the overall coronal flux was lower (but still higher than the coronal flux during solar minimum). The apparent contradiction comes from the fact that the SXV flux of the solar-minimum-period active region studied by Sterling (1997a) was lower than the flux of the corona during the time when Watanabe et al. (1995) measured the MK temperatures. If the flux-temperature correlation holds generally, we would expect that the temperatures seen by Sterling (1997a) to have been lower than those of Watanabe et al. (1995). Sterling (1997a) speculated that the reason his temperatures were higher than Watanabe et al.'s (1995) may be because, in addition to active regions, there was much more diffuse corona present during the time of the Watanabe et al. (1995) study, and that this diffuse corona may have had a lower temperature than that of Sterling's (1997a) active region's corona. When averaged over the entire corona, the overall temperatures would be reduced by the cooler background corona. Our results from the present study support this picture (we are assuming that the SXT and SXV intensities scale in a similar fashion, which seems reasonable). That is, it is indeed likely that the diffuse background corona in the Watanabe et al. (1995) study had temperatures comparable to those we find in Fig. 3c (and presumably those of the "Diffuse Loops" of Fig. 5 also), bringing down the average temperature of the overall corona in their study.
As Yohkoh was launched during the height of solar maximum, it was not possible to see how the diffused regions of the Sun developed. We suspect that much of it comes from old active regions-such as the one studied here-that lose their hotter temperature component, and then become diffuse, e.g., due to the dispersal of the active-region magnetic elements in time as discussed by Leighton (1964). In addition, there may be connections made between these old active regions with different active regions (Tsuneta 1996), thereby forming larger-scale quite Sun coronal features. Another factor possibly contributing to, or augmenting, the magnetic interactions is that the active region studied here was an old solar cycle (cycle 22) region occurring at the start of the new solar cycle (cycle 23), and thus there were magnetic field polarities corresponding to both cycles on the Sun when the region was present (Harvey & Hudson 1998); this factor could distinguish this region from other active regions occurring well within a single cycle. Currently ongoing Yohkoh SXT observations during the buildup of the new solar cycle hopefully will provide us with an opportunity to see if quiet Sun regions near solar cycle maximum actually do develop from old active regions, as suggested by our work here.
In addition, future analysis of observations from the SOHO and TRACE satellites will certainly help us in evaluating many of the topics discussed here. Some preliminary progress has already been made. For example, using SOHO 's Coronal Diagnostic Spectrometer (CDS) instrument, Fludra et al. (1997) and Matthews & Harra-Murnion (1997) discuss relationships between transition region and coronal structures in the same active region we observe here (but during August 1996, when it is in a more active state than discussed in this paper).
© European Southern Observatory (ESO) 1999
Online publication: June 17, 1999