Astron. Astrophys. 364, 859-872 (2000)

4. Location of the source of

In this paper we examine the CaXIX resonance line and the satellite lines. Plasma parameters are calculated from a best fit theoretical spectrum to the observed data. Two sample BCS spectra are shown in Fig. 3 from the time of maximum and the maximum in BCS CaXIX counts. The plots show that the observed data are well represented by a single component fit at both times. Two component fits to the data were attempted and although these yielded similar values, the calculated plasma parameters were unphysical: blue shift velocities were zero and . The BCS CaXIX lightcurve and the derived plasma parameters are shown in Fig. 4. The total count rate in the CaXIX channel has been corrected by a dead-time factor, which varies during the flare, and has a maximum value of two. This is large and results from the large count rate in the SXV channel of Counts per second. Since the SXV and CaXIX channels share the same detector the dead-time correction factor is calculated from the combined counts in the two channels.

Fig. 3a and b. Two sample BCS CaXIX spectra (histogram) and the best fitting theoretical spectrum (solid line). The times are at maximum (a ) and BCS CaXIX counts maximum (b ).

Fig. 4a-e. The results from the spectral fitting of the BCS CaXIX data.

Fig. 5 shows the relative timings of the non-thermal broadenings and the M1 channel HXR flux. The time of peak can be well determined for this event and is seen to occur after the first small HXR peak (18:20UT) and before t he maximum in HXRs.

Fig. 5. The relative timings of the non-thermal broadening and HXR flux.

We also note here that the wavelength shift of the main resonance line changes complementary to the throughout the flare (Fig. 4e). The spectral resolution of the CaXIX detector is Å, therefore the measured centroid shift is approximately two pixels (). If we assume that these line shifts are representative of bulk plasma motions associated with chromospheric evaporation the shift places a lower limit on the up-flow velocity.

The temperature responses of the SXT filters and the BCS CaXIX channel are shown in Fig. 6. This plot shows that although SXT is more sensitive to lower temperature plasma (i.e. ) it also detects plasma at CaXIX temperatures. Hence to locate the dominant source of CaXIX emission we use images taken by SXT through the two thickest filters, Al12 and Be119.

Fig. 6. The temperature responses of BCS CaXIX channel and SXT filters.

We use the SXT to determine the spatially resolved distribution of temperature and emission measure within the flare area. When generating SXT temperature and emission measure images the SXT images are summed over the same time range as the BCS data to improve the counting statistics. This increases the accuracy of the calculated plasma parameters and makes the values derived from both instruments directly comparable.

Fig. 7 shows temperature, emission measure and intensity images calculated from SXT from times around the maximum. Pixels in the SXT temperature maps that had a temperature within one standard deviation of the BCS temperature at that time were averaged to produce the derived SXT parameters. These pixels are enclosed by the contours in the SXT intensity image. The derived SXT temperatures are thus similar to that from BCS. If the emission measures from SXT are similar to that derived from BCS then we can assume that the two instruments are observing approximately the same plasma. At 18:20:47UT (Fig. 7a) the emission measures differ by less than a factor of 2 within the errors and at 18:21:29UT (Fig. 7b) the values are the same to within the errors. Thus early in the flare, around the time of maximum SXT and BCS are observing approximately the same plasma.

Fig. 7a and b. This figure shows an SXT intensity, EM and temperature image of the flare from (a ) 18:20:47 18:21:29UT and (b ) 18:21:2918:21:56UT the times around the peak. The pixels bounded by contours have a temperature within one standard deviation of the BCS temperature for the same time interval. The large error on the BCS temperature at 18:20:47UT is due to the large non-thermal broadening which makes it difficult to obtain a good fit.

Fig. 8 shows a plot of the SXT emission measure and temperature derived this way, along with the BCS temperature and emission measure, as a function of time. The plot shows that early in the flare the derived emission measures differ only slightly and change co-temporally, but as the flare progresses the differences become more significant. The BCS emission measures are also systematically higher than the SXT emission measures. The systematic differences imply that not all the hot component is seen by SXT, some of it is superimposed along the line of sight with cooler plasma (cf. Doschek 1999). The divergence indicates that this effect becomes stronger in the later stages of the flare. Cooler plasma along the line of sight will dominate SXT's response, resulting in lower derived temperatures. Hence pixels in which the hot component may be present along with cooler plasma will not show a high temperature in SXT and will not be within one standard deviation of the BCS temperature. These pixels will not therefore contribute to the SXT derived temperature and emission measure of the hot component but will be included in the BCS derivation, hence the discrepancy. Therefore, only before 18:23:00UT can we reliably say that SXT and BCS are observing essentially the same plasma.

Fig. 8. The variation of temperature and emission measure with time for the SXR flare loops, from SXT (thin solid line) and BCS (thick dashed line).

The hot pixels enclosed by the contours in Fig. 7a and b are spread over the whole flare area. However, the dominant emission will come from those hot pixels where the emission measure is also high. These high emission measure, high temperature pixels are located in an area extended along the flare loops above the footpoint emission.

Wülser et al. (1994) used a similar method of comparing SXT and BCS temperatures and emission measures in the early phase of a solar flare. In their analysis the derived plasma parameters from BCS and SXT differed by similar amounts to the values calculated here and led them to conclude that BCS and SXT were observing the same plasma. This enabled them to determine that the location of up-flowing SXR plasma, detected by BCS, was within a SXR flare loop connecting two HXR footpoints.

To verify our result that the BCS is detecting only the flare loop plasma and no additional sources, we compare the lightcurves of the BCS CaXIX with the SXT pixels (DN/s, Be119 filter) that have a temperature in the range of (the range over which BCS CaXIX is most sensitive). The results are shown in Fig. 9. The fact that these two lightcurves behave similarly (in the early stages of the flare) suggests that both instruments are observing the same plasma and that neither instrument is also detecting another source. Also shown in Fig. 9 is the lightcurve of the SXT cool component (), the accumulation of this lower temperature plasma inhibits the ability of SXT to detect the hot component, hence the discrepancies between the hot component and BCS lightcurves after 18:24UT.

Fig. 9. Comparison of the BCS CaXIX lightcurve (solid line) with the lightcurve of the SXT (Be199 filter) hot (dashed) and cool (dot-dashed) component.

It is now well established that during solar flares the TRACE 195 Å image becomes considerably contaminated by emission from FeXXIV (=192 Å, Feldman et al. 1999; Warren et al. 1999; Warren 2000). FeXXIV and CaXIX ionization fraction functions both peak at the same temperature (Arnaud & Rothenflug 1985), therefore emission in these two spectral lines should originate from approximately the same plasma sources. By comparing TRACE 195 Å images with TRACE 171 Å(FeIX ) and 284 Å(FeXV ) we can determine which features seen in the 195 Å image are due to FeXII and which are from FeXXIV . This is accomplished simply by assuming that any feature that appears in FeXII and not at either FeXV or FeIX must come from FeXXIV. In Fig. 10 TRACE difference images in each EUV filter are shown along side a temperature map from SXT from the time of maximum . The difference images were constructed by subtracting a pre-flare image from 18:00UT. These difference images illustrate that the hot SXT component sources are co-spatial with the FeXXIV emission in the TRACE 195 Å images. With the greater spatial resolution of TRACE we can say more accurately where, within the flare area, this FeXXIV emission comes from. It is slightly but significantly above the flare loops, offset to the south from the loop apex. This is co-spatial with the location of the loop top HXR source. Fig. 11 shows that the HXR loop top source is co-spatial with the hottest area of the SXT temperature map of 18:21UT and that this hot area (and HXR source, Fig. 2e) is located above the SXR loops that form during the main phase.

Fig. 10. TRACE difference images in all three filters and an SXT temperature image. The feature that appears in the FeXII filter and not the other two is judged to be an FeXXIV feature. Note that this is co-spatial with the hot component in the SXT temperature image.

Fig. 11. a The SXT temperature image of 18:21UT with the HXT LO channel image of the same time. b The SXT temperature image of 18:21UT with the SXR intensity image, as contours, during the main phase (18:25UT) showing the SXR flare loops. Contour values are 30, 50 and 80% of maximum.

We have shown using BCS and SXT temperature and emission measure comparisons and TRACE FeXXIV observations, that the dominant source of BCS CaXIX emission at the peak value of is located within and above the SXR flare loops. Fig. 8 shows the variation of temperature and emission measure of the flare loops (i.e. above the footpoints). The emission measure can be seen to increase steadily as the flare progresses. Fig. 12a and b show an SXT intensity, emission measure and temperature map at later times in the flare. These figures illustrate that the area occupied by the hot plasma is expanding. Note that after 18:24UT the BCS derived emission measure becomes significantly greater than SXT, this is because the increased amounts of cooler plasma component () masks SXT's ability to detect the hot component (cf. Fig. 9) as discussed by Doschek (1999). Fig. 13 shows selected SXT images during the rise phase of the SXR emission. The images show that the loops are becoming brighter. We showed in Sect. 3 that this flare is a two-ribbon (CSHKP) flare, this model relies on chromospheric evaporation to create the hot SXR loops. Fig. 8 shows that the emission measure in the loops is increasing, therefore according to the model the loops are filling with SXR plasma. Unfortunately no proper motions are apparent in the SXT or TRACE images. In the TRACE images this likely results from the relatively long exposure times (6 seconds) and inadequate temporal resolution (160 seconds). In the SXT images the spatial resolution (5 arc-seconds) may be too low to observe the evaporation flows.

Fig. 12a and b. SXT intensity, EM and temperature images of the flare from a 18:22:56 18:23:26UT and b 18:26:26 18:26:56UT. The pixels bounded by contours have a temperature within one standard deviation of the BCS temperature for the same time interval. Note in a the BCS and SXT emission measure and temperature are still similar whereas in b the BCS and SXT emission measures now differ significantly (cf. Fig 8).

Fig. 13a-d. Selected SXT images of the flare showing the brightening of the SXR flare loops. Images are individually scaled.

Chromospheric evaporation has become the generally accepted theory to explain the increase in emission measure during a flare. The evaporation of chromospheric plasma occurs as a natural physical response to the release of energy in the corona and the subsequent transport of energy down the field lines to the chromosphere. The plasma is then explosively heated to SXR temperatures and is forced up the flare loop. This process has been successfully modeled by many authors (Fisher et al. 1985a , 1985b , 1985c; Mariska et al. 1989; Yokoyama & Shibata 1998) and agrees well with the observations of SXR flares (Hori et al. 1997 , 1998; Yokoyama & Shibata 1998). The CSHKP model of two-ribbon solar flares, which is believed to well describe this flare, relies on chromospheric evaporation to fill SXR loops. It has been shown that the energy spectrum of the incident electron beam onto the chromosphere affects the efficiency of converting electron energy into chromospheric evaporation. The steeper the spectrum the more input electron energy will be used to drive chromospheric evaporation rather than be radiated or conducted away from the energy deposition site (Mariska et al. 1989; Antonucci et al. 1993; McDonald et al. 1999). Fig. 14 shows the values of the photon spectral index, , at various times during the flare. The photon spectral index is related to the electron energy spectral index (), assuming thick target HXR production, by; . The photon spectrum during the early phases of the flare 18:20:10 18:21:00UT is very steep, hence chromospheric evaporation will occur at this time. However we note that the value of the spectral index may be an over estimation due to contamination of the HXT LO channel by a thermal source.

Fig. 14. The HXT LO and M1 channel flux. Vertical dotted lines indicate accumulation boundaries within which the photon spectral indices were calculated.

The footpoint separation is from TRACE Lyman images. Assuming semicircular loops this gives a half loop length of . We assume that evaporating plasma travels up the loop at speeds ranging from a lower limit set by BCS observations of the line centroid shift () to an upper limit of the sound speed in the SXR loops. The sound speed, , is given by;

where the pressure, , the density, , n is the number density, k the Boltzmann constant, the ratio of specific heats and the ion mass. This expression reduces to,

The temperature of the flare loops is calculated from SXT observations to be which gives a sound speed of . Therefore the time taken for evaporating SXR plasma to reach the loop apex is . During the early stages of evaporation the pressure in the pre-flare loop will be at its lowest level and the pressure in the evaporating plasma will be high. Therefore the up-flow velocity, driven by the pressure gradient, will be at its highest value, close to the sound speed. Hence after the start of evaporation, it will take only for plasma to reach the flare loop apex. Therefore the small HXR burst at 18:20UT (Fig. 5) is consistent with the presence of evaporating plasma near the loop apex at 18:20:30UT.

© European Southern Observatory (ESO) 2000

Online publication: January 29, 2001
helpdesk.link@springer.de