There are two ways in which SXT images might be saturated (see the Instrument Guide). In the case of full-resolution images, saturation most likely occurs in the individual CCD pixels. In this case, the charge that is read out of the device is conserved, but it is distributed incorrectly. In the case of half- or quarter-resolution images, the saturation is most likely in the on-chip summing well. Pixels that are saturated will not produce a meaningful temperature and emission measure estimate. Because of the nature of this problem, there is no means to ``restore'' a saturated image to what it should have looked liked when the photons were initially absorbed in the CCD.
SXT_PREP optionally returns an sat array which will flag saturated pixels. It calls SXT_SATPIX to do this and this will only work if data passed to SXT_PREP have not been decompressed or background subtracted. If the sat array is passed into SXT_TEEM, then the routine will sum all the saturated pixels together and treat them as a ``macro" pixel. The output Te array will contain the resulting temperature and the EM array will contain the corresponding emission measure per pixel. For flare data which are acquired in full-resolution mode, this approach will give a mean estimate for the temperature and emission measure.
Because of the low gain of the SXT CCD camera, the ratio of DN/photon (depending on the filter) is between approximately 2 (Al.1) and 4 (Be119) for . In full-resolution images, the CCD full-well corresponds approximately to DN = 3313. Normally, the exposure time is automatically adjusted so as to keep the peak intensity at approximately half the full-well capacity. So it is difficult to obtain accurate temperature and emission measure estimates in the faint portions of flare images which have inherent dynamic ranges exceeding 100.
One way to improve the signal-to-noise ratio is to trade spatial resolution
for photon flux, for example, by rebining the image. A procedure that does this
is SXT_SUMXY. The calling sequence is:
Another method is to use a user contributed routine called GO_TEEM which allows the user to specify an arbitrarily shaped polygon region.
SXT_TEEM does not make any corrections for scattering or the effects of the point-spread function. The X-ray scattering is a function which increases with X-ray energy (see the Instrument Guide). As a result, scattering is expected to be a bigger problem for the thicker filters. The user of SXT_TEEM should realize that this will sometimes result in higher than expected temperatures at the edges of bright flare kernels.
As mentioned in the Instrument Guide, the Neutral Density filter increases X-ray scattering. Even though the amount is small (about 0.7%), and therefore, does not affect the results in the bright portions of a flare, it can significantly alter the level of the faint emission outside the flare region. One should be careful, therefore, about the interpretation of temperatures in faint regions when one of the filter pairs includes the neutral density filter.