Hi all,
This is just a general plea (and a warning to the
unwary) about accounting for the PSF fraction in spectral fitting of
point sources.
A quick glance at figures 4.6 and 4.9 of the Proposer's Guide illustrates
the problem, which probably affects most observations and is nowhere
discussed in the CIAO threads or analysis guides as far as I can tell,
even at the basic level of suggested minimal extraction regions as
a function of off-axis angle.
Currently, mkarf assumes 100% of the photons detected by chandra
are included in the extraction region. Even for a 2" radius
on-axis extraction region, only 95% of 1.5 keV photons and
85-90% of >4.5 keV photons are within the extraction region.
If your source is in a crowded region or surrounded by bright extended
emission (which for me is almost always the case, and is why Chandra's
resolution is needed), you may want to
restrict your radius to < 1" (not to mention using a tight
annular region for the background, which will contain the
wings of you PSF), in such cases, the energy dependence
of the PSF cannot be ignored if you want anything
at all believable from your spectral fits. Off-axis, this is much
worse. If you have a bright, piled-up source, the most straight-forward
way of dealing with it is to exclude the central pixels where the
pile-up occurs and just fit the wings, then you don't have to worry about
squirly pile-up models, but you really need to account for the
energy dependence of your PSF.
Currently, the only way I know of dealing with this is
to use mkpsf to create a series of model PSFs at various energies,
measure the enclosed fraction at each energy, fit a function to this
and create a correction table, then manually apply the
corrections to your arf file. This seems like something that should be
included automatically in mkarf. Even if the model PSFs aren't well
calibrated, not applying any correction is much worse. At the least, it
seems like a big WARNING should be included in the point-source spectral
thread, and a thread developed for performing a correction such as I
outline above using CIAO (I use a combination of CIAO, FTOOLS, and MIRIAD
to do this, which would be very inconvenient for the average user).
Cheers,
Mallory
This archive was generated by hypermail 2b29 : Thu Dec 13 2012 - 01:00:08 EST