This is to clear up some mis-conceptions about dmarfadd in the
attached and related emails. (I haven't scanned further to see if
there was any followup since).
mkarf and mkgarf handle all detector nonuniformities including
badpixels and chip gaps, and more recently, QE non-uniformities.
These tools average over the dither, mapping detector pixel to sky and
integrating.
mkgarf and mkarf do not include any psf fraction in the region. They
assume 1.0. (and hence are not designed for extended sources)
Dmarfadd was designed for grating arfs, since each chip is an
independent detector, with it's own unique GTI. It is not meant to
add orders, but to stitch the responses together along the orders,
weighting properly by exposure. mkgarf is run per order per grating
per chip. dmarfadd sums a grating, chip set for an order.
Dmarfadd as originally designed almost works for imaging arfs. A few
header-keyword restrictions have been removed, but the documentation
has a big caveat about whether averaging is best as opposed to joint
analysis w/ separate responses, e.g., if dithering between a FI and
BI chip.
I think mkarf is the tool Pat is looking for, but perhaps didn't know
that it averaged over the aspect. mkgarf is preferred for spectral
analysis, though if you don't have many photons and have to assume a
source spectrum, exposure maps can be used. (with some care in the
spectral weighting.)
--Dave
Antonella> ------- Forwarded Messages
Antonella> To: chandra-users@head-cfa.harvard.edu
Antonella> Subject: dmarfadd seems completely unreliable
Antonella> cc: acis_observer@astro.psu.edu
Antonella> Mime-Version: 1.0
Antonella> Date: Tue, 31 Oct 2000 08:35:53 -0500
Antonella> From: Pat Broos <patb@pokum.astro.psu.edu>
Antonella> WARNING to those of you using the tool dmarfadd!
Antonella> If the keywords TG_SCRID or TG_PART are missing from
Antonella> your ARF's (which they are from mine -- who knows why),
Antonella> then dmarfadd will print error messages, but will still
Antonella> produce an output file. It turns out that the output
Antonella> file is simply a copy of the first ARF in your list of
Antonella> ones to add up. Thus one gets incorrect output,
Antonella> instead of the preferred no output at all.
Antonella> I intend to stick to the well-behaved FTOOL addarf for
Antonella> the time being. Addarf is also more flexible, allowing
Antonella> you to average ARFs instead of adding them -- something
Antonella> very useful if you care to worry about the fact that
Antonella> real sources are extracted over a finite region of the
Antonella> detector (which might include a chip gap where the
Antonella> effective area is changing on scales of 10's of
Antonella> pixels), whereas mkarf computes an ARF for a single
Antonella> point on the sky.
Antonella> Regards,
Antonella> Patrick Broos
Antonella> ------- Message 2
Antonella> From: "Keith A. Arnaud" <kaa@genji.gsfc.nasa.gov>
Antonella> To: chandra-users@head-cfa.harvard.edu, acis_observer@astro.psu.edu
Antonella> Subject: Re: dmarfadd seems completely unreliable
Antonella> Pat,
Antonella> It is worth noting that averaging ARFs is only valid if
Antonella> the vignetting does not vary significantly over the
Antonella> region from which the spectrum was extracted. For
Antonella> large regions the correct procedure is to raytrace a
Antonella> spatial model of your source. I'm curious to know
Antonella> whether anyone is actually doing this.
Antonella> Keith Arnaud
Antonella> (Admittedly it took about 5 years for us to add this
Antonella> ability to the ASCA s/w so I don't expect this to be a
Antonella> priority, especially since Chandra's strength is small
Antonella> sources. I expect this to be more of an issue with
Antonella> XMM-Newton).
Antonella> ------- Message 3
Antonella> Date: Tue, 31 Oct 2000 11:52:12 -0500 (EST)
Antonella> From: Jonathan McDowell <jcm@head-cfa.harvard.edu>
Antonella> Message-Id: <200010311652.LAA15893@urania.harvard.edu>
Antonella> To: chandra-users@head-cfa.harvard.edu
Antonella> Subject: dmarfadd problem
Antonella> Pat,
Antonella> I apologize for the hole in our documentation. dmarfadd
Antonella> in CIAO1.1 was developed at the last moment to fix a
Antonella> hole in the grating thread - it was intended only for
Antonella> grating observations but unfortunately this was not
Antonella> made clear in the overhastily constructed help file.
Antonella> I am informed that the forthcoming CIAO2.0 version of
Antonella> the tool can handle imaging ARFs and does not have the
Antonella> keyword problem you mention. The documentation is also
Antonella> improved.
Antonella> The second part of your complaint - that we don't
Antonella> spatially average - I agree with, and is related to the
Antonella> concern among some CXC scientists that the spatial
Antonella> averaging is the 'wrong' thing to do. We do have a
Antonella> package (Alexey Vikhlinin's calcrmf ARF/RMF tools) on
Antonella> the swap page
Antonella> http://asc.harvard.edu/cgi-gen/contributed_software.cgi
Antonella> which does support the spatial averaging capability,
Antonella> but there's no arfadd equivalent there. Can you explain
Antonella> exactly what you were trying to do? (the motivation
Antonella> originally for dmarfadd was to let you add orders in
Antonella> grating data; the path to obtaining an ARF averaged
Antonella> over a region for imaging data is to use the calcarf
Antonella> from the Alexey package, which Alexey and I have
Antonella> modified to work within CIAO1 smoothly and which
Antonella> includes a documentation thread.)
Antonella> In future, I hope you'll also report such problems to helpdesk
Antonella> so that we can try and address them more directly.
Antonella> - Jonathan
Antonella> ------- Message 4
Antonella> To: Jonathan McDowell <jcm@head-cfa.harvard.edu>
Antonella> cc: chandra-users@head-cfa.harvard.edu
Antonella> Subject: Re: dmarfadd problem
Antonella> In-reply-to: Your message of "Tue, 31 Oct 2000 11:52:12 EST."
Antonella> <200010311652.LAA15893@urania.harvard.edu>
Antonella> Mime-Version: 1.0
Antonella> Date: Tue, 31 Oct 2000 12:28:42 -0500
Antonella> From: Pat Broos <patb@pokum.astro.psu.edu>
Antonella> Hi Jonathan,
Antonella> Thanks for your helpful message. Before I talk about
Antonella> the technical stuff I should apologize for the title of
Antonella> my original message "dmarfadd seems completely
Antonella> unreliable". When I _started_ the email (and typed the
Antonella> subject) I didn't yet realize that my output was a copy
Antonella> of the first file -- I thought the computations were
Antonella> just hosed (unreliable). :)
Antonella> Anyway, the issue we're currently struggling with here
Antonella> is how to generate reasonable ARF's for point sources
Antonella> that fall in the chip gaps, where the EA (for a single
Antonella> CCD) can fall from nominal to zero in ~35 pixels. I
Antonella> know the "right" way to do it is extraordinarily
Antonella> complex -- we need something reasonable now. My
Antonella> thought was that computing several ARFs across our
Antonella> extraction region and averaging would be a reasonable
Antonella> 1st order estimation.
Antonella> And, of course, even if one chooses to accept ARFs
Antonella> calculated at single points on the sky, there are lots
Antonella> of such locations that have data from multiple CCDs.
Antonella> The user is forced to run mkarf on each CCD involved,
Antonella> then sum them. (By the way I don't see why mkarf was
Antonella> written to deal with only one CCD at a time.)
Antonella> And, of course, even if one can ignore chip gaps, if
Antonella> you have multiple observations of the same field and
Antonella> choose to fit merged spectra (rather than simultaneous
Antonella> fits of spectra from each observation), then one must
Antonella> sum ARFs.
Antonella> At PSU these are a non-trivial issues. Orion & HDF
Antonella> fields have hundreds of sources, and both are
Antonella> multi-obsid campaigns. The fraction of the merged
Antonella> field covered by chip gaps is substantial, and for dim
Antonella> sources it makes sense to work with merged spectra
Antonella> (requiring summed ARF's).
Antonella> Thanks for the pointers to the swap page -- I will take
Antonella> a look at what others are trying.
Antonella> Regards,
Antonella> Pat
Antonella> ------- End of Forwarded Messages
--
Dave
David Huenemoerder (617-253-4283; fax: 253-8084)
Center for Space Research/Chandra Science Center
MIT NE80-6023, Cambridge, MA 02139
http://space.mit.edu/~dph
This archive was generated by hypermail 2b29 : Tue Feb 12 2013 - 01:00:04 EST