[Fwd: Plate Scale for Hinode/XRT]

[Bish ishibashi <bish@cora.nwra.com>]

Hello, I am resending the email I sent out this morning (03MDT) since
it was too long and I am told that it was rejected by the mail list.
In any case, my apologies if you have somehow received twice.

----------------------------------------------------------------------


Hello, I realize that there has got to be a better thing
to think on Friday night. But it has been bothering me
a bit and I would like to know what others know about the
question on the proper plate scale used for Hinode/XRT.

In a XRT FITS header, the keyword PLATESCL is generally
set to 1.032"/pix. But is it really accurate to, say, a
tenth of a percent?

As many of you are aware, co-alignment of an XRT image
with an image taken with another instrument is proven to
be rather challenging. A chosen co-alignment method of
ours is to pick a SOHO/EIT 284AA (Fe XV, logT ~ 6.3) image
matching with an XRT image taken nearly at the same time,
and then co-align discrete XBP features (while avoiding
any major AR features) in these images spatially. While
doing so, I have noted that the plate scale of 1.032"/pix
does not work well. Say for instance if I am to pick one
discrete feature to co-align the two images and blink them
to see the result, I see decent co-alignment with the
discrete feature selected for spatial correlation, whereas
other discrete features on the XRT image radially move
toward the selected feature. This is a classic case of an
underestimated plate scale. Please see an animated gif
image (plscl1032.gif) below that demonstrates this point.

[the image is found at:

    http://space.mit.edu/home/bish/TMP/plscl1032.gif

]

So I have, via trials and errors, tuned the plate scale
to identify what correlates best with an SOHO/EIT 284AA
image. I should note here that the plate scale of an EIT
image is known to be 2.629+/-0.001"/pix (Auchere, DeForest,
and Artzner, 2000, ApJ, L529, 115). And that is what I
use for my coalignment program (I do fix all FITS/WCS
keywords in an EIT image according to the guideline given
by the FITSIO protocol). Anyway, the derived plate scale
via comparing an EIT 284AA image and an XRT (Al/Poly+Open)
image is 1.0417 ± 0.0005"/pix. The uncertainty is not
based on a formal statistical error; rather it is meant
to be the maximum probable inaccuracy associated with
spatial correlation [1]. Please see another animated gif
image for an improved co-alignment with the plate scale
of 1.0417"/pix below (plscl1042.gif).

[this image is found at:

      http://space.mit.edu/home/bish/TMP/plscl1042.gif

]

[1] the blinking method can usually tell if a discrete
     feature is shifted by a quarter of a pixel. So
     consider that as a total offset error accumulated
     across the FOV of an image. In this example the FOV
     of the XRT image used is 512x512, so the maximum
     error is 1.042" * 0.25/512 ~ 0.0005"/pix.


Having said all that, what is the most current value of
the plate scale for Hinode/XRT? I am aware that the
transit of Mercury event was used to measure a tentative
plate scale of Hinode/XRT (1.056"/pix, if I recall).
But I am not sure what details have been taken into
account (i.e., Hinode's orbital solution, etc) to derive
the value. I am not aware of the error value associated
with the measurement, either.

Could someone please share insights on the plate scale,
please?

Bish Ishibashi


PS. the EIT and XRT datasets used here were taken at 0106UT
     on 02-Feb-2007. I've done the same analysis on the XRT
     and EIT images taken at also 0106UT on 08-Nov-2006, which
     resulted in the same outcome.