Subject: Re: The meaning of variance maps

Hello Roland:

The present version of midisky (midi_b5) produces fits files
with two or four extensions. If no reliable source candidates
are found by midisky only a variance and an intensity map will
be output. The variance map are the raw counts backprojected to
a given sky pixel. If you take the squareroot of this number
you ought to get the sigma value for Poission statistics in the
intensity map. (I have not included a factor 1.06 as I should have
done to account for the extra variance introduced in the intensity
map when removing the background - this should be done). The
intensity map is essentially the same as the variance map after
fitting with some simple background model and finally removing
large scale structures by subtracting an average of 18 pixels
 - out of 20 selected in a fixed pattern around each pixel in a
pattern like this:
     B
    B.B
   B...B
  B.....B
 B.......B
B....X....B
 B.......B
  B.....B
   B...B
    B.B
     B

where X is the pixel to be coreccted and the B's are pixels used to
form the average. The two B-pixels with the highest values are excluded
from the average to avoid negative ghost patterns around strong sources.

You mention that the variance map you looked at had about 2400 counts
in the center this would correspond to an rms of 50 near the center,
but it could well be that the average rms over the whole field is
only 20 because over most of the field the number of counts are
much smaller than 2400.

When good source candidates are found two more extensions are
produced, one is a "residual intensity map" (the same variance map
still applies) and one is a "sources found" map.

I will be in ISDC on Wednesday and Thursday next week. I hope to
see you there and we can discuss how we should proceed.

niels




> Dear Niels,
>
> What Niels-Joergen proposed is probably close to what i am doing.
>
> What i still need is the meaning of the input images. In the files
> provided by Jerome there are two images for each pointing ("intensity"
> and "variance"):
>
>
> - the "intensity" is flat, with a mean of about zero and a standard
>    deviation of about 20.
>
>
> - the "variance" image varies between 0 at the border to about 2400
>    at the center of the field.
>
>
> iguess the"intensity" image is S_i but have no idea how to interpret the
> "variance" image. Could you tell me how those two images relate to S_i
> and sigma_i ?
>
> Thanks,
>
>
> --
> Roland Walter                      phone:  +41 22 950 91 28 (direct)
> INTEGRAL Science Data Centre               +41 22 950 91 00 (switchb.)
> Chemin d'Ecogia 16                 fax:    +41 22 950 91 33
> CH-1290 VERSOIX                    e-mail: Roland.Walter@obs.unige.ch
> Switzerland                        WWW:    http://isdc.unige.ch/