HOTSPOT HANDLING DISCUSSION 1

Date: Tue, 27 Jul 2004 13:26:37 +0200
From: Søren Brandt 
To: Peter Kretschmar 
Cc: oxborrow@dsri.dk, njw@dsri.dk, carl@dsri.dk
Subject: JEM-X hot spot handling

Peter,

We have discussed the handling of hot spots by Carol Anne (it should
work).
She wants to see some examples, but unfortunately they are often 
active during other peoples observations.
Recently there was a "good" one in orbit 217, pid 26, getting
quite strong in pid 27.
You may be able to check that it works ....

Søren

From: Peter Kretschmar 
Subject: Re: JEM-X hot spot handling
To: Søren Brandt 
Cc: oxborrow@dsri.dk, njw@dsri.dk, carl@dsri.dk

> We have discussed the handling of hot spots by Carol Anne (it should
> work).
> She wants to see some examples, but unfortunately they are often 
> active during other peoples observations.

   You could ask Chris/Arvind for permission to get the data
of selected Science Windows, then we could copy that data out.

> Recently there was a "good" one in orbit 217, pid 26, getting
> quite strong in pid 27.
> You may be able to check that it works ....

   Using OSA4 on the Ops. Network it seems to work quite well,
though a fraction of events remain unfiltered. See attachment
for an example - PID 26, the PID 27 looks very similar.

   Cheers,
   Peter     figure

Date: Wed, 28 Jul 2004 10:00:03 +0200 (MEST)
From: Carol Anne Oxborrow 
Subject: Re: JEM-X hot spot handling
To: sb@dsri.dk, peter.kretschmar@obs.unige.ch
Cc: oxborrow@dsri.dk, njw@dsri.dk, carl@dsri.dk

Dear Peter,

many thanks for the hotspot images, both with and without flagged events. This
shows me what I need to know, namely that the hotspot removal basically works.
The algorithm I've used to locate hotspots requires that a pixel meets certain
requirements of both excessive intensity, and having similarly excessive 
neighbours. In some regions at the edge of a hotspot one or other of these
requirements may not be met so that a small number of hotspot events slip 
through here. This is to avoid the program removing spots from the mask pattern,
say in a very bright off-axis source, which would be unhelpful to say the least.

I'm satisfied with the level of filtering that's shown in these images. It
should be remembered that the histogram viewer will always scale the image
so that the strongest pixel is a particular colour (white/lemon in this case)
no matter it's strength, so that the two pale yellow pixels left in the
filtered image (left) are far less populated than similar coloured pixels
in the right hand image. 

If there is anyone who is not satisfied with the current level of filtering,
please let me know and I can always turn down the criteria a little. However,
the more enthusiastically we harvest events from one place the more likely it
is that we will start taking innocent events from another place.

All comments on this subject are of course welcome.

Best wishes,

Carol Anne