Subject: Re: Jem-X light curve



Hello Stephane:

Thanks for your mail and your analysis of the Crab light curve
from orbit 43(?) using j_ima_iros.

The comments below are based solely on your mail and the three
figures you attached. I have not studied the complete output of
your run.

I dont want to dismiss your results, but I am sure that you are
trying to use j_ima_iros in a way for which is was not intended.

At the outset you should realize that a program like j_ima_iros
when searching an image for a potential source peak will search and
iterate to optimize (actually maximize!) the source signal. So you
must expect that as the signal to noise ratio degrades the results 
will be gradually more and more skewed in the direction of over-
estimating the flux. I am not sure that this is a complete explanation 
of the gradual increase in flux you observe, but I just want to 
emphasize that j_ima_iros knows nothing about the Crab, and has no 
preference for locating a source at its nominal place.

I have been curious about the many erratic positions and flux values 
from your run. But I suspect that you have defined a sequence of 
identical time intervals in which you want to build your light curve. 
Unfortunately the INTEGRAL observation style with its relatively short
'science windows' does not match well with this type of interval 
definitions. In many cases you will get only a very short useable
time interval within your time window, and I suspect that this is
the cause of the many poor fits even during the first pointings
where the Crab is apparently nearly on axis (hex dither pattern?). 
You should check (in the log file) the total number of counts used in 
each of your science windows. You should also be aware that the 
on-board data buffer used by JEM-X means we often loose data from
the last few 100 seconds of a science window, you cannot assume 
that the available JEM-X events are distributed uniformly along
the duration of the science window. The JEM-X software will
correctly compensate for this when calculating fluxes - but it
cannot compensate for the absense of sufficient data in one of
your light curve interval - there it will just produce very poor
results. 

But another major problem making light curves with JEM-X is the
very strong vignetting effect acting during the majority of the 
pointings in a 5 by 5 dither. This means that for the majority of
the pointings you cannot get reliable results even with 1000 s
of integration time per point. And in the corners of the 5 by 5
there is even no guarantee that you find the Crab in a 2200 s
science window.

So the conclusion is the that j_ima_iros is not a good tool for
light curves. For that purpose should use the light curve
program written by Stefan Larsson, or alternatively use software
which extract the source flux from the images produced by j_ima_iros
but making use of the known source position. This conclusion
must be described in some detail in our user manual.

I shall be interested to know if my assumptions about your procedure
are reasonably correct - or we maybe have a more serious problem.

with best regards

niels