Re: V image fainter than I

[droege@snapmail.us]

Paul and all,

The current coded scheme is that we must have both a V and an I detection
to keep the measurement.  This has the advantage that it does a pretty
good job of getting rid of cosmic rays.  To try to make up for the losses,
I set the detection limit low and accept a few false detections from the
noise.  This should just produce a background smear of dim detections at
random points in the sky.  There are not very many of these as a cut on
single hits at low brightness will reveal.  Well, there might be a few
million.  I have never counted them.

A disadvantage is that we lose red stars when the V measurement drops
below the limit of detection.  For slowly varying red stars the curve
disappears when there is still obviously a lot of I data left.  Given all
the data on such a star one could get a pretty good guess on what to do
with the non paired measurements if we kept them.

I admit that the way we are running throws away a lot of data.  If someone
is interested in this data, I will welcome code that processes this data
to files.  Now is the time to do something.  By fall I will be running 10
cameras for production and will not want to change anything.

I will still keep raw images, darks and flats on DVD so someone else could
later process to get this data.  But I doubt anything that I don't do on
the first pass will ever get done.

Tom Droege




> Just a question: can't you give an "upper limit" instead of "no
> detection", this "upper limit" being carefully estimated from skey,
> noise etc. ? "no detection" already indicates that it is not a very
> blue star, better than nothing!
>
> best whishes for the coming eater days.     Paul
>
> On Sat, 19 Mar 2005 11:21:43 -0600 (CST), droege@snapmail.us
> <droege@snapmail.us> wrote:
>> Arne and all,
>>
>> We are not (now) keeping the unmatched I detections.  We have been
>> keeping
>> the dark and flat corrected images, so they could be reprocessed to get
>> the unsaved detections.
>>
>> Looking at the V and Ic filter bandpasses superimposed on the CCD qe
>> curve, it makes sense to me that their should be roughly twice the I
>> sensitivity as
>> v.
>>
>> This is supported by the sky background of the I images which is about a
>> factor of 2 larger than the V images.
>>
>> For Arne's argument below, it should be noted that stars are lost at
>> each
>> end.  We lose I detections because they are faint and there is not
>> matching V detection.  We lose I detections because they are too bright
>> and thus the good matching V detection is thrown out because the I
>> detection is saturated.
>>
>> Arne writes:
>>
>> > One obvious solution is to change the exposure length of the V-band
>> > camera,
>> > making its exposure a factor of two or more longer than the I-band
>> camera.
>> > This is the best solution to gain matches, but complicates the survey
>> > process.  The best solution is to find out why your current V-band
>> > measures
>> > are so poor.
>>
>> Seems like there are three posibilities:
>>
>> 1) Put a neutral density filter in the I path
>> 2) Use different V and I exposures.
>> 3) Stop down the I camera.
>> 4) Use the lumigen coated CCDs in the V cameras.
>>
>> Arne says 2) "complicates the survey".  In what way?  One complication
>> is
>> that the present software can't do it.  Are there others?
>>
>> I don't like 1) beacuse I have had not so good experience with neutral
>> density filters in the past, and it adds another pair of surfaces to
>> keep
>> clean.
>>
>> So far I like 3) best.  It is easy to do to any level.  One can stop
>> down
>> just enough to save most of the bright I stars.
>>
>> I have two lumigen coated CCDs.  These should pick up the V response a
>> little (looking at the qe curve).  But I only have two and there are
>> five
>> dual systems.
>>
>> Now is the time to make such decisions.  The engineering run is over.
>> This is the "real" run.  If we are to do things like save the unmatched
>> detections, then we need software to do it now.  So "Heros" please step
>> forward.
>>
>> Tom Droege
>>
>>
>> > droege@snapmail.us wrote:
>> >> Thanks to Michael for the nice discussion below.  I think I am
>> finally
>> >> understanding what is going on.
>> >>
>> >> It seems to me that the best solution is to minimize the missing
>> >> detections.
>> >>
>> >> We could as Michael suggests assign some average (V-I).  I think this
>> >> would be a little smaller than the 3-4 Michael suggests, but the
>> result
>> >> is
>> >> still probably garbage.
>> >>
>> >> Another way might be to save all the measurements and then go
>> somewhere
>> >> else to find the (V - I) color term.  Then the (mostly I)
>> measurements
>> >> could be corrected.  This assumes we have such a catalog for the I
>> >> stars.
>> >> Seems to me that one of the problems with our survey is that we don't
>> >> have
>> >> a good I catalog.  Sigh!
>> >>
>> >> So I vote to adjust something to make the detections match for the
>> two
>> >> telescopes. We have:
>> >>
>> >> a)Exposure
>> >> b)Aperture
>> >>
>> >> to make them match.  (Anything else?) Which is best to use?  It is
>> >> really
>> >> more convenient to adjust the aperture.  This assumes that a simple
>> >> objective end mask will do it.
>> >>
>> > I will rephrase my answer once again to see if I can make it clearer.
>> > Let's take a simple example from one of Tom's earlier messages:
>> >    time    Idet    Vdet
>> > 34452056  1827    465
>> >
>> > If you decrease the I-band camera aperture but leave the exposure for
>> > both V and Ic identical, then you will still have 465 V detections,
>> but
>> > some smaller number of Ic detections - lets say you change it so that
>> > Vdet = Idet.  There are two problems with this:
>> >    (1) the V/Ic matches do *not* necessarily match a V-band measure
>> with
>> > an equivalently bright Ic measure.  You will lose some matches with
>> faint
>> > Ic measures.
>> >    (2) you lose the ~1300 Ic detections you currently have that are
>> > unmatched with V.  These *are* good detections; the software finds
>> them,
>> > they are just not matched (but presumably saved).  The net effect of
>> > decreasing the I-band camera aperture is to reduce the amount of
>> science
>> > coming from your camera.  Software does just as good a job of
>> "removing"
>> > the unmatched stars as would your aperture mask.
>> >
>> > One obvious solution is to change the exposure length of the V-band
>> > camera,
>> > making its exposure a factor of two or more longer than the I-band
>> camera.
>> > This is the best solution to gain matches, but complicates the survey
>> > process.  The best solution is to find out why your current V-band
>> > measures
>> > are so poor.
>> > Arne
>> >
>> >
>> >
>>
>>
>
>
> --
> ****************************************************
> Paul Bartholdi
> 11 chemin de la Barillette
> CH-1260 NYON
> Suisse                      tel +41 22 361 0222
> ****************************************************
>
>