Re: fun with coadding images on Disk Set 19

[Tom Droege <tdroege@veriomail.com>]

Arne and all,

I am trying to learn and to take data that I can start reducing to useful 
science.

Since we are not going very deep with the Mark IV, it seems to me that we 
need to go as deep as we can in order to get any useful new 
information.  If we don't go as deep as we can, then we run the risk of 
only measuring previously well studied objects.

It appears that at my location, 100 seconds is about as long an exposure as 
I wish to take.  This usually produces 4000 counts of sky.  A 400 second 
exposure would more than half fill the pixels.  So I think in terms of 
taking a 400 second exposure in four pieces to keep the dynamic range of 
the measurements in bound.  Yes, this takes readout time.  I try to gain by 
taking a continuous exposure string.  This means that I do not have to 
clear after each exposure.

Sigh!  The co-adding seems to take a lot of time.  I co-added the first and 
last exposure from the DS193 data and it took a really long time.  But I 
had to provide big offsets to the program I was using.  When the offset is 
well understood between frames, this process should go faster.  On the plus 
side, computers keep getting faster so computation time may not matter that 
much.

I am trying to sort out a productive science program for my site.  Data set 
DS19 is my first attempt of a collection program.  How is it as a 
plan?  What could be improved?

Tom Droege



At 08:42 AM 10/17/01 -0700, you wrote:
>Coadding to go deeper is a useful tool, and I recommend
>it if someone wants to look for fainter variables in Tom's set.
>Coadding has pros and cons, however.  Since there is drift
>in the images, using shorter exposures and stacking will result
>in smaller star profiles, a good thing.  For us high-altitude
>sites, coadding permits cosmic ray removal.  However, it does
>mean that multiple readout times are involved, meaning
>less open-shutter time per night compared with single exposures.
>The software effort is more time-consuming and complex.  The
>data storage is more.
>   The limiting factor is going to be the pixel size and blending.
>I think V=16-17 is certainly possible away from the galactic plane,
>but in the plane you might consider some of the fancier differential
>image subtraction algorithms.  I also think you will get more return
>per hour spent by looking at brighter sources rather than trying
>to see how faint you can work.
>Arne