[Issues] overscan correction methods

[Ewout M. Helmich]

G.A. Verdoes Kleijn wrote:

>Dear Astro-Wisers,
>
>Data reduction processes in the Astro-Wise system have the option to use
>subtractions of overscan regions, i.e., 'overscan corrections'. 
>Bias frames: if no overscan correction is used the masterbias frame will
>contain both a typical overall bias level and the smaller scale bias
>variations. The average of the overall bias level is a positive number
>(mostly >100) while the average of the small-scale bias variations is
>very close to zero (mostly <<1). Thus there are two main types of
>masterbias frames those with the overall bias level (type MB0) and those
>without (type MB1). Eight different ways are defined for performing
>overscan corrections. In other words, MB1 consists of 8 subtypes. 
>
>Thus for reduction processes which involve subtraction of bias frames it
>must be assured that the appropriate masterbias is subtracted.
>Currently, the Astro-Wise system does not enforce selection of an
>appropriate masterbias (a warning is issued in some situations).
>Espcially for automated processes involving bias subtraction (e..g, with
>the recipes and TargetProcessing tools) it would be highly convenient to
>have a check on overscan correction. With this in place it will be
>possible as well to improve the default calibration files (i.e., the
>"forevers").
>
>Best wishes, Gijs
>
>_______________________________________________
>Issues mailing list
>Issues at astro-wise.org
>http://listman.astro-wise.org/mailman/listinfo/issues
>  
>
Hi everyone,

The default overscan correction method, used in all Tasks will be 
changed shortly. When you update your opipe checkout later today, and 
start awe, you should see a news message mentioning this. From then on, 
the following applies:

Currently the default is "0", or "no overscan correction". This will be 
changed to "6", or "per-row overscan correction". If you are used to 
reducing data without specifying an overscan option, you must from now 
on specify overscan=0 or oc=0 or -oc 0 etc. in relevant arguments. On 
the other hand if you happen to use per-row overscan correction, it will 
no longer be necessary to specify e.g. overscan=6.

In addition, checks have been built into the classes to prevent mixing 
of different methods, but only in those cases where it will produce an 
obviously wrong result. In particular: in ReducedScienceFrame it is not 
checked whether the overscan correction methods for the input BiasFrame 
and the input FringeFrame are the same. However, it is checked whether 
the methods of the BiasFrame and the ReducedScienceFrame differ, because 
in this case either no net bias, or twice the bias could be subtracted 
from the ReducedScienceFrame.


Some examples if you want to keep the previous default behaviour:

awe> dpu.run('Bias', date='2000-04-28', instrument='WFI', chip='ccd50', 
overscan=0)

awe> task = Bias(date='2000-04-28', instrument='WFI', chip='ccd50', 
overscan=0)
awe> task.execute()

linux>awe $AWPIPE/DBRecipes/Bias.py -d 2000-04-28 -i WFI -c ccd50 -oc 0


Regards,
Ewout