Details
-
Type:
Story
-
Status: Done
-
Resolution: Done
-
Fix Version/s: None
-
Component/s: obs_decam
-
Labels:None
-
Story Points:5
-
Epic Link:
-
Team:Data Facility
Description
Currently, the default overscan correction from IsrTask is used for processing DECam raw data. Overscan subtraction is done one amplifier at a time.
However, a bias jump occurs due to the simultaneous readout of the smaller ancillary CCDs on DECam, some images show discontinuity in the y direction across one amplifier, as in the example screen shot.
This ticket is to improve overscan correction for DECam data so to mitigate this discontinuity in the ISR processing.
Arrangement of CCDs on DECam: http://www.ctio.noao.edu/noao/sites/default/files/DECam/DECamPixelOrientation.png
More details:
There are 6 backplanes in the readout system, shown by the colors in DECamPixelOrientation.png. In raw data files, the CCD's backplane is noted in the header keyword "FPA". Examination of some images suggests that science CCDs on orange and yellow backplanes show bias jump at 2098 pixels from the y readout. That is the y size of the focus CCDs.
Actions:
For CCDs on the affected backplanes, divide the array into two pieces at the jump location, and do overscan correction on the upper and lower pieces separately.
Attachments
Issue Links
Activity
It might help to plot the overscan, which you can do by using the --debug command-line argument, and put in your debug.py (which sits somewhere on your PYTHONPATH):
import lsstDebug
|
print "Importing debug settings..."
|
def DebugInfo(name):
|
di = lsstDebug.getInfo(name) # N.b. lsstDebug.Info(name) would call us recursively
|
if name in (
|
"lsst.ip.isr.isr",
|
):
|
di.display = True
|
return di
|
lsstDebug.Info = DebugInfo
|
Would you please post one of the plots here?
Paul Price
added a comment - It might help to plot the overscan, which you can do by using the --debug command-line argument, and put in your debug.py (which sits somewhere on your PYTHONPATH ):
import lsstDebug
print "Importing debug settings..."
def DebugInfo(name):
di = lsstDebug.getInfo(name) # N.b. lsstDebug.Info(name) would call us recursively
if name in (
"lsst.ip.isr.isr",
):
di.display = True
return di
lsstDebug.Info = DebugInfo
Would you please post one of the plots here? Useful debug feature I didn't know about before! Thanks for mentioning it!
This is with isr.overscanFitType='AKIMA_SPLINE' isr.overscanOrder=30 for ccdnum 52 (one with larger jump)
Hsin-Fang Chiang
added a comment - Useful debug feature I didn't know about before! Thanks for mentioning it!
This is with isr.overscanFitType='AKIMA_SPLINE' isr.overscanOrder=30 for ccdnum 52 (one with larger jump)
Paul Price would you mind reviewing this?
Hsin-Fang Chiang
added a comment - Paul Price would you mind reviewing this? Very nice! You were more thorough than I would have been (checking the header for the backplane, checking the readout corner). The only comment I would make is that I think you should use the entire 110 char width, e.g., write this:
lowerDataBBox = afwGeom.Box2I(
|
dataBox.getBegin(),
|
afwGeom.Extent2I(dataBox.getWidth(), yLower))
|
as this:
lowerDataBBox = afwGeom.Box2I(dataBox.getBegin(), afwGeom.Extent2I(dataBox.getWidth(), yLower))
|
I think the latter is easier to read. But if you disagree, just ignore me and merge anyway.
Paul Price
added a comment - Very nice! You were more thorough than I would have been (checking the header for the backplane, checking the readout corner). The only comment I would make is that I think you should use the entire 110 char width, e.g., write this:
lowerDataBBox = afwGeom.Box2I(
dataBox.getBegin(),
afwGeom.Extent2I(dataBox.getWidth(), yLower))
as this:
lowerDataBBox = afwGeom.Box2I(dataBox.getBegin(), afwGeom.Extent2I(dataBox.getWidth(), yLower))
I think the latter is easier to read. But if you disagree, just ignore me and merge anyway. Oh, would you mind posting an "after" picture showing that the bias jump is fixed?
Paul Price
added a comment - Oh, would you mind posting an "after" picture showing that the bias jump is fixed? Thank you for your review!
Added an "after" screen shot of the same image.
Changes merged to master.
Hsin-Fang Chiang
added a comment - Thank you for your review!
Added an "after" screen shot of the same image.
Changes merged to master. That looks nice — well done!
Paul Price
added a comment - That looks nice — well done! 
Tried high order splines and the jump is still quite obvious.
I'll proceed with Paul Price's suggestion "for the troublesome amps only, cut them in two vertically and then feed each piece into IsrTask.overscanCorrection separately."