Date: 20 Oct 2004 1100 PDT

Attendees: Sean, Mark L., Jason, Matt, Lexi, Bill, Dan

Agenda:
   1) Update on progress in developing background estimators
   2) Update on fitting routines
   3) New business

Update on background estimators: Sean posted an IDL procedure to flag artifacts on the muxbleed tiger team web site. The procedure identifies pixels that trigger artifacts by applying per channel thresholds in DN for muxbleed, column pulldown and banding (where applicable). The procedure then updates the dmask file, flagging pixels that contain muxbleed, pulldown, etc. The procedure was developed using the 30 Doradus data to guide the identification of artifacts. Lexi expressed a valid concern that this procedure would not work very well for muxbleed that cannot be identified in a single BCD. The group consensus was that we will probably have to have an empirical fit to handle the low level muxbleed artifacts which may be determined from fitting appropriate ensembles of BCDs.

Mark L. has written an IDL procedure to estimate the background in masked pixels. The routine performs a set of 1d interpolations over affected pixels with the interpolation direction determined by the artifact. The procedure works pretty well for both the Elais N1 and 30 Dor images. Mark and Sean are continuing to work on improving the background estimator, but the current version appears to be good enough to use in developing fits.

Fitting routines: Lexi reported progress in working with the data processed without the pipeline muxbleed correction. He generated time history images on a per output basis analogous to the ones previously posted. Lexi and Dan are continuing to work on developing fitting routines. There was some discussion of whether or not to back out the flat-field before fitting muxbleed (also see the minutes from meeting 2). Jason estimated that the flat could cause up to a 20% variation for pixels at the edge of FPA 1 array. To first order, we could mulitply by the superflat to remove the existing flat field, as to report the actual flat used for any given AOR will involve some bookkeeping.

Muxbleed time dependance: Jason noted that he sees a variation in the level of stripiness due to muxbleed in BCDs from his GTO AORs. The AORs image ~1 Jy sources which produce muxbleed that seems to intensify (affect more pixels) throughout the 10-15 minute AORs. This is the first reported instance of time variability of muxbleed. No temporal variation was noted in the deep images inspected by Lexi, but the source fluences are much lower than in Jason's AORs. The discussion expanded to sampling dependent variations which have been noted in HDR AORs. It was noted that no example of muxbleed existed where the muxbleed extended pass the top or bottom of the array to wrap around. Bill talked about how hot pixels will affect the observed muxbleed. As hot pixels also produce muxbleed, the observed muxbleed will be modified by the presence of hot pixels and the fowler sampling used. Hot pixels will cause the character of the muxbleed to change.

Action items:

1) Mark and Sean to continue working on background estimators

2) Dan and Lexi to continue working on fitting routines

3) Jason to post his images showing the temporal variation of muxbleed

4) Matt, Bill and Sean to fix the artifact vs. fowler IER

Next meeting: Planned for 04 Nov 2004 at 1400 PDT