Date: 04 Nov 2004 1100 PDT

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

Agenda:
   1) Update on progress in developing background estimators
   2) Update on fitting routines
   3) Artifact vs. fowler test.

Update on background estimators: Sean has an IDL procedure that seems to produce good estimates of the sky in pixels affected by artifacts. The routine identifies affected pixels by generating masks around pixels above a specified threshold. For each masked pixel, the routine estimates the flux using a Gaussian kernel applied to a window (5x5) of nearby pixels. If there are not enough pixels to perform a reliable estimation, a larger window (11x11) is used. The routine seems to produce acceptable results for channels 1 and 2 when applied to the 30 Doradus images.

Fitting routines: Lexi showed some results of applying a characterization and fitting procedure he has written. The procedure identifies muxbleed causing pixels by thresholding above a user specified threshold. The procedure then identifies cases of muxbleed which are not blended with other muxbleed. The procedure creates a 2d image with each instance of muxbleed comprising an image row. The image is compressed along rows using outlier rejection to produce a muxbleed profile. A separate procedure can use the profile to correct for muxbleed. The profile is fit to each instance of muxbleed using the model, sky + muxbleed = background + alpha * profile, where the background is determined from the median of an image after rejecting sources. Alpha is a constant which is derived from fitting the model to the data. Both procedures return a wealth of diagnostic information. Lexi's procedures work extremely well on the GOODS UDF data. The derived muxbleed profiles appear to be constant for all instances of muxbleed examined. The GOODS data only displays muxbleed over a limited range of triggering pixel flux; therefore, further exploration of parameter space (source flux, fowler sampling) is indicated. Tiger team is to explore more instances of muxbleed using the tools Lexi has developed.

Artifact vs. fowler test: The IER for the test has been revised and submitted for validation on the OSTL. The on-board test should occur sometime around 20 Dec 2004.

Action items:

1) Sean and Jason to distribute the muxbleed lookup tables used by the pipeline.

2) Lexi to distribute his muxbleed characterization and fitting routine.

3) Sean to distribute his background fitting procedure.

4) Sean, Jason and Mark to investigate representative datasets using Lexi's
routines.

4) Lexi to try to further characterize the dependancy of the shape of muxbleed
on source flux, output.

Next meeting: TBD, sometime after Thanksgiving