over-subtraction of Mosaic3 z-band imaging

[siyue]

Many z-band images of nearby galaxies have some "fence" patterns throughout the galaxy. Below should you one example NGC2742:

Meanwhile, the sky background is over subtracted:

The region enclosed by the green curves are obviously (for me) darker than other sky regions!
I didn't find the same problem in the g-band, r-band, and DECam z-band data.
This problem may make the photometry based on MzLS z-band data suspicious.
Is the DESI team aware of the problem and are there any solutions?
Otherwise I have to remove all the galaxies observed with MzLS from my sample. very upset.

[dstndstn]

Yes, we're very aware of this issue. The Mosaic3 camera has pattern noise, and these artifacts arise from the algorithm that tries to remove it.

[siyue]

Thanks! just learn more about this infuriating noise! Then I will remove all these galaxies from my sample. If I may make a comment, I would suggest making this potential problem explicit in the SGA's description!

[moustakas]

@siyue, yes, this issue is frustrating and was also flagged here as it has impacted my intracluster light work, too--
legacysurvey/legacypipe#666

I agree that we should add a note to the "Known Issues" page.

[siyue]

YES! They are too green!! This is also how I found the problem!!