Ö÷Ìâ: Questions on u-band data

Yipeng, Shenshiyin and Zhou,


We've done some initial, independent comparisons with the u-band data 
that you shared with us on June 12th.  I'll include some of that below.
But a number of our photometry experts will be converging on Berkeley
this coming week, so this would be the opportunity to look at this more
carefully.  For that, I was wondering if we could request the same images
from Zhou but before re-interpolating.  The approach we're looking at
(forward modeling with "The Tractor") assumes that the pixel-level data
are independent.


Attached are some comparisons to SDSS data from co-adding ~50 epochs
from the stripe 82 data.  This is using the DR8 data, where the photometry
and flat-fields were as good as we could make them.


The first plot shows the straight comparison between the SCUSS "Bertin"
magnitudes and SDSS co-add.  The quoted SCUSS errors are 0.02 mag at u~20,
but the scatter implies 0.08 mag.  (The "DAOphot" magnitudes have a larger
 scatter.)
The increased scatter could imply that the data aren't as deep as expected,
or that we're seeing differences between the calibration or color terms
between the surveys.


The 2nd plot shows the trends with color.  There is some at the few percent 
level, but this is a match better match between the systems than one might 
hope for.  So very good news, and it simply means this trend should be
removed when making this comparison.


The residuals with RA,Dec are the most interesting.  SDSS DR8 has systematics
in Dec. at the level of 0.5%, but in RA they'll be good to < 0.2% (because 
of the drift scanning and averaging over many runs).  I think this shows
there's a real opportunity to improve the SCUSS data with the flat-fields.  
(Could there also be a problem with the interpolationif it's not 
flux-conserving?)


In detail, the SDSS data uses 6 u-band filters.  They're very similar, but 
we know the differences between them.  Figure 8 and Table 3 on this paper
tabulate these:

   http://iopscience.iop.org/1538-3881/139/4/1628/fulltext/


If you have the non-interpolated images to share, we can see what we learn
this week, and would certainly let you know what we find.  If you don't have
the SDSS co-added photometry from the final data release (DR8), we should
share those too.  We don't have anything like that on the DR8 web site 
(maybe we should!).


Thanks!
David