I share Dave Monet's concern about agglutinating detections into objects
at the DB level.
This is fine for bright objects in constant seeing, but a total
nightmare
for faint blended things as the seeing changes.
I believe that the way we intend to process the deep LSST data is:
1/ Search for variable/moving objects on each frame (or rather
frame pair).
2/ Possibly measure the properties of known objects (i.e. at
known positions) on each frame
3/ Occasionally, process the entire stack of exposures. At
this point the primary item being detected is an object NOT
a detection; the entire stack will be processed to determine
the properties of these objects which will therefore have
a consistent definition even if not detected in a given band/exposure
So what we do at basecamp is not closely related to the final catalogs.
Much of the per-exposure science (maybe all?) will come from item 1, and
we do think about how to match these detections (variables and moving
beasties) to the master catalog from [the previous iteration of] step 3.
R
_______________________________________________
LSST-data mailing list
[email protected]
http://www.lsstmail.org/mailman/listinfo/lsst-data
