Hello everyone, thanks for the very hard work with commissioning the observatory and compiling the SV data set.
Is there a very rough estimate on how large the DP2 data set is going to be, in terms of TB on disk? In particular we (DESC) are interested on a guesstimate for coadd images. Thanks!
Hi @nsevilla, I’m not sure there’s yet a published estimate for DP2, no. But there is a summary of the data obtained during commissioning, and in particular for the science validation surveys on which DP2 will be based, that we can use as an upper limit.
Science Validation Survey Progress Report, 2025-09-30
Based on the summary linked above, it looks like a DP2 based on the science validation surveys might cover about 1000 square degrees (very bottom of the page). This is a very rough estimate, as the final decisions based on science validation, etc., are of course not done yet. But you did say you only wanted a guesstimate.
If I grab a DP1 deep coadd image (79 sq arcmin) from the butler and save it (as a FITS; mask planes, headers, and everything), it’s about 230 MB. So in six filters, over 1000 square degrees, the coadded images alone might add up to a little over 62 TB.
If this guesstimate for the on-disk size of the coadded images is sufficient for your needs, please mark this reply post as the solution – and if it’s not, let’s continue the discussion.
The SV summary page mentions that about 15000 visit images (this paragraph), total, were obtained. The standard estimate is 20 GB per LSSTCam image, so ~300 TB is the ballpark guesstimate for DP2 visit images.
One small clarification - the 1000 square degrees referenced at the bottom of the survey progress report linked above is the area that reached approximately half of the one year exposure time for the LSST WFD. The total SV area was a bit over 3,000 square degrees and while I can’t predict exactly how much of this will be useful for coadds, you might want plan for on the order of 3x Melissa’s estimate above.
(I suppose also, in astronomy this is “about the same” 
)
Thanks, indeed, we put in some padding in our request for space of a factor of a few.
Best
N