In article <[EMAIL PROTECTED]>,
 Jacek Becla <[EMAIL PROTECTED]> wrote:

> >    - do either of the current prototype schemas (UW and sergei's)
> >      have time partitioning in mind?
> 
> Sergei, Nicole and Russ, can you answer this?

What is time partitioning?

The UW database has a TAI field for each image. Object link to images, 
so it's easy to find the time at which a given object was measured.

If integration time varies across the image (as is likely for LSST if a 
shutter is used, as somebody pointed out at a recent phone con), then:
- Additional information is needed in the object table to allow 
computing time as a function of position on the image.
- The time of each source should probably be computed as sources are 
detected, and stored as a separate TAI field in the source table.

We have run into this issue ourselves. Time varies across the image for 
SDSS data (drift-scanned data). Our Image table has TAI and RowTime such 
that TAI of source = TAI of image + RowTime*CtrY. However, we have not 
yet added a TAI field to our Object table.

-- Russell

_______________________________________________
LSST-data mailing list
[email protected]
http://www.lsstmail.org/mailman/listinfo/lsst-data

Reply via email to