Versions Compared


  • This line was added.
  • This line was removed.
  • Formatting was changed.


WISE data loadingIgor Gaponenko
  • Received input from Xiuqin with information about image metadata tables to be loaded
  • Renaming previously-loaded tables to be closer to their original names instead of the nominal LSST names
  • UPDATED on 2017-05-08: 8 out of 11 tables are now available in PDAC. This covers all relevant image metadata tables. Three other multi-billion row tables representing single exposure sources of All-Sky, 3-Band Cryo and Post-Crio missions are currently on hold due to unavailability of data from IRSA. See the current status of all tables in: PDAC v2 data list
  • Brian Van Klaveren working on organizing the metadata, main issue is the data transformation before loading
  • John Gates working on problem with joins between tables in different databases in Qserv (applies to WISE "Reject"-to-ForcedSource joins, and on WISE-to-SDSS spatial joins) - issue is in sub-chunk handling

MetaservBrian Van Klaveren
  • metaserv v0 is up and running but doesn't provide the expected APIs, mostly just table names at this point
  • column metadata is basically in "CREATE TABLE" format, does not include UCDs or units
  • Planning on working on metaserv v1 this month, though there are scheduling issues with other meetings/activities

Portal status
  • IPAC wants to start using the current metaserv information next week; not sure whether to wait for metaserv v1, will decide shortly and let Brian Van Klaveren know
  • Working on time-series and other improvements for IRSA; the non-IRSA-specific parts of this will show up in PDAC as well, as well as some new features
  • Cone search starting from a list of objects/positions has been deployed in PDAC but this PDAC-specific implementation hasn't been fully tested yet in-house
    • The usual IRSA way of doing this relies on the ability to upload a temporary table and can't be used against the PDAC DAX services, which are read-only.
    • IPAC has developed an alternate implementation for PDAC which loops over single-object searches.
  • Have gone through Krzysztof Suberlak's writeup; lots of good feedback, will be making tickets from it. Briefly discussed today:
    • Looking into supporting a natural UI for extending object name searches beyond just NED and Simbad
    • Will improve parsing of negative coordinate values in the main catalog search field
    •  Unknown User (xiuqin) Can you provide a list of and/or tag the tickets that were triggered by user feedback?   label "pdac-feedback" will be used to track those tickets. Filter
      jqlQueryteam="Science User Interface" AND labels in (pdac-feedback)

User feedback

Discussed several issues Chris has observed:

  • Discussion of UI for finding time series data for a specific object
    • Observed that cone searches on the ForcedSource table can't be guaranteed to produce a time series for a single Object. Seems awkward to require a WHERE clause to clean this up.
    • Unknown User (xiuqin) There is already a dedicated button on ForcedSource search result screens that you can use to generate a time series for the specific Object that is behind any row in the table.
    • Gregory Dubois-Felsmann Driving this from the Object ("DeepSource" in the temporary language of PDAC) table instead is what the Portal is supposed to do, to lead users to the data.
      • Unknown User (xiuqin) Yes, but this didn't work out well when first tried because so many Objects didn't lead to ForcedSources at all. Gregory Dubois-Felsmann: this requires SDSS-Stripe82-Summer2013-specific logic: only the i band Objects were used to seed forced photometry (in all bands). So what's needed is a button on the Object result screen that allows this action to be taken and that provides user feedback to educate them about the nature of the data. (This is what the Portal is all about!)
      •  Unknown User (xiuqin) will work with Tatiana Goldina to develop a button that is clickable when an i-band Object row is selected, greyed-out when a row from a different band is selected, and provides a (long) tooltip that explains the point about forced photometry being limited to i-band Objects. The has been scheduled in June 2017.
      • We should also remind the users somehow in the UI that the Objects in the different bands are not in any way linked to each other. (No associations were performed.)
  • Discussion of connection problem reported on #dm-pdac
    • First time connecting to PDAC after a long gap, the first search seems to fail; future ones are OK. Error reported is “Fail to load table. Error: edu.caltech.ipac.firefly.server.query.DataAccessException: DataAccessException:ERROR:DAX Error: OperationalError from:unknown” which suggests a DAX origin.
    • Brian Van Klaveren has an idea what could be causing this, associated with the maintenance of persistent database connections.
  • Discussion of coordinate rotation issues in external searches on M81 from multiple archives
    • Reported by Krzysztof Suberlak at this link on #dm-pdac
    • Basic solution is: the retrieved images have different pixel grid orientations and are shown in their natural orientation by default; click on "WCS match" to rotate them to a common orientation
      • NB: this issue will arise with LSST single-epoch images, which will all be at different rotation angles. With user feedback, we will have to decide what users are most likely to want when requesting an image cutout series.
    • Secondary issue: on the specific images selected, even after "WCS match" the coordinate grids don't appear to be exactly parallel, perhaps a few degrees out of alignment.
      • The example was constructed from images of m81 from WISE coadd band 1, SDSS (DR7?) u band, 2MASS J, and DSS "red"
      • (Updated: issue has been investigated, see
      •  Unknown User (xiuqin) will investigate with the IRSA archive science team and report back   

InfrastructureUnknown User (pdomagala)

PlanningGregory Dubois-Felsmann