Deep Drilling Fields and Data Management

On behalf of the LSST Data Management team, we would like to open up this forum to discuss the processing of proposed “Deep Drilling” and/or “Mini-Survey” programs. This conversation between the science community and the LSST DM team was inspired by the breakout session on “Deep Drilling Fields and Other LSST Mini-Surveys” at the LSST Project and Community Workshop in 2016. The relevant DM-DDF issues are outlined in Gregory Dubois-Felsmann’s talk from that session (available at, or from the breakout session website, including these questions that we encourage the community to keep in mind while designing their DDF and Mini-Survey proposals:

  1. What additional processing beyond that currently planned by the DM team (alerting relative to an annually created template) would greatly enhance the DDF science goals?
  2. Are there DDF or Mini-Surveys specific aspects of the Level 3 system that would add significant value if provided? “Level 3” is the LSST-provided capability that enables non-DM, user-driven, processing of LSST data at the LSST Archive center (or remotely).
  3. Are there aspects of the Science User Interface & Tools (SUIT) that need to be developed in order to enhance the usefulness of DDF data products
  4. To what degree should the DDF or Mini-Survey imaging could/should be incorporated into the main survey’s deep stacks and associated data products (as opposed to being processed as separate data products)?

The following resources may also be of use to the community:

  1. LSST Data Products Definitions Document (DPDD).
  2. The LSST Data Management System, Juric et al. (2015),

For weak lensing, I think we’re going to want to:

  • process at least some deep drilling fields (high-latitude ones) in the same way we process a full data release production before running the full data release production, so we can use the results to build priors and/or calibrate shear estimates on the wide survey;

  • process various wide-depth subsets of some deep drilling fields (again, high-latitude ones) using the regular DRP pipeline. We’ll definitely want best-seeing, worst-seeing, and probably a couple of independent typical-seeing subsets, but there may be other ways we’d want to subdivide as well.