Location: Sesto, Italy
Date: 3-7 February 2020
Attendees: Yusra AlSayyad and Robert Lupton attended the meeting, representing the Rubin Observatory
Presentations:
Robert gave two talks: "Rubin Observatory Pipelines" and "Deblending in LSST"
Yusra gave a talk, "LSST Data Products", and ran a discussion session. The LSB features that were of most interest to the participants were: LSB galaxies such as those being found by Johnny Greco and Van Dokkum's Firefly Intra-Cluster Light (ICL) Outer parts of well-resolved (e.g. NGC) galaxies; e.g. truncated disks IR Cirrus (predominantly as a contamination) The IAC (Instituto de Astrofísica de Canarias) group has been reprocessing SDSS Stripe 82 and HSC data, and are expecting to have to redo the ISR for LSST. We need to make sure that this is not the case -- they are basing their assesment on the behaviour of HSC flat fielding, which is not as good as it should be (partially due to limitations of the HSC calibration lamps, and the issues with amp-to-amp offsets that we still don't understand). In particular, the IAC group likes sky flats, but doesn't appear to understand the distinction between flats that flatten the sky, and those which deliver correct photometry. For example, they were apparently unaware of the strong ghosting in the SDSS sky flats. We do need to think about how we will obtain truly flat backgrounds using lsstCam, and the possible inclusion of median sky frames in the strategy for removing large-scale gradients. There was great interest in background matching. Current techniques to avoid this include using single per-CCD background levels (IAC) and taking sky data close in time to science data. It became clear that subtraction of bright star wings is required and moderately well developed. Claims were made that the outer parts of the wings were time variable, although the data were not presented convincingly. There was no discussion of the separation of part of the PSF came from ghosties/ghoulies, and how much came from the atmosphere. The former are strongly position dependent (cf. the HSC Arcturus movie), but should be stable; the latter would be expected to *not* vary with position, but potentially vary depending on atmospheric conditions. Both contributions are probably chromatic (due to coatings, filters, and the properties of atmospheric scatterers). There is also a contribution from the detectors, especially in y and (probably) z, which will be contained within single CCDs. The current PSF-wing work does not take these considerations into account. We can investigate this using HSC data, but need to be sure that we'll get enough bright-star imaging with SST to characterise the former. Raul Infante-Sainz from IAC will be visiting Princeton in early March to work with Morgan Schmitz on star subtraction in LSST/HSC. This is important; failure to remove wings of stars makes the deblender work far too hard. We explained many times that the problem facing LSB science is how we split the "sky" from the "objects". We promised that we would be producing coadded frames that had had (almost all of) the "terrestrial" background removed (the quotes are because we can potentially remove some of e.g. the Zodi too, but this is just a detail). We also discussed that these would not be the frames on which most LSST science would be based, but that we would be removing some sort of model of the extended light. We promised that it would be possible to get this model back, but for some reason this didn't seem to sink in. We asked for help with defining the model of sky that we would remove to make it as useful as possible to LSB science (e.g. by making the model searchable as a set of extended structures, e.g. wavelet models, rather than as pixel grids). The UK (under Sugata Kaviraj) has funding from STFC to work on this sort of thing; the lucky victim is Aaron Watkins who's moving from Finland to Liverpool in 2020-03. We discussed the angular scales of objects of interest, but didn't make a lot of progress. The IAC people, and possibly other groups, are using Muhammad Akhlagi's definition of sky, which is basically (in RHL's opinion) looking for the darkest part of the sky. The Rubin/LSST group would benefit from careful thought about how we want to define the sky, and the interaction between the definition and the demands put upon the deblender. I think that we can also make progress with more aggressive masking of objects (possibly following multi-scale detection), and that this are interesting opportunities for collaboration. The meeting decided that the LSST Galaxies Science Collaboration would do experiments injecting low-surface brightness things into HSC data. They nominated Lee Kelvin to do this on his science time.
References
The workshop schedule, including slides, is available here: https://docs.google.com/spreadsheets/d/1T_vYW5fyQf-NOPDapYqVWk-5r7MRdUzKlyunb48mIP0/edit#gid=0 Do you want me to pull out our talks? Hmm https://drive.google.com/open?id=1-l1Eljgf8zVRzkGnJx_M8W_aaENAe3Fp RHL Deblenders https://drive.google.com/open?id=1FkD0Xysl0WlPxpiM6FQxYzBvtqkCNNrm RHL deblenders Yusra ran a discussion session and LSP demo. https://docs.google.com/document/d/1nG2PX6y2ugcBGCSwgJNRpTCdEeehn1m0BfVtvLkESNQ/edit?usp=sharing R