| Table of Contents |
|---|
Introduction
New WPs
New project schedule
Those with ideas for future newsletter items should contact the LSST:UK Project Managers (George Beckett and Terry Sloanlusc_pm@mlist.is.ed.ac.uk), while everyone is encouraged to subscribe to the Rubin Observatory Digest for more general news from the US observatory team.
LSST:UK All-Hands Meeting
2021 Mid-Year Junior Associates Selection Round
Broker Workshop
I am old enough to remember when talks were given with the aid of an overhead projector and transparent acetate sheets written by hand. Since then Powerpoint and Keynote became the preferred medium, but it was quite different at the LSSTC enabling science 2021 broker workshop, where presentations merged with tutorials in a github page consisting of notebooks and markdown, all times in UTC because the 100 or so participants were all on different continents.
Ken Smith presented on behalf of LSST:UK: a number of notebooks on Google Colab, that use the new Lasair API. You can try them here, and read about the API here.
Since the Rubin Observatory is not yet operational, there is no LSST data, so most of the brokers are using the Zwicky Transient Facility (ZTF) as a prototype. ZTF delivers something like 2% to 4% of the expected data rate from LSST, partly because there are fewer transients, partly because the data packets are smaller than LSST.
The Alerce and ANTARES brokers are the big players, with significant resources from the Chilean and US governments. In Europe, there is the Fink from France, AMPEL from Germany, and Lasair from the UK. The variety of approaches shows the wisdom of the Rubin Observatory in outsourcing this aspect of the project to a wide community: some brokers have sophisticated classifiers of variable stars and transients, some expect users to write code, while other expect users to build SQL queries.
The life-cycle of a transient alert is emerging through consensus. The telescope makes a source detection and sends out an alert packet; a broker processes the packet and annotates it with context and classification, then the users of the broker are informed of alerts that satisfy their criteria. That user may be a machine, a so-called marshall system, that allows a group of scientists to share opinions, and the marshall may be connect to TOMS (Target and Observation Manager Software) that can initiate follow-up observation of the most interesting sources.
There were many technology innovations on display. As noted above, new ways to combine code, tutorial, and presentation; ways to build sustainable code with pre-commit git hooks and automated testing; Javascript tools for websites such as React/Redux; and authentication services via COManage. There was a discussion section on NoSQL databases, which promise scalability even with very large amounts of data.
The LSST:UK Photometric Catalogue Cross-Match Service
While the Rubin Observatory’s LSST data will be great by themselves, the real benefit will be in combining the data with those from other telescopes. We have therefore been implementing a service to provide the cross-matches between LSST data releases and a set of other datasets, like Gaia for its proper motions or those in the IR such as WISE or the VISTA survey for the extended wavelength coverage.
However, we expect LSST to simply be too crowded for “traditional” cross-matching methods (such as the nearest neighbour method, shown in (a) to the right) to provide good matches for faint objects – and a significant number of LSST sources are what we might consider “faint”! With WISE as our precursor to LSST – both having roughly the same number of sources per point-spread-function area, coincidentally – we have developed a method to overcome the effects of this crowding. In (b) on the right, for an example of Gaia (catalogue positions denoted with red xs) vs WISE (background image, catalogue position as a green +), we can see an additional Gaia source that we don’t see in WISE due to its low angular resolution, shown by the WISE image in the background, where the fainter WISE object can (just about) be seen as a (very slight) elongation of the background PSF to the right of the image. This object has been absorbed into the detection of the bright object we do detect, but affects the position we measure for the source.
What these hidden objects do is move the measured positions of our detected sources.
Modelling this effect, we can correct for it in the questions we ask during the cross-match process, the most important of which is “what are the chances that these two objects are counterparts to one another, and two detections of one actual sky object, given the recorded distance between them on the sky?” If we assumed that neglected this effect didn’t exist, the movements of where we measure an object to be on the sky by these hidden objects would cause us to think that the objects, now much too far apart to be explainable as being the same object detected twice, were two separate objects which happen to be near to one another on the sky. The user of these matches would then think “okay, the infrared brightness of this source must be below that detectable by this survey” and could, for example, use upper flux limits in a spectral energy distribution fit – when really the source is very bright, just a little further away than we deemed okay!
This effect is very important for WISE – the figure on the left shows a comparison between our matches, including this effect where hidden sources move the recorded positions of other objects, and those provided by the Gaia team, where they don’t include the effect. We recover many more sources beyond half an arcsecond separation, which would otherwise be missed from the composite dataset created from the lists of matching pairs.
Accounting for these hidden contaminant sources is important for two reasons. First, if anyone wanted to create a combined dataset of LSST with other surveys which suffer from significant crowding, their composite photometric catalogue would be incomplete. Second, as we anticipate LSST itself suffering from high levels of crowding due to the sheer number of sources that will be detected down to 27th magnitude, we must model the LSST data as being a combination of several unresolved objects. Conveniently, however, this modelling allows us to provide information on how much too bright the objects are, providing users will an estimate of the flux of these perturbing sources.
Recent LSST:UK outputs
LSST:UK has recently produced the following technical reports.
Title | Author | Description |
|---|---|---|
Forthcoming meetings of interest
The global pandemic has led to almost all face-to-face meetings being cancelled. However, in light of continued restrictions on travel, Rubin Observatory business has moved online and we aim to maintain a list of relevant/ interesting upcoming meetings on our Confluence site.