We’ve had 3 days of learning and sprinting. Here are some of the key results from the meeting.

Andy Rivkin – thoughts about filters for LSST asteroid taxonomy. If you only have 3 out of 4 filters; gri is pretty good if precision is high; grz is okay on average; Sloan filters are not the same as LSST

Josef Durech – experimented with code to get shapes, spin poles, and colors. Took ATLAS data (3 years of data) and took only 50 points (SNR>30) in 2 colors covering a range of phase angles, recovered correct period, color, shape, and spin pole.

Mario Juric – developed updated scheme (data model) for LSST Solar System Data Products and getting reading for DM (Data Management) to implement. For first year, could assume same phase curve slope for all colors. Collaboration Workspace worked well (LSST stack installed already for SSSC Spring Jupyter Hub).

Kat Volk – Toy model of resonant TNO populations and how the North Ecliptic Spur to work. For 2:1 resonance, leading and trailing islands. In baseline sims, 300-500 detected. Without NES, lose much more in leading island than in trailing island; e.g., detections down to 420 and 120. This is ~best case resonance, other resonances will be even more strongly unbalanced.

David Gerdes – DECam Solar System Survey ideas for survey proposal due. Focused on science, but LSST testbed too. Deep TNO survey (26 maybe 26.5 with shift-and-stack), focus on Neptune Trojans; 300 colors / light curves for MBAs; small end size distribution. For LSST, this helps think about a mini deep drilling survey, light curves. Next steps, gather collaborators.

Wes Fraser, Audrey Thirouin, Darin Ragozzine – colors and light curves metric; hardest part is to get correct period. Use LSST cadence from OpSim and power spectrum to make sure that there is a significant peak that matches the period. If there are 30 or more points in given apparition, then the period can usually be recovered. To get resolved light-curve colors, get 30 colors in 2 filters. Number of KBOs that get light curves and colors is about 90% (H=4) on the bright end and 0% at H=8.

Chad Trujillo – outer solar system deep fields; 10, 100, or 1000 square degrees. 2 x 600s images (reaching 26.7). Factor increase of t^(q-1)/4 = ~100 in number. Will post online so others can help. Super-NES could discover ~100 times as many TNOs!

Henry Hsieh – Read the Solar System Roadmap! White paper is coming up that organizes all the things we want. Lays the groundwork for future proposals by identifying software needs and priorities in detail.

Mike Kelley – Cometary Metrics. Added cometary magnitudes; some simulated LSST observations. Jupyter Notebook. One path forward, looking for “blind spots” and “sweet spots” in orbital element space. Goal is to get LSST completeness limits for various cometary populations and writing this up in a papers. Looking at whether NES is important for main belt comets. Looking at Asteroids in Cometary Orbits.

Michael Aye – plans on creating conda package for OpenOrb

Michael Mommert – python OpenOrb, get more information for metrics (e.g., heliocentric longitude, latitude, true anomaly, etc.)

Bryce Bolin – working with Lynne and Ed Lu, simulating detection of impacting asteroids with LSST; with Josef Durech adding real shape information to generate more realistic lightcurves

Thank you to the LSST Corporation and B612 Foundation for supporting and making the 1st LSST Solar System Readiness Sprint possible.


SSSC Sprint Liveblog – SSSC Update by Schwamb and Trilling

Meg Schwamb and David Trilling are Solar System Science Collaboration Co-Chairs.

LSST will produce incredible solar system science and we need to prepare and support our science. We’ve already provided our SSSC Science Roadmap.

Please join and participate in the working groups, currently mostly by email. For example, polls, which currently haven’t had as much participation as we would like.

Also working on white papers (due Nov) to request Northern Ecliptic Spur and other cadence and survey conditions. Here are white paper ideas and putative leaders so far:

  1. Deep Drilling Fields (Trillling)
  2. Northern Ecliptic Spur (Schwamb)
  3. Defense of two observations per field per night (and in what filters)
  4. Two snaps per visit or one?
  5. Rolling cadence?

Goal: get together and spend 3 days making something or writing something that helps us prepare for LSST. You can lead and pitch a project today.

Meg gave some great tips about how to work together and act during the Sprint. Thanks Meg for all your hard work in organizing the Sprint!

SSSC Sprint Liveblog – Lynne Jones

Lynne Jones (UW/DIRAC/LSST) – OpSim Simulations & Small Body Metrics (with MAF = Metrics Analysis Framework)

OpSim generates simulated pointing histories under realistic conditions; it will eventually be the actual LSST scheduler. Pointing history includes 5-sigma depth, seeing, filter, etc. Generating one OpSim output is ~50 CPU-hours.

sims_movingObjects takes a set of orbits/objects and an output from OpSim to generate observations LSST would acquire (passes along OpSim results). Reports SNR (including trailing losses) for a given H. Current status: it is there and it works, but it may not be efficient. Generating 2000 objects (single thread) is ~4 CPU-hours.

Metrics Analysis Framework (MAF) analyses these observations and applies a “metric” to all observations of the same object. Can separate size and orbit properties, as if cloning same orbit over a range of H magnitudes; can also keep H-orbit correlations if you want. For a discovery metric, for example, ~0.1 CPU-hours.

Current metrics (see Cahpter 3 of Community Observing Strategy Evaluation Paper (COSEP) include: discovery (time of first discovery, number of opportunities for discovery – summarized as completeness as a function of H over population); number of observations or total arc length; light curve inversion; color determination.; likelihood of detecting activity for a given object assuming random distribution of activity times. All of these can be summarized as fraction of population as a function of H.

Potential metrics: number of objects lost; accuracy of orbital parameters as a function of time; does an object receive enough observations to determine rotation period / spin pole / phase curve? Will an object need followup on any given night? Improvements on exiting metrics.

You can calculate metrics over H or over orbital properties, etc. These metrics define performance of a given survey strategy in a quantitative way, allowing for quantitative decision making in cadence and other calculations. [Need also a metric to sceince conversion, e.g., metric X going above level Y enables science Z.] Survey Strategy White Papers are going to explore this in detail; many questions for solar system. Snaps (2x15s vs. 1x30s), Wide-Fast-Deep survey cadence? North Ecliptic Spur (NES) footprint and cadence? Pairs in same or different filters? Filter selection criteria? Other mini surveys for solar system? [For reference, a single visit gets several thousand asteroids in a single visit and pointing at the ecliptic.]

The Survey Strategy Committee is going to review all these and provide a recommendation for the initial survey strategy. As the survey progresses, things could be re-evaluated. COSEP is living document, but White Papers are for our best guesses now.

For us to do: be familiar with all these things and contribute. Contribute metrics to sims_maf_contrib. Add sections to COSEP to include supporting text. Defend your science!

So great to have Lynne’s expertise!


SSSC Sprint Liveblog – Mario Juric

Mario Juric – “LSST Solar System Processing Product Owner” for the LSST Solar System Processing Team.

LSST provides 3 types of products to enable Solar System science: Prompt Products = Real-time (60 s) and Daily Catalog; Data Release Orbit Catalog (annual). Data Products Definition Document LSC-163. Daily Catalog is best estimate at that time. Annual Data Release Catalog is like a very well done, self-consistent, single-software, best-result Daily Catalog; will reprocess all the previous data. For example, the Annual Data Release is the most convenient for survey completeness analyses.

Real-time alerts are not precision science, but to enable very rapid follow-up. DIASources (= Difference Image Analysis Sources) are associated with Solar System Objects when known. Every DIASource will be fit with a trailed source model. Comparing PSF models to trailed models should be an excellent method for identifying trailed objects.

Daily (Orbit) Catalog. Why daily? So that LSST can identify objects in tomorrow’s data. Product is orbits, physical properties (H,G), and other things we can study here at the Sprint. The goal is for this catalog to be as complete as possible and cross-matched to external catalogs. This is still to enable rapid follow-up, so software will be updated as needed (e.g., bug fixes). It is not designed for all goals. GAIA catalog. Not clear how the template process is going to work in the early part (first year) of analysis.

Data Release (Annual) Catalog – best for precision photometry and astrometry since computed with self-consistent software and best knowledge of calibration. Note that this will include LSST-data only (not past MPC data).

Current Plan: After observing, find tracklets, run MOPS, recompute catalog (including merging and precovery), make updated SSObject table (“LSSTORB”), and predict for next night. But, currently don’t have firm plan to cross-reference the LSST catalog to the MPC catalog.

New plan would be to send new discoveries to MPC, they produce MPCORB, and then LSST uses MPCORB for its next night predictions. This is a proposal that will probably be detailed by October and approved by the end of the year or so.

Some discussion on possible risks with relying on the MPC, but Mario points out that there are many “third-party risks” in LSST and that there will be back-up plans.

Single tracklets (or even doublets) are, in the initial plan, not reported anywhere. When MPC is able to handle tracklets, those can be passed too.

Team is Joachim Moeyens (40%), Siegfried Eggl (100%), Mario Juric (25%). Lynne Jones and Eric Bellm help. This time is just for solar system; this doesn’t include the other LSST resources which provide all the undergirding.

Glad Mario is on our team!


Ready, Set, SSSC Sprint! – Liveblog

Hello SSSC! Meg has asked me to liveblog the SSSC Sprint today (Tues, 7/10), as we get started. About 20 of us have assembled and we’ve started getting some background from LSST.

John Swinbank, Deputy Project Manager, LSST Data Management
“LSST is not just a telescope, it is an integrated survey system.”

It’s exciting to see LSST coming together physically! John’s showing some good pictures and great updates. For example, the Auxiliary Telescope, the nearby 1.2-m telescope used to help characterize atmospheric absorption, has first light and early data processing happening in the next year (mid 2019).

LSST Camera is the size of a small car: 3.2 gigpixels, 2 second readout, 0.2″ pixels. Focal plane is 63 cm diameter of 189 separate CCDs into 21 3×3 rafts. (Kinda similar to my other favorite, the Kepler Space Telescope!) Each raft is independent. 10 rafts have already been assembled and accepted, all done by January 2019. Rafts can be tested and commissioned independently on telescope, starting mid-2019. Commissioning camera will include science verification surveys, but possibly not (publishable) science.

Data Management (DM) System turns raw data into prompt (e.g. “alert”) and data release data products. 11 Data Releases in 10 years (first one after 6 months). (Note that, at the annual data release, all the prompt products are reproduced, just without alerts.) LSST Science Platform designed to let scientists handle the mountain of data. Includes a portal (queries, etc.), Jupyter-style notebooks to process LSST data yourself, and web API for Data Access Centers based on Virtual Observatory (VO) standards.

All DM code is explicitly open-source and scientists are encouraged to use it. The image processing problem is far more complex than other surveys: many different shapes, blended, etc.

LSE-163 describes all the DM Pipelines and Products. I’m impressed with how well thought out it is.

Current status: none of the pipelines are complete, but many parts of the system are quite usable. Primitives and Algorithms – a rich collection of high-performance tools for working with astronomical data you can pick up and use today. Data Release Processing and Alerts, Alert Distribution – working on test data (Hyper Suprime-Came, DECam, ZTF). Some MOPS is available.

DM also manages the Petascale computing facilities doing data processsing. Science Platform coming together, being used someone.

LSST development and construction is on track and feels like it is accelerating and real!