LSST: From Science Drivers To Reference Design and Anticipated Data Products

Draft version August 13, 2019
Typeset using LATEX twocolumn style in AASTeX62
LSST: from Science Drivers to Reference Design and Anticipated Data Products
Željko Ivezić,1 Steven M. Kahn,2, 3 J. Anthony Tyson,4 Bob Abel,5 Emily Acosta,2 Robyn Allsman,2
David Alonso,6 Yusra AlSayyad,7 Scott F. Anderson,1 John Andrew,2 James Roger P. Angel,8
George Z. Angeli,9 Reza Ansari,10 Pierre Antilogus,11 Constanza Araujo,2 Robert Armstrong,7
Kirk T. Arndt,6 Pierre Astier,11 Éric Aubourg,12 Nicole Auza,2 Tim S. Axelrod,8 Deborah J. Bard,13
Jeff D. Barr,2 Aurelian Barrau,14 James G. Bartlett,12 Amanda E. Bauer,2 Brian J. Bauman,15
Sylvain Baumont,16, 11 Andrew C. Becker,1 Jacek Becla,13 Cristina Beldica,17 Steve Bellavia,18
Federica B. Bianco,19, 20 Rahul Biswas,21 Guillaume Blanc,10, 22 Jonathan Blazek,23, 24 Roger D. Blandford,3
Josh S. Bloom,25 Joanne Bogart,3 Tim W. Bond,13 Anders W. Borgland,13 Kirk Borne,26 James F. Bosch,7
arXiv:0805.2366v5 [astro-ph] 23 May 2018
Dominique Boutigny,27 Craig A. Brackett,13 Andrew Bradshaw,4 William Nielsen Brandt,28

Michael E. Brown,29 James S. Bullock,30 Patricia Burchat,3 David L. Burke,3 Gianpietro Cagnoli,31
Daniel Calabrese,2 Shawn Callahan,2 Alice L. Callen,13 Srinivasan Chandrasekharan,32
Glenaver Charles-Emerson,2 Steve Chesley,33 Elliott C. Cheu,34 Hsin-Fang Chiang,17 James Chiang,3
Carol Chirino,2 Derek Chow,13 David R. Ciardi,35 Charles F. Claver,2 Johann Cohen-Tanugi,36
Joseph J. Cockrum,2 Rebecca Coles,23 Andrew J. Connolly,1 Kem H. Cook,37 Asantha Cooray,30
Kevin R. Covey,38 Chris Cribbs,17 Wei Cui,39 Roc Cutri,35 Philip N. Daly,40 Scott F. Daniel,1 Felipe Daruich,2
Guillaume Daubard,11 Greg Daues,17 William Dawson,15 Francisco Delgado,2 Alfred Dellapenna,18
Robert de Peyster,13 Miguel de Val-Borro,7 Seth W. Digel,13 Peter Doherty,41 Richard Dubois,13
Gregory P. Dubois-Felsmann,35 Josef Durech,42 Frossie Economou,2 Michael Eracleous,28 Henry Ferguson,43
Enrique Figueroa,2 Merlin Fisher-Levine,7 Warren Focke,13 Michael D. Foss,13 James Frank,18
Michael D. Freemon,17 Emmanuel Gangler,44 Eric Gawiser,45 John C. Geary,46 Perry Gee,4 Marla Geha,47
Charles J. B. Gessner,2 Robert R. Gibson,1 D. Kirk Gilmore,3 Thomas Glanzman,13 William Glick,17
Tatiana Goldina,35 Daniel A. Goldstein,25, 48 Iain Goodenow,2 Melissa L. Graham,1 William J. Gressler,2
Philippe Gris,44 Leanne P. Guy,2 Augustin Guyonnet,41 Gunther Haller,13 Ron Harris,49 Patrick A. Hascall,13
Justine Haupt,18 Fabio Hernandez,50 Sven Herrmann,13 Edward Hileman,2 Joshua Hoblitt,2 John A. Hodgson,13
Craig Hogan,51 Dajun Huang,18 Michael E. Huffer,3 Patrick Ingraham,2 Walter R. Innes,3 Suzanne H. Jacoby,2
Bhuvnesh Jain,52 Fabrice Jammes,44 James Jee,4 Tim Jenness,2 Garrett Jernigan,53 Darko Jevremović,54
Kenneth Johns,34 Anthony S. Johnson,13 Margaret W. G. Johnson,17 R. Lynne Jones,1 Claire Juramy-Gilles,11
Mario Jurić,1 Jason S. Kalirai,43 Nitya J. Kallivayalil,55 Bryce Kalmbach,1 Jeffrey P. Kantor,2
Pierre Karst,56 Mansi M. Kasliwal,57 Heather Kelly,13 Richard Kessler,51 Veronica Kinnison,2
David Kirkby,58 Lloyd Knox,4 Ivan V. Kotov,18 Victor L. Krabbendam,2 K. Simon Krughoff,2 Petr Kubánek,59
John Kuczewski,18 Shri Kulkarni,57 John Ku,13 Nadine R. Kurita,13 Craig S. Lage,4 Ron Lambert,2, 60
Travis Lange,13 J. Brian Langton,13 Laurent Le Guillou,16, 11 Deborah Levine,35 Ming Liang,2 Kian-Tat Lim,13
Chris J. Lintott,6 Kevin E. Long,61 Margaux Lopez,13 Paul J. Lotz,2 Robert H. Lupton,7 Nate B. Lust,7
Lauren A. MacArthur,7 Ashish Mahabal,57 Rachel Mandelbaum,62 Darren S. Marsh,13 Philip J. Marshall,3
Stuart Marshall,3 Morgan May,18 Robert McKercher,2 Michelle McQueen,18 Joshua Meyers,7
Myriam Migliore,14 Michelle Miller,49 David J. Mills,2 Connor Miraval,18 Joachim Moeyens,1
David G. Monet,63 Marc Moniez,10 Serge Monkewitz,35 Christopher Montgomery,2 Fritz Mueller,13
Gary P. Muller,2 Freddy Muñoz Arancibia,2 Douglas R. Neill,2 Scott P. Newbry,13 Jean-Yves Nief,50
Andrei Nomerotski,18 Martin Nordby,13 Paul O’Connor,18 John Oliver,41, 64 Scot S. Olivier,15 Knut Olsen,49
William O’Mullane,2 Sandra Ortiz,2 Shawn Osier,13 Russell E. Owen,1 Reynald Pain,11 Paul E. Palecek,18
John K. Parejko,1 James B. Parsons,17 Nathan M. Pease,13 J. Matt Peterson,2 John R. Peterson,39
Donald L. Petravick,17 M. E. Libby Petrick,2 Cathy E. Petry,2 Francesco Pierfederici,65
Stephen Pietrowicz,17 Rob Pike,66 Philip A. Pinto,8 Raymond Plante,17 Stephen Plate,18 Paul A. Price,7
Michael Prouza,59 Veljko Radeka,18 Jayadev Rajagopal,49 Andrew P. Rasmussen,39 Nicolas Regnault,11
Kevin A. Reil,13 David J. Reiss,1 Michael A. Reuter,2 Stephen T. Ridgway,49 Vincent J. Riot,15 Steve Ritz,67
Sean Robinson,18 William Roby,35 Aaron Roodman,13 Wayne Rosing,68 Cecille Roucelle,12
Matthew R. Rumore,18 Stefano Russo,13 Abhijit Saha,49 Benoit Sassolas,31 Terry L. Schalk,67
Pim Schellart,7, 69 Rafe H. Schindler,3 Samuel Schmidt,4 Donald P. Schneider,28 Michael D. Schneider,15
William Schoening,2 German Schumacher,2, 60 Megan E. Schwamb,70 Jacques Sebag,2 Brian Selvy,2
Glenn H. Sembroski,39 Lynn G. Seppala,15 Andrew Serio,2 Eduardo Serrano,2 Richard A. Shaw,43 Ian Shipsey,6
Jonathan Sick,2 Nicole Silvestri,1 Colin T. Slater,1 J. Allyn Smith,71 R. Chris Smith,60 Shahram Sobhani,72
Christine Soldahl,13 Lisa Storrie-Lombardi,35 Edward Stover,2 Michael A. Strauss,7 Rachel A. Street,68
Christopher W. Stubbs,41, 64 Ian S. Sullivan,1 Donald Sweeney,2 John D. Swinbank,1, 7 Alexander Szalay,73
Peter Takacs,18 Stephen A. Tether,13 Jon J. Thaler,74 John Gregg Thayer,13 Sandrine Thomas,2
Vaikunth Thukral,13 Jeffrey Tice,13 David E. Trilling,75 Max Turri,13 Richard Van Berg,13, 52
Daniel Vanden Berk,76 Kurt Vetter,18 Francoise Virieux,12 Tomislav Vucina,2 William Wahl,18
2 Ivezić, Kahn, Tyson, Abel, Acosta, Allsman, Alonso, AlSayyad, Anderson, et al.
Lucianne Walkowicz,77, 78 Brian Walsh,18 Christopher W. Walter,79 Daniel L. Wang,13 Shin-Yawn Wang,35
Michael Warner,60 Oliver Wiecha,2 Beth Willman,2, 8 Scott E. Winters,15 David Wittman,4 Sidney C. Wolff,2
W. Michael Wood-Vasey,80 Xiuqin Wu,35 Bo Xin,2 Peter Yoachim,1 and Hu Zhan81
1 University of Washington, Dept. of Astronomy, Box 351580, Seattle, WA 98195
2 LSST Project Office, 950 N. Cherry Avenue, Tucson, AZ 85719
3 Kavli Institute for Particle Astrophysics and Cosmology, SLAC National Accelerator Laboratory, Stanford University, Stanford, CA
94025
4 Physics Department, University of California, One Shields Avenue, Davis, CA 95616
5 Olympic College, 1600 Chester Ave., Bremerton, WA 98337-1699
6 Department of Physics, University of Oxford, Denys Wilkinson Building, Keble Road, Oxford, OX1 3RH, UK
7 Department of Astrophysical Sciences, Princeton University, Princeton, NJ 08544
8 Steward Observatory, The University of Arizona, 933 N Cherry Ave., Tucson, AZ 85721
9 Giant Magellan Telescope Organization (GMTO), 465 N. Halstead Street, Suite 250, Pasadena, CA 91107
10 Laboratoire de l’Accélérateur Linéaire, CNRS/IN2P3, Université de Paris-Sud, B.P. 34, 91898 Orsay Cedex, France
11 Laboratoire de Physique Nucléaire et des Hautes Energies, Université Pierre et Marie Curie, Université Paris Diderot, CNRS/IN2P3, 4
place Jussieu, 75005 Paris, France

12 AstroParticule et Cosmologie, Université Paris Diderot, CNRS/IN2P3, CEA/lrfu, Observatoire de Paris, Sorbonne Paris Cité, 10, rue
Alice Domon et Léonie Duquet, Paris Cedex 13, France

13 SLAC National Accelerator Laboratory, 2575 Sand Hill Rd, Menlo Park CA 94025
14 Laboratoire de Physique Subatomique et de Cosmologie, Université Grenoble-Alpes, CNRS/IN2P3, 53 av. des Martyrs, 38026 Grenoble
cedex, France
15 Lawrence Livermore National Laboratory, 7000 East Avenue, Livermore, CA 94550
16 Sorbonne Universités, UPMC Univ Paris 06, UMR 7585, LPNHE, F-75005, Paris, France
17 NCSA, University of Illinois at Urbana-Champaign, 1205 W. Clark St., Urbana, IL 61801
18 Brookhaven National Laboratory, Upton, NY 11973
19 Center for Urban Science & Progress, New York University, Brooklyn, NY 11021
20 Center for Cosmology & Particle Physics, New York University, New York, 10012
21 Oskar Klein Centre, Department of Physics, Stockholm University, SE 106 91 Stockholm, Sweden
22 Université Paris Diderot, Sorbonne Paris Cité, F-75013 Paris, France
23 Center for Cosmology and Astro-Particle Physics, The Ohio State University, Columbus, OH 43210, USA
24 Institute of Physics, Laboratory of Astrophysics, École Polytechnique Fedèrale de Lausanne (EPFL), Observatoire de Sauverny, 1290
Versoix, Switzerland
25 Astronomy Department, University of California, 601 Campbell Hall, Berkeley, CA 94720
26 School of Physics, Astronomy and Computational Sciences, George Mason University, 4400 University Drive, Fairfax, VA 22030
27 Université Grenoble-Alpes, Université Savoie Mont Blanc, CNRS/IN2P3 Laboratoire d’Annecy-le-Vieux de Physique des Particules, 9
Chemin de Bellevue – BP 110, 74940 Annecy-le-Vieux Cedex, France

28 Department of Astronomy and Astrophysics, The Pennsylvania State University, 525 Davey Lab, University Park, PA 16802
29 Division of Geological and Planetary Sciences, California Institute of Technology, Pasadena, CA 91125
30 Center for Cosmology, University of California, Irvine, CA 92697
31 Laboratoire des Materiaux Avances (LMA), CNRS/IN2P3, Université de Lyon, F-69622 Villeurbanne, Lyon, France
32 Department of Computer Science, The University of Arizona, 1040 E 4th St, Tucson, AZ 85719
33 Jet Propulsion Laboratory, California Institute of Technology, Pasadena, CA 91109
34 Department of Physics, University of Arizona, 1118 E. Fourth Street, Tucson, AZ 85721
35 IPAC, California Institute of Technology, MS 100-22, Pasadena, CA 91125
36 Laboratoire Univers et Particules de Montpellier (LUPM), Université de Montpellier, CNRS/IN2P3, Place Eugène Bataillon, 34095
Montpellier, France
37 Cook Astronomical Consulting, 220 Duxbury CT, San Ramon, CA 94583, USA
38 Western Washington University, 516 High Street, Bellingham, WA 98225
39 Department of Physics and Astronomy, Purdue University, 525 Northwestern Ave., West Lafayette, IN 47907
40 University of Arizona, Tucson, AZ 85721
41 Department of Physics, Harvard University, 17 Oxford St, Cambridge MA 02138
42 Astronomical Institute, Charles University, Praha, Czech Republic
43 Space Telescope Science Institute, 3700 San Martin Drive, Baltimore, MD 21218
44 Université Clermont Auvergne, CNRS, Laboratoire de Physique de Clermont, F-63000 Clermont-Ferrand, France
45 Department of Physics and Astronomy, Rutgers University, 136 Frelinghuysen Rd, Piscataway, NJ 08854
46 Smithsonian Astrophysical Observatory, 60 Garden St., Cambridge MA 02138
47 Astronomy Department, Yale University, New Haven, CT 06520
The Large Synoptic Survey Telescope 3
48 Lawrence Berkeley National Laboratory, 1 Cyclotron Road, Berkeley, CA 94720, USA

49 National Optical Astronomy Observatory, 950 N. Cherry Ave, Tucson, AZ 85719
50 CNRS, CC-IN2P3, 21 avenue Pierre de Coubertin, CS70202, 69627 Villeurbanne cedex, France
51 Department of Astronomy and Astrophysics, University of Chicago, 5640 South Ellis Avenue, Chicago, IL 60637
52 Department of Physics & Astronomy, University of Pennsylvania, 209 South 33rd Street, Philadelphia, PA 19104-6396
53 Space Sciences Lab, University of California, 7 Gauss Way, Berkeley, CA 94720-7450
54 Astronomical Observatory, Volgina 7, P.O. Box 74, 11060 Belgrade, Serbia
55 Department of Astronomy, University of Virginia, Charlottesville, VA 22904
56 Aix Marseille Univ, CNRS/IN2P3, CPPM, Marseille, France
57 Astronomy Department, California Institute of Technology, 1200 East California Blvd., Pasadena CA 91125
58 Department of Physics and Astronomy, University of California, 4129 Frederick Reines Hall, Irvine, CA 92697
59 Institute of Physics, Academy of Sciences of the Czech Republic, Na Slovance 2, 182 21 Praha 8, Czech Republic
60 Cerro Tololo InterAmerican Observatory, La Serena, Chile
61 Longhorn Industries, Ellensburg, WA 98926
62 McWilliams Center for Cosmology, Department of Physics, Carnegie Mellon University, Pittsburgh, PA 15213, USA
63 U.S. Naval Observatory Flagstaff Station, 10391 Naval Observatory Road, Flagstaff, AZ 86001
64 Department of Astronomy, Center for Astrophysics, Harvard University, 60 Garden St., Cambridge, MA 02138
65 Instituto de Radioastronomı́a Milimétrica, Av. Divina Pastora 7, Núcleo Central, E-18012 Granada, Spain
66 Google Inc., 1600 Amphitheatre Parkway Mountain View, CA 94043
67 Santa Cruz Institute for Particle Physics and Physics Department, University of California–Santa Cruz, 1156 High St., Santa Cruz,
CA 95064
68 Las Cumbres Observatory, 6740 Cortona Dr. Suite 102, Santa Barbara, CA 93117
69 Department of Astrophysics/IMAPP, Radboud University Nijmegen, P.O. Box 9010, 6500 GL Nijmegen, The Netherlands
70 Gemini Observatory, Northern Operations Center, 670 North A’ohoku Place, Hilo, HI 96720, USA
71 Austin Peay State University, Clarksville, TN 37044
72 Belldex IT Consulting, Tucson, AZ 85742
73 Department of Physics and Astronomy, The John Hopkins University, 3701 San Martin Drive, Baltimore, MD 21218
74 University of Illinois, Physics and Astronomy Departments, 1110 W. Green St., Urbana, IL 61801
75 Department of Physics and Astronomy, Northern Arizona University, PO Box 6010, Flagstaff, AZ 86011, USA
76 Saint Vincent College, Department of Physics, 300 Fraser Purchase Road, Latrobe, PA 15650
77 Library of Congress, 101 Independence Ave SE, Washington, DC 20540
78 The Adler Planetarium, 1300 South Lakeshore Ave, Chicago, IL 60605, USA
79 Department of Physics, Duke University, Durham, NC 27708
80 Department of Physics and Astronomy, University of Pittsburgh, 3941 O’Hara Street, Pittsburgh PA 15260
81 Key Laboratory of Optical Astronomy, National Astronomical Observatories, Chinese Academy of Sciences, 20A Datun Road,
Chaoyang District, Beijing 100012, China
ABSTRACT
Major advances in our understanding of the Universe frequently arise from dramatic improvements
in our ability to accurately measure astronomical quantities. Aided by rapid progress in information
technology, current sky surveys are changing the way we view and study the Universe. Next-generation
surveys will maintain this revolutionary progress. We describe here the most ambitious survey currently
planned in the optical, the Large Synoptic Survey Telescope (LSST). A vast array of science will be
enabled by a single wide-deep-fast sky survey, and LSST will have unique survey capability in the
faint time domain. The LSST design is driven by four main science themes: probing dark energy and
dark matter, taking an inventory of the Solar System, exploring the transient optical sky, and mapping
the Milky Way. LSST will be a large, wide-field ground-based system designed to obtain repeated
images covering the sky visible from Cerro Pachón in northern Chile. The telescope will have an 8.4
m (6.5 m effective) primary mirror, a 9.6 deg2 field of view, and a 3.2 Gigapixel camera. The standard
observing sequence will consist of pairs of 15-second exposures in a given field, with two such visits in
each pointing in a given night to identify and constrain the orbits of asteroids. With these repeats,
the LSST system is capable of imaging about 10,000 square degrees of sky in a single filter in three
clear nights. The typical 5σ point-source depth in a single visit in r will be ∼ 24.5 (AB). The system
is designed to yield high image quality as well as superb astrometric and photometric accuracy. The
project is in the construction phase and will begin regular survey operations by 2022. The survey area
will be contained within 30,000 deg2 with δ < +34.5◦ , and will be imaged multiple times in six bands,
ugrizy, covering the wavelength range 320–1050 nm. About 90% of the observing time will be devoted
to a deep-wide-fast survey mode which will uniformly observe a 18,000 deg2 region about 800 times
(summed over all six bands) during the anticipated 10 years of operations, and will yield a coadded
map to r ∼ 27.5. These data will result in databases including 20 billion galaxies and a similar number
of stars, and will serve the majority of the primary science programs. The remaining 10% of the
observing time will be allocated to special projects such as a Very Deep and Fast time domain survey,
whose details are currently under discussion. We illustrate how the LSST science drivers led to these
choices of system parameters, and describe the expected data products and their characteristics. The
goal is to make LSST data products including a relational database of about 32 trillion observations
of 40 billion objects available to the public and scientists around the world – everyone will be able to
view and study a high-definition color movie of the deep Universe.
Keywords: astronomical data bases: atlases, catalogs, surveys — Solar System — stars — the Galaxy
— galaxies — cosmology
1. INTRODUCTION nearly all fields of astronomy – and several areas of fun-

Major advances in our understanding of the Universe damental physics. In addition, the world-wide atten-
have historically arisen from dramatic improvements in tion received by Sky in Google Earth1 (Scranton et al.
our ability to “see”. We have developed progressively 2007), the World Wide Telescope2 , and the hundreds
larger telescopes over the past century, allowing us to of thousands of volunteers classifying galaxies in the
peer further into space, and further back in time. With Galaxy Zoo project (Lintott et al. 2011) and its ex-
the development of advanced instrumentation – imagers, tensions demonstrate that the impact of sky surveys
spectrographs, and polarimeters – we have been able extends far beyond fundamental science progress and
to parse radiation detected from distant sources over reaches all of society.
the full electromagnetic spectrum in increasingly sub- Motivated by the evident scientific progress enabled
tle ways. These data have provided the detailed infor- by large sky surveys, three nationally-endorsed reports
mation needed to construct physical models of planets, by the U.S. National Academy of Sciences (National
stars, galaxies, quasars, and larger structures, and to Research Council 2001, 2003a,b) concluded that a ded-
probe the new physics of dark matter and dark energy. icated ground-based wide-field imaging telescope with
Until recently, most astronomical investigations have an effective aperture of 6–8 meters “is a high priority
focused on small samples of cosmic sources or individual for planetary science, astronomy, and physics over the
objects. This is because our largest telescope facilities next decade.” The Large Synoptic Survey Telescope
typically had rather small fields of view, and those with (LSST) described here is such a system. Located on
large fields of view could not detect very faint sources. Cerro Pachón in northern Chile, the LSST will be a
With all of our existing telescope facilities, we have still large, wide-field ground-based telescope designed to ob-
surveyed only a small fraction of the observable Universe tain multi-band images over a substantial fraction of the
(except when considering the most luminous quasars). sky every few nights. The survey will yield contiguous
Over the past two decades, however, advances in tech- overlapping imaging of over half the sky in six optical
nology have made it possible to move beyond the tradi- bands, with each sky location visited close to 1000 times
tional observational paradigm and to undertake large- over 10 years. The 2010 report “New Worlds, New Hori-
scale sky surveys. As vividly demonstrated by sur- zons in Astronomy and Astrophysics” by the NRC Com-
veys such as the Sloan Digital Sky Survey (SDSS; York mittee for a Decadal Survey of Astronomy and Astro-
et al. 2000), the Two Micron All Sky Survey (2MASS; physics (National Research Council 2010) ranked LSST
Skrutskie et al. 2006), the Galaxy Evolution Explorer as its top priority for large ground-based projects, and
(GALEX; Martin et al. 2005), and Gaia (Gaia Collab- in May 2014 the National Science Board approved the
oration et al. 2016) to name but a few, sensitive and project for construction. As of this writing, the LSST
accurate multi-color surveys over a large fraction of the construction phase is close to the peak of activity. Af-
sky enable an extremely broad range of new scientific
investigations. These projects, based on a synergy of 1 https://www.google.com/sky/
advances in telescope construction, detectors, and above 2 http://worldwidetelescope.org/home
all, information technology, have dramatically impacted
ter initial tests with a commissioning camera and full 90% of the observing time will be devoted to a deep-
commissioning with the main camera, the ten year sky wide-fast (main) survey mode. The working paradigm
survey is projected to begin in 2022. is that all scientific investigations will utilize a common
The purpose of this paper is to provide an overall sum- database constructed from an optimized observing pro-
mary of the main LSST science drivers and how they gram (the main survey mode), such as that discussed in
led to the current system design parameters (§ 2), to § 3.1. Here we briefly describe these science goals and
describe the anticipated data products (§ 3), and to pro- the most challenging requirements for the telescope and
vide a few examples of the science programs that LSST instrument that are derived from those goals, which will
will enable (§ 4). The community involvement is dis- inform the overall system design decisions discussed be-
cussed in § 5, and broad educational and societal im- low. For a more detailed discussion, we refer the reader
pacts of the project in § 6. Concluding remarks are pre- to the LSST Science Requirements Document (Ivezić
sented in § 7. This publication will be maintained at the & The LSST Science Collaboration 2011), the LSST
arXiv.org site3 , and will also be available from the LSST Science Book (LSST Science Collaboration et al. 2009,
website (http://www.lsst.org). The latest arXiv version hereafter SciBook), and links to technical papers and
of this paper should be consulted and referenced for the presentations at https://www.lsst.org/scientists.
most up-to-date information about the LSST system.
2.1. The Main Science Drivers
2. FROM SCIENCE DRIVERS TO REFERENCE
The main science drivers are used to optimize various
DESIGN
system parameters. Ultimately, in this high-dimensional
The most important characteristic that determines parameter space, there is a manifold defined by the to-
the speed at which a system can survey a given sky area tal project cost. The science drivers must both justify
to a given flux limit (i.e., its depth) is its étendue (or this cost, as well as provide guidance on how to opti-
grasp), the product of its primary mirror area and the mize various parameters while staying within the cost
angular area of its field of view (for a given set of ob- envelope.
serving conditions, such as seeing and sky brightness). Here we summarize the dozen or so most important
The effective étendue for LSST will be greater than 300 interlocking constraints on data and system properties
m2 deg2 , which is more than an order of magnitude placed by the four main science themes:
larger than that of any existing facility. For example,
the SDSS, with its 2.5-m telescope (Gunn et al. 2006) 1. The depth of a single visit to a given field;
and a camera with 30 imaging CCDs (Gunn et al. 1998),
2. Image quality;
has an effective étendue of only 5.9 m2 deg2 .
The range of scientific investigations which will be en- 3. Photometric accuracy;
abled by such a dramatic improvement in survey capa-
bility is extremely broad. Guided by the community- 4. Astrometric accuracy;
wide input assembled in the report of the Science Work- 5. Optimal exposure time;
ing Group of the LSST in 2004 (Science Working Group
of the LSST & Strauss 2004), the LSST is designed to 6. The filter complement;
achieve goals set by four main science themes:
7. The distribution of revisit times (i.e., the cadence
1. Probing Dark Energy and Dark Matter; of observations), including the survey lifetime;
2. Taking an Inventory of the Solar System; 8. The total number of visits to a given area of sky;
3. Exploring the Transient Optical Sky; 9. The coadded survey depth;
4. Mapping the Milky Way. 10. The distribution of visits on the sky, and the total
sky coverage;
Each of these four themes itself encompasses a vari-
ety of analyses, with varying sensitivity to instrumental 11. The distribution of visits per filter; and
and system parameters. These themes fully exercise the
technical capabilities of the system, such as photomet- 12. Parameters characterizing data processing and
ric and astrometric accuracy and image quality. About data access (such as the maximum time allowed
after each exposure to report transient sources,
and the maximum allowed software contribution
3 https://arxiv.org/abs/0805.2366 to measurement errors).
We present a detailed discussion of how these science- Weak lensing (WL) techniques can be used to map the
driven data properties are transformed to system pa- distribution of mass as a function of redshift and thereby
rameters below. trace the history of both the expansion of the Uni-
verse and the growth of structure (e.g., Hu & Tegmark
2.1.1. Probing Dark Energy and Dark Matter 1999; for recent reviews see Kilbinger 2015; Mandel-
baum 2017). Measurements of cosmic shear as a func-
Current models of cosmology require the existence of
tion of redshift allow determination of angular distances
both dark matter and dark energy to match observa-
versus cosmic time, providing multiple independent con-
tional constraints (Riess et al. 2007; Komatsu et al. 2009;
straints on the nature of dark energy. These investiga-
Percival et al. 2010; LSST Dark Energy Science Col-
tions require deep wide-area multi-color imaging with
laboration 2012; Weinberg et al. 2015), and references
stringent requirements on shear systematics in at least
therein). Dark energy affects the cosmic history of both
two bands, and excellent photometry in at least five
the Hubble expansion and mass clustering. Distinguish-
bands to measure photometric redshifts (a requirement
ing competing models for the physical nature of dark en-
shared with LSS, and indeed all extragalactic science
ergy, or alternative explanations involving modifications
drivers). The strongest constraints on the LSST im-
of the General Theory of Relativity, will require percent
age quality arise from this science program. In order
level measurements of both the cosmic expansion and
to control systematic errors in shear measurement, the
the growth of dark matter structure as a function of
desired depth must be achieved with many short expo-
redshift. Any given cosmological probe is sensitive to,
sures (allowing for systematics in the measurement of
and thus constrains degenerate combinations of, several
galaxy shapes related to the PSF and telescope point-
cosmological and astrophysical/systematic parameters.
ing to be diagnosed and removed). Detailed simula-
Therefore the most robust cosmological constraints are
tions of weak lensing techniques show that imaging over
the result of using interlocking combinations of probes.
∼ 20, 000 deg2 to a 5σ point-source depth of rAB ∼ 27.5
The most powerful probes include weak gravitational
gives adequate signal to measure shapes for of order 2
lens cosmic shear (WL), galaxy clustering and baryon
billion galaxies for weak lensing. These numbers are
acoustic oscillations (LSS), the mass function and clus-
adequate to reach Stage IV goals for dark energy, as de-
tering of clusters of galaxies, time delays in lensed quasar
fined by the Dark Energy Task Force (Albrecht et al.
and supernova systems (SL), and photometry of type
2006). This depth, and the corresponding deep sur-
Ia supernovae (SN) – all as functions of redshift. Us-
face brightness limit, optimize the number of galaxies
ing the cosmic microwave background fluctuations as
with measured shapes in ground-based seeing, and al-
the normalization, the combination of these probes can
low their detection in significant numbers to beyond a
yield the needed precision to distinguish among mod-
redshift of two. Analyzing these data will require sophis-
els of dark energy (see e.g., Zhan 2006, and references
ticated data processing techniques. For example, rather
therein). The challenge is to turn this available precision
than simply coadding all images in a given region of sky,
into accuracy, by careful modeling and marginalization
the individual exposures, each with their own PSF and
over a variety of systematic effects (see e.g., Krause &
noise characteristics, should be analyzed simultaneously
Eifler 2017).
to optimally measure the shapes of galaxies (Tyson et al.
Meanwhile, there are a number of astrophysical probes
2008; Jee & Tyson 2011).
of the fundamental properties of dark matter worth ex-
Type Ia supernovae provided the first robust evidence
ploring, including, for example, weak and strong lensing
that the expansion of the Universe is accelerating (Riess
observations of the mass distribution in galaxies and iso-
et al. 1998; Perlmutter et al. 1999). To fully exploit
lated and merging clusters, in conjunction with dynami-
the supernova science potential, light curves sampled in
cal and X-ray observations (see e.g., Dawson et al. 2012;
multiple bands every few days over the course of a few
Newman et al. 2013; Rocha et al. 2013), the numbers
months are required. This is essential to search for sys-
and gamma-ray emission from dwarf satellite galaxies
tematic differences in supernova populations (e.g., due
(see e.g., Hargis et al. 2014; Drlica-Wagner et al. 2015),
to differing progenitor channels) which may masquerade
the subtle perturbations of stellar streams in the Milky
as cosmological effects, as well as to determine photo-
Way halo by dark matter substructure (Belokurov &
metric redshifts from the supernovae themselves. Unlike
Koposov 2016), and massive compact halo object mi-
other cosmological probes, even a single object gives in-
crolensing (Alcock et al. 2001).
formation on the relationship between redshift and dis-
Three of the primary Dark Energy probes, WL, LSS
tance. Thus a large number of SN across the sky allows
and SN, provide unique and independent constraints on
one to search for any dependence of dark energy prop-
the LSST system design (SciBook Ch. 11–15).
erties on direction, which would be an indicator of new The small-body populations in the Solar System,
physics. The results from this method can be compared such as asteroids, trans-Neptunian objects (TNOs) and
with similar measures of anisotropy from the combina- comets, are remnants of its early assembly. The history
tion of WL and LSS (Zhan et al. 2009). Given the ex- of accretion, collisional grinding, and perturbation by
pected SN flux distribution at the redshifts where dark existing and vanished giant planets is preserved in the
energy is important, the single visit depth should be at orbital elements and size distributions of those objects.
least r ∼ 24. Good image quality is required to separate Cataloging the orbital parameters, size distributions,
SN photometrically from their host galaxies. Observa- colors and light curves of these small-body populations
tions in at least five photometric bands will allow proper requires a large number of observations in multiple fil-
K-corrected light curves to be measured over a range ters, and will lead to insights into planetary formation
of redshift. Carrying out these K-corrections requires and evolution by providing the basis and constraints
that the calibration of the relative offsets in photomet- for new theoretical models. In addition, collisions in the
ric zero points between filters and the system response main asteroid belt between Mars and Jupiter still occur,
functions, especially near the edges of bandpasses, be and occasionally eject objects on orbits that may place
accurate to about 1% (Wood-Vasey et al. 2007), sim- them on a collision course with Earth. Studying the
ilar to the requirements from photometric redshifts of properties of main belt asteroids at sub-kilometer sizes
galaxies. Deeper data (r > 26) for small areas of the is important for linking the near-Earth Object (NEO)
sky can extend the discovery of SN to a mean redshift population with its source in the main belt. About 20%
of 0.7 (from ∼ 0.5 for the main survey), with some ob- of NEOs, the potentially hazardous asteroids (PHAs),
jects beyond z ∼1 (Garnavich et al. 2004; Pinto et al. are in orbits that pass sufficiently close to Earth’s orbit,
2004, SciBook Ch. 11). The added statistical leverage on to within 0.05 AU, that perturbations on time scales
the “pre-acceleration” era (z & 1) would improve con- of a century can lead to the possibility of collision. In
straints on the properties of dark energy as a function December 2005, the U.S. Congress directed5 NASA to
of redshift. implement a survey that would catalog 90% of NEOs
Finally, there will be powerful cross checks and com- with diameters larger than 140 meters by 2020.
plementarities with other planned or proposed surveys, Discovering and linking objects in the Solar System
such as Euclid (Laureijs et al. 2011) and WFIRST moving with a wide range of apparent velocities (from
(Spergel et al. 2015), which will provide wide-field several degrees per day for NEOs to a few arc seconds
optical-IR imaging from space; DESI (Levi et al. 2013) per day for the most distant TNOs) places strong con-
and PFS (Takada et al. 2014), which will measure spec- straints on the cadence of observations, requiring closely
troscopic BAO with millions of galaxies; and SKA4 spaced pairs of observations (two or preferably three
(radio). Large survey volumes are key to probing dy- times per lunation) in order to link detections unam-
namical dark energy models (with sub-horizon dark biguously and derive orbits (SciBook Ch. 5). Individual
energy clustering or anisotropic stresses). The cross- exposures should be shorter than about 30 seconds to
correlation of the three-dimensional mass distribution – minimize the effects of trailing for the majority of mov-
as probed by neutral hydrogen in CHIME (Newburgh ing objects. The images must be well sampled to en-
et al. 2014), HIRAX (Newburgh et al. 2016) or SKA, able accurate astrometry, with absolute accuracy of at
or galaxies in DESI and PFS – with the gravitational least 0.1 arcsec in order to measure orbital parameters
growth probed by tomographic shear in LSST will be a of TNOs with enough precision to constrain theoreti-
complementary way to constrain dark energy properties cal models and enable prediction of occultations. The
beyond simply characterizing its equation of state and photometry should be better than 1–2% to measure as-
to test the underlying theory of gravity. Current and fu- teroids’ colors and thus determine their types. The dif-
ture ground-based CMB experiments, such as Advanced ferent filters should be observed over a short time span
ACT (De Bernardis et al. 2016), SPT-3G (Benson et al. to reduce apparent variations in color due to changes in
2014), Simons Observatory, and CMB Stage-4 (Abaza- observing geometry, but should be repeated over many
jian et al. 2016), will also offer invaluable opportunities lunations in order to determine phase curves and allow
for cross-correlations with secondary CMB anisotropies. shape modeling.
The Congressional mandate can be fulfilled with a
2.1.2. Taking an Inventory of the Solar System 10-meter-class telescope equipped with a multi-gigapixel
5 For details see http://neo.jpl.nasa.gov/neo/report2007.html

4 https://www.skatelescope.org
camera, and a sophisticated and robust data processing to detect the parallax microlensing signal of intermedi-
system (Ivezić et al. 2007a). The images should reach a ate mass black holes and measure their masses (Gould
depth of at least 24.5 (5σ for point sources) in the r band 1992). It would open the possibility of discovering popu-
to reach high completeness down to the 140 m mandate lations of binaries and planets via transits (e.g., Beaulieu
for NEOs. Such an instrument would probe the ∼100 et al. 2006; Drake et al. 2010; Choi et al. 2013; Batista
m size range at main-belt distances, and discover rare et al. 2014), as well as obtaining spectra of lensed stars
distant TNOs such as Sedna (Brown et al. 2004) and in distant galaxies.
2012 VP113 (Trujillo & Sheppard 2014). A deep and persistent survey will discover precursors
of explosive and eruptive transients, generate large sam-
2.1.3. Exploring the Transient Optical Sky ples of transients whose study has thus far been limited
by small sample size (e.g., different subtypes of core col-
Recent surveys have shown the power of measuring
lapse SN, Bianco et al. 2014.)
variability of celestial sources for studying gravitational
Time series ranging between one minute and ten years
lensing, searching for supernovae, determining the phys-
cadence should be probed over a significant fraction
ical properties of gamma-ray burst sources, discovering
of the sky. The survey’s cadence will be sufficient,
gravitational wave counterparts, probing the structure
combined with the large coverage, to serendipitously
of active galactic nuclei, studying variable star popula-
catch very short-lived events, such as eclipses in ultra-
tions, discovering exoplanets, and many other subjects
compact double degenerate binary systems (Anderson
at the forefront of astrophysics (SciBook Ch. 8; Law
et al. 2005), to constrain the properties of fast faint tran-
et al. 2009; Djorgovski et al. 2012; Rowe et al. 2014).
sients (such as optical flashes associated with gamma-
Time-domain science has diverse requirements for
ray bursts; Bloom et al. 2008), to detect electromag-
transient and variable phenomena that are physically
netic counterparts to gravitational wave sources (Nis-
and phenomenologically heterogeneous. It requires large
sanke et al. 2013; Scolnic et al. 2018) and to further con-
area coverage to enhance the probability of detecting
strain the properties of new classes of transients discov-
rare events; good image quality to enable differencing of
ered by programs such as the Deep Lens Survey (Becker
images, especially in crowded fields; good time sampling,
et al. 2004), the Catalina Real-time Transient Survey
necessary to distinguish different types of variables and
(Drake et al. 2009), the Palomar Transient Factory (Law
to infer their properties (e.g., determining the intrinsic
et al. 2009), and the Zwicky Transient Factory (Bellm
peak luminosity of Type Ia supernovae requires measur-
2014). Observations over a decade will enable the study
ing their light curve shape); accurate color information
of long period variables, intermediate mass black holes,
to classify variable objects; long term persistent obser-
and quasars (Kaspi et al. 2007; MacLeod et al. 2010;
vations to characterize slow-evolving transients (e.g.,
Graham et al. 2014; Chapline & Frampton 2016).
tidal disruption events, super luminous supernovae at
The next frontier in this field will require measuring
high redshift, and luminous blue variables); and rapid
the colors of fast transients, and probing variability at
data reduction, classification, and reporting to the com-
faint magnitudes. Classification of transients in close-
munity to allow immediate follow-up with spectroscopy,
to-real time will require access to the full photometric
further optical photometry, and imaging in other wave-
history of the objects, both before and after the transient
bands.
event (e.g., Mahabal et al. 2011).
Wide area, dense temporal coverage to deep limiting
magnitudes will enable the discovery and analysis of rare
2.1.4. Mapping the Milky Way
and exotic objects such as neutron stars and black hole
binaries, novae and stellar flares, gamma-ray bursts and A major challenge in extragalactic cosmology today
X-ray flashes, active galactic nuclei, stellar disruptions concerns the formation of structure on sub-galactic
by black holes (Bloom et al. 2011; Gezari et al. 2012), scales, where baryon physics becomes important, and
and possibly new classes of transients, such as binary the nature of dark matter may manifest itself in ob-
mergers of supermassive black holes (Shields & Bonning servable ways (e.g. Weinberg et al. 2015). The Milky
2008), chaotic eruptions on stellar surfaces (Arnett & Way and its environment provide a unique dataset for
Meakin 2011), and, further yet, completely unexpected understanding the detailed processes that shape galaxy
phenomena. formation and for testing the small-scale predictions of
Such a survey would likely detect microlensing by stars our standard cosmological model. New insights into
and compact objects in the Milky Way, but also in the the nature and evolution of the Milky Way will re-
Local Group and perhaps beyond (de Jong et al. 2008). quire wide-field surveys to constrain its structure and
Given the duration of the LSST it will also be possible accretion history. Further insights into the stellar pop-
ulations that make up the Milky Way can be gained

with a comprehensive census of the stars within a few
hundred pc of the Sun.
Mapping the Galaxy requires large area coverage, ex-
cellent image quality to maximize photometric and as-
trometric accuracy, especially in crowded fields, photo-
metric precision of at least 1% to separate main sequence
and giant stars (e.g., Helmi et al. 2003) as well as to iden-
tify variable stars such as RR Lyrae (Sesar et al. 2010;
Sharma et al. 2011), and astrometric precision of about
10 mas per observation to enable parallax and proper
motion measurements (SciBook Ch. 6,7). In order to
probe the halo out to its presumed edge at ∼ 100 kpc
(Ivezić et al. 2004) with main-sequence stars, the total
coadded depth must reach r > 27, with a similar depth
in the g band. The metallicity distribution of stars can
be studied photometrically in the Sgr tidal stream (e.g.,
see Majewski et al. 2003; Chou et al. 2007) and other
halo substructures (∼ 30 kpc, Carollo et al. 2007), yield- Figure 1. The image quality distribution measured at the
ing new insights into how they formed. Our ability to Cerro Pachón site using a differential image motion monitor
measure these metallicities is limited by the coadded (DIMM) at λ = 500 nm, and corrected using an outer scale
depth in the u band; to probe the outer parts of the parameter of 30 m over an 8.4 m aperture. For details about
stellar halo, one must reach u ∼ 24.5. To detect RR the outer scale correction see Tokovinin (2002). The observed
Lyrae stars beyond the Galaxy’s tidal radius at ∼ 300 distribution is well described by a log-normal distribution,
with the parameters shown in the figure.
kpc, the single-visit depth must be r ∼ 24.5.
In order to measure the tangential velocity of stars at
a distance of 10 kpc, where the halo dominates over the tailed specification of various probability density distri-
disk, to within 10 km s−1 (comparable with the accu- bution functions, please see the LSST Science Require-
racy of large-scale radial velocity surveys), the proper ments Document (Ivezić & The LSST Science Collabo-
motion accuracy should be 0.2 mas yr−1 or better. This ration 2011) and the LSST Science Book (LSST Science
is the same accuracy as will be delivered by the Gaia mis- Collaboration et al. 2009).
sion6 (Perryman et al. 2001; de Bruijne 2012) at its faint
1. The single visit depth should reach r ∼ 24.5. This
limit (r ∼ 20). In order to measure distances to solar
limit is primarily driven by the search for NEOs,
neighborhood stars out to a distance of 300 pc (the thin
variable sources (e.g., SN, RR Lyrae stars), and
disk scale height), with geometric distance accuracy of
by proper motion and trigonometric parallax mea-
at least 30%, trigonometric parallax measurements accu-
surements for stars. Indirectly, it is also driven
rate to 1 mas (1σ) are required over 10 years. To achieve
by the requirements on the coadded survey depth
the required proper motion and parallax accuracy with
and the minimum number of exposures required
an assumed astrometric accuracy of 10 mas per obser-
by WL science. We plan to split a single visit into
vation per coordinate, approximately 1,000 separate ob-
two exposures of equal length to identify and re-
servations are required. This requirement for a large
move cosmic rays.
number of observations is similar to that from minimiz-
ing systematics in weak lensing observations (§ 2.1.1). 2. Image quality should maintain the limit set by the
atmosphere (the median free-air seeing is 0.65 arc-
2.1.5. A Summary and Synthesis of Science-driven
sec in the r band at the chosen site, see Fig. 1), and
Constraints on Data Properties
not be degraded appreciably by the hardware. In
The goals of all the science programs discussed above addition to stringent constraints from weak lens-
(and many more, of course) can be accomplished by sat- ing, good image quality is driven by the required
isfying the minimal constraints listed below. For a more survey depth for point sources and by image dif-
elaborate listing of various constraints, including de- ferencing techniques.
3. Photometric repeatability should achieve 5 mmag
6 http://sci.esa.int/gaia/ precision at the bright end, with zeropoint stabil-
ity across the sky of 10 mmag and band-to-band 9. The coadded survey depth should reach r ∼ 27.5,
calibration errors not larger than 5 mmag. These with sufficient signal-to-noise ratio in other bands
requirements are driven by the need for high pho- to address both extragalactic and Galactic science
tometric redshift accuracy, the separation of stellar drivers.
populations, detection of low-amplitude variable
objects (such as eclipsing planetary systems), and 10. The distribution of visits per filter should en-
the search for systematic effects in type Ia super- able accurate photometric redshifts, separation of
nova light curves. stellar populations, and sufficient depth to en-
able detection of faint extremely red sources (e.g.,
4. Astrometric precision should maintain the limit brown dwarfs and high-redshift quasars). Detailed
set by the atmosphere, of about 10 mas per visit simulations of photometric redshift uncertainties
at the bright end (on scales below 20 arcmin). This suggest roughly similar number of visits among
precision is driven by the desire to achieve a proper bandpasses (but because the system throughput
motion accuracy of 0.2 mas yr−1 and parallax ac- and atmospheric properties are wavelength depen-
curacy of 1.0 mas over the course of a 10-year sur- dent, the achieved depths are different in different
vey (see § 3.2.3). bands). The adopted time allocation (see Table 1)
5. The single visit exposure time should be less than includes a slight preference to the r and i bands
about a minute to prevent trailing of fast moving because of their dominant role in star/galaxy sep-
objects and to aid control of various systematic aration and weak lensing measurements.
effects induced by the atmosphere. It should be
11. The distribution of visits on the sky should ex-
longer than ∼20 seconds to avoid significant effi-
tend over at least ∼18,000 deg2 to obtain the re-
ciency losses due to finite readout, slew time, and
quired number of galaxies for WL studies, with
read noise. As described above, we are planning
attention paid to include “special” regions such as
to split each visit into two exposures.
the Ecliptic and Galactic planes, and the Large
6. The filter complement should include at least six and Small Magellanic Clouds (if in the Southern
filters in the wavelength range limited by atmo- Hemisphere). For comparison, the full area that
spheric absorption and silicon detection efficiency can be observed at airmass less than 2.0 from any
(320–1050 nm), with roughly rectangular filters mid-latitude site is about 30,000 deg2 .
and no large gaps in the coverage, in order to
enable robust and accurate photometric redshifts 12. Data processing, data products and data access
and stellar typing. An SDSS-like u band (Fukugita should result in data products that approach the
et al. 1996) is extremely important for separating statistical uncertainties in the raw data; i.e., the
low-redshift quasars from hot stars, and for es- processing must be close to optimal. To enable
timating the metallicities of F/G main sequence fast and efficient response to transient sources, the
stars. A bandpass with an effective wavelength processing latency for variable sources should be
of about 1 micron would enable studies of sub- less than a minute, with a robust and accurate
stellar objects, high-redshift quasars (to redshifts preliminary characterization
of ∼7.5), and regions of the Galaxy that are ob- of all reported variables.
scured by interstellar dust.
Remarkably, even with these joint requirements, none
7. The revisit time distribution should enable deter-
of the individual science programs is severely over-
mination of orbits of Solar System objects and
designed, i.e., despite their significant scientific diver-
sample SN light curves every few days, while ac-
sity, these programs are highly compatible in terms of
commodating constraints set by proper motion
desired data characteristics. Indeed, any one of the four
and trigonometric parallax measurements.
main science drivers could be removed, and the remain-
8. The total number of visits of any given area of sky, ing three would still yield very similar requirements for
when accounting for all filters, should be of the most system parameters. As a result, the LSST system
order of 1,000, as mandated by WL science, the can adopt a highly efficient survey strategy in which
search for NEOs, and proper motion and trigono- a single dataset serves most science programs (instead
metric parallax measurements. Studies of tran- of science-specific surveys executed in series). One can
sient sources also benefit from a large number of view this project as massively parallel astrophysics. The
visits. vast majority (about 90%) of the observing time will be
Table 1. The LSST Baseline Design and Survey Parameters delivered image quality be dominated by atmospheric
seeing at the chosen site (Cerro Pachón in Northern
Quantity Baseline Design Specification Chile). A larger field-of-view would lead to unaccept-
Optical Config. 3-mirror modified Paul-Baker able deterioration of the image quality. This constraint
Mount Config. Alt-azimuth leaves the primary mirror diameter and survey lifetime
Final f-ratio, aperture f/1.234, 8.4 m as free parameters. The adopted survey lifetime of 10
Field of view, étendue 9.6 deg2 , 319 m2 deg2 years is a compromise between a shorter time that leads
Plate Scale 50.9 µm/arcsec (0.2” pix) to an excessively large and expensive mirror (15 m for
Pixel count 3.2 Gigapix a 3 year survey and 12 m for a 5 year survey) and not
Wavelength Coverage 320 – 1050 nm, ugrizy as effective proper motion measurements, and a smaller
Single visit depths, designa 23.9, 25.0, 24.7, 24.0, 23.3, 22.1 telescope that would require more time to complete the
Single visit depths, min.b 23.4, 24.6, 24.3, 23.6, 22.9, 21.7 survey, with the associated increase in operations cost.
Mean number of visitsc 56, 80, 184, 184, 160, 160
The primary mirror size is a function of the required
survey depth and the desired sky coverage. By and
Final (coadded) depthsd 26.1, 27.4, 27.5, 26.8, 26.1, 24.9
large, the anticipated science outcome scales with the
a Design specification from the Science Requirements Document
number of detected sources. For practically all astro-
(SRD; Ivezić & The LSST Science Collaboration 2011) for 5σ nomical source populations, in order to maximize the
depths for point sources in the ugrizy bands, respectively. The
number of detected sources, it is more advantageous to
listed values are expressed on the AB magnitude scale, and cor-
respond to point sources and fiducial zenith observations (about
maximize the area first, and then the detection depth7 .
0.2 mag loss of depth is expected for realistic airmass distribu- For this reason, the sky area for the main survey is max-
tions, see Table 2 for more details). imized to its practical limit, 18,000 deg2 , determined by
b Minimum specification from the Science Requirements Docu- the requirement to avoid airmasses less than 1.5, which
ment for 5σ depths.
would substantially deteriorate the image quality and
the survey depth (see eq. 6).
c An illustration of the distribution of the number of visits as a
With the adopted field-of-view area, the sky cover-
function of bandpass, taken from Table 24 in the SRD.
age and the survey lifetime fixed, the primary mirror
d Idealized depth of coadded images, based on design specification diameter is fully driven by the required survey depth.
for 5σ depth and the number of visits in the penultimate row There are two depth requirements: the final (coadded)
(taken from Table 24 in the SRD). survey depth, r ∼ 27.5, and the depth of a single visit,
r ∼ 24.5. The two requirements are compatible if the
number of visits is several hundred per band, which is
in good agreement with independent science-driven re-
devoted to a deep-wide-fast survey mode of the sort we quirements on the latter.
have just described, with the remaining 10% allocated The required coadded survey depth provides a direct
to special programs which will also address multiple sci- constraint, independent of the details of survey execu-
ence goals. Before describing these surveys in detail, we tion such as the exposure time per visit, on the minimum
discuss the main system parameters. effective primary mirror diameter of 6.4 m, as illustrated
in Fig. 2.
2.2. The Main System Design Parameters 2.2.2. The Optimal Exposure Time
Given the minimum science-driven constraints on the The single visit depth depends on both the primary
data properties listed in the previous section, we now mirror diameter and the chosen exposure time, tvis . In
discuss how they are translated into constraints on the turn, the exposure time determines the time interval
main system design parameters: the aperture size, the to revisit a given sky position and the total number of
survey lifetime, the optimal exposure time, and the filter
complement.
7 If the total exposure time is doubled and used to double the
2.2.1. The Aperture Size survey area, the number of sources increases by a factor of two.
If the survey area is kept fixed, the increased exposure time will
The product of the system’s étendue and the survey result in ∼0.4 mag deeper data (see eq. 6). For cumulative source
lifetime, for given observing conditions, determines the counts described by log(N ) = C + k ∗ m, the number of sources
will increase by more than a factor of two only if k > 0.75. Apart
sky area that can be surveyed to a given depth. The from z < 2 quasars, practically all populations have k at most 0.6
LSST field-of-view area is maximized to its practical (the Euclidean value), and faint stars and galaxies have k < 0.5.
limit, ∼10 deg2 , determined by the requirement that the For more details, please see Nemiroff (2003).
visits, and each of these quantities has its own science

drivers. We summarize these simultaneous constraints
in terms of the single-visit exposure time:
• The single-visit exposure time should not be longer

than about a minute to prevent trailing of fast So-
lar System moving objects, and to enable efficient
control of atmospheric systematics.
• The mean revisit time (assuming uniform cadence)

for a given position on the sky, n, scales as
10 deg2

tvis Asky
n= days,
10 sec 10, 000 deg2 AFOV
(1)
where two visits per night are assumed (required
for efficient detection of Solar System objects, see
below), and the losses for realistic observing condi-
tions have been taken into account (with the aid of
the Operations Simulator described below). Sci-
ence drivers such as supernova light curves and Figure 2. The coadded depth in the r band (AB magni-
moving objects in the Solar System require that tudes) vs. the effective aperture and the survey lifetime. It
n < 4 days, or equivalently tvis < 40 seconds for is assumed that 22% of the total observing time (corrected
for weather and other losses) is allocated for the r band, and
the nominal values of Asky and AF OV .
that the ratio of the surveyed sky area to the field-of-view
area is 2,000.
• The number of visits to a given position on the sky,
Nvisit , with losses for realistic observing conditions
taken into account, is given by Taking these constraints simultaneously into account,
as summarized in Fig. 3, yielded the following reference
design:

3000 T
Nvisit = . (2)
n 10 yr 1. A primary mirror effective diameter of ∼6.5 m.
The requirement Nvisit > 800 again implies that With the adopted optical design, described below,
n < 4 and tvis < 40 seconds if the survey lifetime, this effective diameter corresponds to a geometri-
T is about 10 years. cal diameter of ∼8 m. Motivated by characteristics
of the existing equipment at the Steward Mirror
• These three requirements place a firm upper limit Laboratory, which fabricated the primary mirror,
on the optimal visit exposure time of tvis < 40 the adopted geometrical diameter is set to 8.4 m.
seconds. Surveying efficiency (the ratio of open- 2. A visit exposure time of 30 seconds (using two 15
shutter time to the total time spent per visit) con- second exposures to efficiently reject cosmic rays;
siderations place a lower limit on tvis due to fi- the possibility of a single exposure per visit, to im-
nite detector read-out and telescope slew time (the prove observing efficiency, will be investigated dur-
longest acceptable read-out time is set to 2 sec- ing the commissioning phase), yielding = 77%.
onds, the shutter open-and-close time is 2 seconds,
and the slew and settle time is set to 5 seconds, in- 3. A revisit time of 3 days on average for 10,000 deg2
cluding the read-out time for the second exposure of sky, with two visits per night.
in a visit):
To summarize, the chosen primary mirror diameter is
tvis
= . (3) the minimum diameter that simultaneously satisfies the
tvis + 9 sec depth (r ∼ 24.5 for single visit and r ∼ 27.5 for coadded
To maintain efficiency losses below ∼30% (i.e., at depth) and cadence (revisit time of 3–4 days, with 30
least below the limit set by the weather patterns), seconds per visit) constraints described above.
and to minimize the read noise impact, tvis > 20
2.3. System Design Trade-offs
seconds is required.
Pan-STARRS survey (PS4). With an étendue about 6

times smaller than that of LSST (effective diameters of
6.4 m and 3.0 m, and a field-of-view area of 9.6 deg2
vs. 7.2 deg2 ), and all observing conditions being equal,
the PS4 system could in principle use a cadence identi-
cal to that of LSST. The main difference in the datasets
would be a faint limit shallower by about 1 mag in a
given survey lifetime. As a result, for Euclidean popu-
lations the sample sizes would go down by a factor of 4,
while for populations of objects with a shallower slope
of the number-magnitude relation (e.g., galaxies around
redshift of 1) the samples would be smaller by a fac-
tor 2–3. The distance limits for nearby sources, such
as Milky Way stars, would drop to 60% of their corre-
sponding LSST values, and the NEO completeness level
mandated by the U.S. Congress would not be reached.
If instead the survey coadded depth were to be main-
tained, then the survey sky area would have to be 6 times
smaller (∼3,500 deg2 ). If the survey single-visit depth
were to be maintained, then the exposure time would
Figure 3. The single-visit depth in the r band (5σ detection have to be about 6 times longer (ignoring the slight dif-
for point sources, AB magnitudes) vs. revisit time, n (days), ference in the field-of-view area and simply scaling by
as a function of the effective aperture size. With a cover- the étendue ratio), resulting in non-negligible trailing
age of 10,000 deg2 in two bands, the revisit time directly losses for Solar System objects, and either i) a factor
constrains the visit exposure time, tvis = 10 n seconds. In of six smaller sky area observed within n = 3 days,
addition to direct constraints on optimal exposure time, tvis or ii) the same sky area revisited every n = 18 days.
is also driven by requirements on the revisit time, n, the to-
Given these conflicts, one solution would be to split the
tal number of visits per sky position over the survey lifetime,
Nvisit , and the survey efficiency, (see eqs.1-3). Note that observing time and allocate it to individual specialized
these constraints result in a fairly narrow range of allowed programs (e.g., large sky area vs. deep coadded data vs.
tvis for the main deep-wide-fast survey. deep single-visit data vs. small n data, etc.), as is being
done by the PS1 Consortium9 .
We note that the Pan-STARRS project (Kaiser et al. In summary, given the science requirements as stated
2002, 2010), with similar science goals as LSST, envi- here, there is a minimum étendue of ∼300 deg2 m2 which
sions a distributed aperture design, where the total sys- enables our seemingly disparate science goals to be ad-
tem étendue is a sum of étendue values for an array dressed with a single dataset. A system with a smaller
of small 1.8 m telescopes8 . Similarly, the LSST system étendue would require separate specialized surveys to
could perhaps be made as two smaller copies with 6m address the science goals, which results in a loss of sur-
mirrors, or 4 copies with 4m mirrors, or 16 copies with veying efficiency10 . The LSST is designed to reach this
2m mirrors. Each of these clones would have to have minimum étendue for the science goals stated in its Sci-
its own 3 Gigapixel camera (see below), and given the ence Requirements Document.
added risk and complexity (e.g., maintenance, data pro-
cessing), the monolithic design seems advantageous for 2.4. The Filter Complement
a system with such a large étendue as LSST. The LSST filter complement (ugrizy, see Fig. 4) is
It is informative to consider the tradeoffs that would modeled after the Sloan Digital Sky Survey (SDSS)
be required for a system with a smaller aperture, if the system (Fukugita et al. 1996) because of its demon-
science requirements were to be maintained. For this strated success in a wide variety of applications, includ-
comparison, we consider a four-telescope version of the
9 More information about Pan-STARRS is available from http:

8 The first of these telescopes, PS1, has been operational for //pswww.ifa.hawaii.edu/pswww/.
some time (Chambers et al. 2016), and has an étendue 1/24th 10 The converse is also true: for every étendue there is a set of
that of LSST. optimal science goals that such a system can address with a high
efficiency.
1.0 5.5
x1e-14
Airmass 1.2 5.0
0.8 u g r z y
i
Flux (ergs/cm2/s/Angstrom)
4.5
4.0
Throughput (0-1)
0.6
3.5
3.0
0.4
2.5
0.2 2.0
1.5
0.0300 400 500 600 700 800 900 1000 1100 1.0
5000 6000 7000 8000 9000 10000 11000
Wavelength (nm) Wavelength (Angstrom)
Figure 4. The LSST bandpasses. The vertical axis shows Figure 5. An example of determination of the atmospheric
the total throughput. The computation includes the atmo- opacity by simultaneously fitting a three-parameter stellar
spheric transmission (assuming an airmass of 1.2, dotted model SED (Kurucz 1979) and six physical parameters of
line), optics, and the detector sensitivity. a sophisticated atmospheric model (MODTRAN, Anderson
et al. 1999) to an observed F-type stellar spectrum (Fλ ). The
black line is the observed spectrum and the red line is the
ing photometric redshifts of galaxies (Budavári et al.
best fit. Note that the atmospheric water feature around
2003), separation of stellar populations (Lenz et al. 1998; 0.9–1.0 µm is exquisitely well fit. The components of the
Helmi et al. 2003), and photometric selection of quasars best-fit atmospheric opacity are shown in Fig. 6. Adapted
(Richards et al. 2002; Ross et al. 2012). The extension of from Burke et al. (2010).
the SDSS system to longer wavelengths (the y band at
∼1 micron) is driven by the increased effective redshift
range achievable with the LSST due to deeper imag-
ing, the desire to study sub-stellar objects, high-redshift
quasars, and regions of the Galaxy that are obscured by
interstellar dust, and the scientific opportunity enabled
by modern CCDs with high quantum efficiency in the
near infrared.
The chosen filter complement corresponds to a design
“sweet spot”. We have investigated the possibility of
replacing the ugrizy system with a filter complement
that includes only five filters. For example, each filter
width could be increased by 20% over the same wave-
length range (neither a shorter wavelength range, nor
gaps in the wavelength coverage are desirable options),
but this option is not satisfactory. Placing the red edge
of the u band blueward of the Balmer break allows op-
timal separation of stars and quasars, and the telluric
water absorption feature at 9500 Å effectively defines the
Figure 6. The components of the best-fit atmospheric opac-
blue edge of the y band. Of the remaining four filters
ity used to model the observed stellar spectrum shown in
(griz), the g band is already quite wide. As a last op- Fig. 5. The atmosphere model (MODTRAN, Anderson et al.
tion, the riz bands could be redesigned as two wider 1999) includes six components: water vapor (blue), oxy-
bands. However, this option is also undesirable because gen and other trace molecules (green), ozone (red), Rayleigh
the r and i bands are the primary bands for weak lensing scattering (cyan), a gray term with a transmission of 0.989
studies and for star/galaxy separation, and chromatic (not shown) and an aerosol contribution proportional to λ−1
atmospheric refraction would worsen the point spread and extinction of 1.3% at λ=0.675 µm (not shown). The
black line shows all six components combined. Adapted from
function for a wider bandpass.
Burke et al. (2010).
2.5. The Calibration Methods
Precise determination of the point spread function bration from the task of assigning absolute optical flux
across each image, accurate photometric and astromet- to celestial objects.
ric calibration, and continuous monitoring of system Celestial sources will be used to refine the internal
performance and observing conditions will be needed to photometric system and to monitor stability and uni-
reach the full potential of the LSST mission. Extensive formity of the photometric data. We expect to use Gaia
precursor data including the SDSS dataset and our own Collaboration et al. (2016) photometry, utilising the BP
data obtained using telescopes close to the LSST site of and RP photometric measurements as well as the G
Cerro Pachón (e.g., the SOAR and Gemini South tele- magnitudes; for a subset of stars (e.g. F-subdwarfs)
scopes), as well as telescopes of similar aperture (e.g., we expect to be able to transfer this rigid photomet-
Subaru), indicate that the photometric and astrometric ric system above the atmosphere to objects observed
accuracy will be limited not by our instrumentation or by LSST. There will be >100 main-sequence stars with
software, but rather by atmospheric effects. 17 < r < 20 per detector (14×14 arcmin2 ) even at
The overall photometric calibration philosophy (Stubbs high Galactic latitudes. Standardization of photomet-
& Tonry 2006) is to measure explicitly, at 1 nm resolu- ric scales will be achieved through direct observation of
tion, the instrumental sensitivity as a function of wave- stars with well-understood spectral energy distributions
length using light from a monochromatic source injected (SEDs), in conjunction with the in-dome calibration sys-
into the telescope pupil. The dose of delivered photons tem and the atmospheric transmission spectra.
is measured using a calibration photodiode whose quan- Astrometric calibration will be based on the results
tum efficiency is known to high accuracy. In addition, from the Gaia mission (Gaia Collaboration et al. 2016),
the LSST system will explicitly measure the atmospheric which will provide numerous high-accuracy astrometric
transmission spectrum associated with each image ac- standards in every LSST field.
quired. A dedicated 1.2-meter auxiliary calibration
telescope will obtain spectra of standard stars in LSST 2.6. The LSST Reference Design
fields, calibrating the atmospheric throughput as a func- We briefly describe the reference design for the main
tion of wavelength (Stubbs et al. 2007, see Figs. 5 and LSST system components. Detailed discussion of the
6). The LSST auxiliary telescope will take data at lower flow-down from science requirements to system design
spectral resolution (R ∼ 150) but wider spectral cov- parameters, and extensive system engineering analysis
erage (340nm — 1.05µm) than shown in these figures, can be found in the LSST Science Book (Ch. 2–3).
using a slitless spectrograph and an LSST corner-raft
CCD. Celestial spectrophotometric standard stars can 2.6.1. Telescope and Site
be used as a separate means of photometric calibration,
The large LSST étendue is achieved in a novel three-
albeit only through the comparison of band-integrated
mirror design (modified Paul-Baker Mersenne-Schmidt
fluxes with synthetic photometry calculations.
system; Angel et al. 2000) with a very fast f /1.234 beam.
A similar calibration process has been undertaken by
The optical design has been optimized to yield a large
the Dark Energy Survey (DES) team, which has been
field of view (9.6 deg2 ), with seeing-limited image qual-
approaching a calibration precision of 5 mmag (Burke
ity, across a wide wavelength band (320–1050 nm). In-
et al. 2018).
cident light is collected by an annular primary mirror,
SDSS, PS1, and DES data taken in good photomet-
having an outer diameter of 8.4 m and inner diameter
ric conditions have approached the LSST requirement of
of 5.0 m, creating an effective filled aperture of ∼6.4 m
1% photometric calibration (Padmanabhan et al. 2008;
in diameter once vignetting is taken into account. The
Schlafly et al. 2012; Burke et al. 2018), although mea-
collected light is reflected to a 3.4 m convex secondary,
surements with ground-based telescopes typically pro-
then onto a 5 m concave tertiary, and finally into the
duce data with errors a factor of two or so larger. Anal-
three refractive lenses of the camera (see Fig. 7). In
ysis of repeated SDSS scans obtained in varying ob-
broad terms, the primary-secondary mirror pair acts as
serving conditions demonstrates that data obtained in
a beam condenser, while the aspheric portion of the sec-
non-photometric conditions can also be calibrated with
ondary and tertiary mirror acts as a Schmidt camera.
sufficient accuracy (Ivezić et al. 2007b), as long as high-
The three-element refractive optics of the camera cor-
quality photometric data also exist in the region. The
rect for the chromatic aberrations induced by the ne-
LSST calibration plan builds on this experience gained
cessity of a thick dewar window and flatten the focal
from the SDSS and other surveys.
surface. During design optimization, the primary and
The planned calibration process decouples the estab-
tertiary mirror surfaces were placed such that the pri-
lishment of a stable and uniform internal relative cali-
mary’s inner diameter coincides with the tertiary’s outer
Figure 9. The baseline design for the LSST telescope. The

small focal ratio allows for a very squat telescope, and thus
a very stiff structure.
ror was cast and polished by the Richard F. Caris Mirror

Lab at the University of Arizona in Tucson before being
inspected and accepted by LSST in April 2015 (Araujo-
Figure 7. The LSST baseline optical design (modified three- Hauck et al. 2016). The primary-tertiary mirror cell was
mirror Paul-Baker) with its unique monolithic mirror: the fabricated by CAID in Tucson and is undergoing accep-
primary and tertiary mirrors are positioned such that they tance tests. The integration of the actuators and final
form a continuous compound surface, allowing them to be
tests with the mirror is scheduled for early 2018.
polished from a single substrate.
The LSST Observing Facility (Fig. 10), consisting of
the telescope enclosure and summit support building, is
being constructed atop Cerro Pachón in northern Chile,
sharing the ridge with the Gemini South and SOAR tele-
scopes11 (latitude: S 30◦ 140 40.6800 ; longitude: W 70◦
440 57.9000 ; elevation: 2652 m; Mamajek 2012). The tele-
scope enclosure houses a compact, stiff telescope struc-
ture (see Fig. 9) atop a 15 m high concrete pier with a
fundamental frequency of 8 Hz, that is crucial for achiev-
ing the required fast slew-and-settle times. The height of
the pier was set to place the telescope above the degrad-
ing effects of the turbulent ground layer. Capping the
telescope enclosure is a 30 m diameter dome with exten-
sive ventilation to reduce dome seeing and to maintain
a uniform thermal environment over the course of the
Figure 8. The polishing of the primary-tertiary mirror pair
night. Furthermore, the summit support building has
at the Richard F. Caris Mirror Lab at the University of Ari-
zona Tucson. been oriented with respect to the prevailing winds to
shed its turbulence away from the telescope enclosure.
diameter, thus making it possible to fabricate the mir- The summit support building includes a coating cham-
ror pair from a single monolithic blank using spin-cast ber for recoating the three LSST mirrors and clean room
borosilicate technology. The secondary mirror is fab- facilities for maintaining and servicing the camera.
ricated from a thin 100 mm thick meniscus substrate, 2.6.2. Camera
made from Corning’s ultra-low expansion material. All
three mirrors will be actively supported to control wave-
front distortions introduced by gravity and environmen- 11 Coordinates listed in older versions of this paper were incor-
tal stresses on the telescope. The primary-tertiary mir- rect. We thank E. Mamajek for pointing out this error to us.
Figure 11. A cutaway view of LSST camera. Not shown

are the shutter, which is positioned between the filter and
lens L3, and the filter exchange system.
in an optics structure at the front of the camera body,

which also contains a mechanical shutter, and a carousel
assembly that holds five large optical filters. The sixth
optical filter can replace any of the five via a procedure
accomplished during daylight hours.
Each of the 21 rafts will host 3 front end electronic
boards (REB) operating in the cryostat (at −10◦ C),
that read in parallel a total of 9×16 segments per CCD
Figure 10. Top: artist’s rendering of the dome enclosure
(144 video channels reading one million pixels each).
with the attached summit support building on Cerro Pachón. This very high parallelization is the key to allow for a
The LSST auxiliary calibration telescope is shown on an ad- fast readout (2 seconds) of the entire focal plane. To
jacent rise to the right. Bottom: Photograph of the LSST reach this performance with a reasonably-sized board,
Observatory as of Summer 2017. Note the different perspec- a special low-noise (<3 electrons), low-crosstalk be-
tive from the artist’s rendering. The main LSST telescope tween channels (<0.02%) and low-power dissipation (25
building is on the right, waiting for the dome to be installed. mW/channel) Analog Signal Processing Integrated Cir-
The auxiliary telescope building is on the left with its dome
cuit (ASPIC), hosting 8 channels per chip, has been de-
being installed.
veloped, which is able to read the CCDs with a linearity
better than 0.1% (Antilogus et al. 2017).
The LSST camera provides a 3.2 Gigapixel flat focal
plane array, tiled by 189 4K×4K CCD science sensors 2.6.3. Data Management
with 10 µm pixels (see Figs. 11 and 12). This pixel count The rapid cadence and scale of the LSST observing
is a direct consequence of sampling the 9.6 deg2 field- program will produce approximately 15 TB per night
of-view (0.64 m diameter) with 0.2×0.2 arcsec2 pixels of raw imaging data12 (about 20 TB with calibration
(Nyquist sampling in the best expected seeing of ∼0.4 exposures). As with all large modern surveys, the large
arcsec). The sensors are deep depleted high resistivity data volume, the real-time aspects, and the complexity
silicon back-illuminated devices with a highly segmented of processing involved requires that the survey itself
architecture that enables the entire array to be read in 2 take on the task of fully reducing the data. The data
seconds. The detectors are grouped into 3×3 rafts (see collected by the LSST system will be automatically re-
Fig. 13); each contains its own dedicated electronics. duced to scientifically useful catalogs and images by the
The rafts are mounted on a silicon carbide grid inside a LSST Data Management (DM; Jurić et al. 2015) system.
vacuum cryostat, with a custom thermal control system
that maintains the CCDs at an operating temperature
of around 173 K. The entrance window to the cryostat
12 For comparison, the volume of all imaging data collected over
is the third (L3) of the three refractive lenses in the
a decade by the SDSS-I/II projects and published in SDSS Data
camera. The other two lenses (L1 and L2) are mounted Release 7 (Abazajian et al. 2009) is approximately 16 TB.
The detailed outputs of the LSST Data Management

system are described in § 3.3. The principal functions
of the system are to:
• Process, in real time, the incoming stream of
images generated by the camera system during
observing by archiving raw images, generating
alerts to new sources or sources whose properties
have changed, and updating the relevant catalogs
(Prompt products; § 3.3).
• Process each night’s data during the day and de-

termine or refine orbits for all asteroids found in
the imaging.
• Periodically process the accumulated survey data

to provide a uniform photometric and astrometric
calibration, measure the properties of all detected
objects, and characterize objects based on their
time-dependent behavior. The results of such a
processing run form a Data Release (DR), which
is a static, self-consistent dataset suitable for use
in performing scientific analyses of LSST data and
Figure 12. The LSST Camera focal plane array. Each cyan publication of the results (the data release prod-
square represents one 4K× 4K pixel sensor. Nine sensors are
ucts; § 3.3). We are planning two data releases
assembled into a raft; the 21 rafts are outlined in red. There
are 189 science sensors, for a total of 3.2 gigapixels. Also
covering the first year of full operations, and an-
shown are the four corner rafts, where the guide sensors and nual data releases thereafter.
wavefront sensors are located.
• Facilitate the creation of data products gener-
ated by the science community, by providing suit-
able software, application programming interfaces
(APIs), and computing infrastructure at the LSST
data access centers.
• Make all LSST data available through an inter-

face that utilizes community-based standards to
the maximum possible extent. Provide enough
processing, storage, and network bandwidth to en-
able user analyses of the data without the need for
petabyte-scale data transfers.
Over the ten years of LSST operations and 11 data
releases, this processing will result in a cumulative pro-
cessed data size approaching 500 petabytes (PB) for
imaging, and over 50 PB for the catalog databases. The
final data release catalog database alone is expected to
be approximately 15 PB in size.
The DM system will span four key facilities on three

continents: the Summit Facility on Cerro Pachón in
Figure 13. The LSST Camera raft module, corresponding Chile (where the initial detector cross-talk correction
to the red squares in Fig. 12, with 9 sensors, integrated elec- will be performed); the Base Facility in La Serena, Chile
tronics, and thermal connections. Raft modules are designed (which will serve as a retransmission for data uploads
to be replaceable. to North America, as well as the Data Access Center
for the Chilean community); the Data Processing and
Figure 14. The LSST data flow from the mountain facilities in Chile to the data access center and processing center in the
U.S., and the satellite processing center in France.
Archiving Facility at the National Center for Supercom- necessary data access and orchestration middleware, as
puting Applications (NCSA) in Champaign-Urbana, IL; well as the database and user interface components.
and the Satellite Processing Facility at CC-IN2P3 in Algorithm development for the LSST software builds
Lyon, France. All real-time data processing and half on the expertise and experience of prior large astronom-
the data release product processing will take place at ical surveys (including SDSS, Pan-STARRS, DES, Su-
the Data Processing and Archiving Facility, which will perMACHO, ESSENCE, DLS, CFHTLS, and UKIDSS).
also serve as the Data Access Center for the US commu- The pipelines written for these surveys have demon-
nity. The other half of the data release processing will strated that it is possible to carry out largely au-
be done at CC-IN2P3, which will also have the role of tonomous data reduction of large datasets, automated
“Long-term Storage” facility. detection of sources and objects, and the extraction
The data will be transported between the centers over of scientifically useful characteristics of those objects.
existing and new high-speed optical fiber links from While firmly footed in this prior history, the LSST soft-
South America to the U.S. (see Fig. 14). The data pro- ware stack has largely been written anew, for reasons
cessing center demands stable, well-tested technology to of performance, extendability, and maintainability. All
ensure smooth operations. Hence, while LSST is mak- LSST codes have been designed and implemented follow-
ing a novel use of advances in information technology, ing software engineering best practices, including mod-
it is not pushing the expected technology to the limit, ularity, clear definition of interfaces, continuous integra-
reducing the overall risk to the project. tion, utilization of unit testing, and a single set of doc-
umentation and coding standards (Jenness et al. 2018).
2.6.4. The LSST software stack The primary implementation language is Python and,
where necessary for performance reasons, C++13 .
The LSST Software Stack is the data processing and
The LSST data management software has been proto-
analysis system developed by the LSST Project to en-
typed for over eight years. Besides processing simulated
able LSST survey data reduction and delivery. It com-
prises all science pipelines needed to accomplish LSST
data processing tasks (e.g., calibration, single frame 13 All components implemented in C++ have been wrapped and
processing, coaddition, image differencing, multi-epoch exposed as Python modules to the rest of the system. Typical
users should not have to work directly with the C++ layer.
measurement, asteroid orbit determination, etc.), the
scientific algorithms, public. Our primary goals in pub-

licizing the code are to simplify reproducibility of LSST
data products and to provide insight into algorithms
used to create them. Achieving these goals requires that
the source code is not only available, but appropriately
documented at all levels. Given that, most of the LSST
software stack is licensed under the terms of the GNU
General Public License (GPL), Version 3, and can be
found at https://github.com/lsst. The documentation
for the LSST Science Pipelines components of the stack
is available at https://pipelines.lsst.io.
The LSST Software Stack may be of interest and
(re)used beyond the LSST project (e.g., by other survey
projects, or by individual LSST end-users). Enabling or
supporting such applications goes beyond LSSTs con-
struction requirements; however, when developing the
LSST codes we strongly prefer design choices that en-
able future generalization. As an example of such re-use,
a pipeline derived from the present-day LSST software
Figure 15. A small region in the vicinity of globular stack prototypes has been used to reduce data taken
cluster M2, taken from a coadd of SDSS Stripe 82 data pro- with the HSC camera (Miyazaki et al. 2018) on Subaru
duced with LSST software stack prototypes. The coaddi- as part of the large SSP survey (http://hsc.mtk.nao.ac.
tion employs a novel “background matching” technique that jp/ssp/survey; Aihara et al. (2018); Bosch et al. (2018);
improves background estimation and preserves the diffuse
see Fig. 16).
structures in the resulting coadd.
2.6.5. The LSST database design: Qserv
The scale of the LSST data release catalogs, in com-
bination with desired targets for user concurrency and
query response times, present some engineering chal-
lenges. The LSST project has been developing Qserv,
a shared-nothing MPP (massively parallel processing)
database system, to meet these needs (Wang et al. 2011;
Becla & Wang 2014). Catalog data within Qserv is spa-
tially partitioned, and hosted on shard servers running
on dedicated hardware resources within the LSST Data
Facility. The shard servers locally leverage conventional
RDBMS (relational database management system) tech-
nologies, running behind custom front-end codes which
Figure 16. A small portion, 40 ×60 , of the HSC gri imaging
handle query analysis, rewrite, distribution, and result
of the COSMOS field. The limiting magnitude is about 27.5, aggregation. The Qserv shard servers also provide a fa-
roughly equivalent to 10-year LSST depth. cility for cross-user synchronization of full-table scans in
order to provide predictable query response times when
LSST data (§ 2.7.3), it has been used to process im- serving many users concurrently. More details about
ages from CFHTLS (Cuillandre et al. 2012) and SDSS Qserv can be found in the LSST document LDM-135
(Abazajian et al. 2009). As an example, Fig. 15 shows (Becla et al. 2017).
a small region in the vicinity of M2 taken from a large
coaddition of SDSS Stripe 82 data, generated with LSST 2.7. Simulating the LSST System
software stack prototypes (Jurić et al. 2013). Throughout its design, construction and commission-
ing, the LSST needs to be able to demonstrate that it
Other than when prohibited by licensing, security, or can achieve the requirements laid out in the Science Re-
other similar considerations, the LSST makes all newly quirements Documents (SRD) given its design and as-
developed source code, and especially that pertaining to delivered components, that the system can be calibrated
to the required level of fidelity, that the data manage- the correlation between, for example, seasonal weather
ment software can extract the appropriate astrophysical patterns and observing conditions at any given point on
signals, and that this can be achieved with sufficient effi- the sky. In addition, down time for observatory mainte-
ciency such that the telescope can complete its primary nance is also included.
objectives within a ten-year survey. The signal-to-noise ratio of each observation is deter-
Realizing these objectives requires that the project mined using a sky background model which includes the
can characterize the performance of the LSST includ- dark sky brightness in each filter, the effects of seeing
ing the performance of the opto-mechanical systems, and atmospheric transparency, and a detailed model for
the response of the detectors and their electronics, and scattered light from the Moon and/or twilight at each
the capabilities of the analysis software. A simulation observation (Yoachim et al. 2016). The time taken to
framework provides such a capability; delivering a vir- move from one observation to the next is given by a de-
tual prototype LSST against which design decisions, op- tailed model of the camera, telescope, and dome. It in-
timizations (including descoping), and trade studies can cludes such effects as the acceleration/deceleration pro-
be evaluated (Connolly et al. 2014). files employed in moving the telescope, the dome, and
The framework underlying the LSST simulations is the wind screen, the time needed to damp vibrations ex-
designed to be extensible and scalable (i.e., capable of cited by each slew, cable wrap, the time taken for active
being run on a single processor or across many-thousand optics lock and correction as a function of slew distance,
core compute clusters). It comprises four primary com- and the time for filter changes and focal plane readout.
ponents: a simulation of the survey scheduler (§ 2.7.1), Observations are scheduled by a ranking algorithm.
databases of simulated astrophysical catalogs of stars, After a given exposure, all possible next observations
galaxies, quasars and Solar System objects (§ 2.7.2), a are assigned a score which depends upon their locations,
system for generating observations based on the point- times, and filters according to a set of scientific require-
ing of the telescope, and a system for generating realistic ments which can vary with time and location. For ex-
LSST images of a given area of sky (§ 2.7.3). Compu- ample, if an ecliptic field has been observed in the r
tationally intensive routines are written in C/C++ with band, the score for another r-band observation of the
the overall framework and database interactions using same field will initially be quite low, but it will rise in
P ython. The purpose of this design is to enable the time to peak about an hour after the first observation,
generation of a wide range of data products for use by and decline thereafter. This algorithm results in obser-
the collaboration; from all-sky catalogs used in simula- vations being acquired as pairs roughly an hour apart,
tions of the LSST calibration pipeline, to studies of the which enables efficient association of NEO detections.
impact of survey cadence on recovering variability, to To ensure uniform sky coverage, fields with fewer previ-
simulated images of a single LSST focal plane. ous observations will be scored more highly than those
which have already been observed more frequently.
2.7.1. The LSST Operations Simulator Once all possible next observations have been scored
for scientific priority, their scores are modified according
The LSST Operations Simulator (Delgado et al. 2014)
to observing conditions (e.g., seeing, airmass, and sky
was developed to enable a detailed quantitative analysis
brightness) and to criteria such as slew time to move
of the various science tradeoffs described in this paper.
from the current position, time required to change fil-
It contains detailed models of site conditions, hardware
ters, etc. The highest-ranked observation is then per-
and software performance, and an algorithm for schedul-
formed, and the cycle repeats. The result of a simulator
ing observations which will, eventually, drive the largely
run is a detailed history of which locations on the sky
robotic observatory. Observing conditions include a
were observed when, in what filter, and with what sky
model for seeing derived from an extensive body of on-
background, seeing and other observing conditions. It
site MASS/DIMM (Multi-Aperture Scintillation Sensor
takes a few days to produce a decade-long simulation
and Differential Image Motion Monitor) measurements
using an average PC.
obtained during site selection and characterization (see
Results of the simulated surveys can be visualized
Fig. 1). It not only reproduces the observed seeing dis-
and analyzed using a Python-based package called the
tribution, but includes the auto-correlation spectrum of
Metrics Analysis Framework (MAF; Jones et al. 2014).
seeing with time over intervals from minutes to seasons.
MAF provides tools to analyze the properties of a survey
Weather data are taken from ten years of hourly mea-
(e.g. the distribution of airmasses) through the creation
surements at nearby Cerro Tololo. Thus the simulator
of functions or metrics that are applied to OpSim out-
correctly represents the variation of limiting magnitude
puts. These metrics can express the expected technical
between pairs of observations used to detect NEOs and
performance of the survey, such as the number of visits assigned to each star using the three-dimensional Galac-
per field or the integrated depth after 10 years, as well tic model of Amôres & Lépine (2005). To provide con-
as the science capabilities or a survey, such as the num- sistency with the modeling of extragalactic fluxes in the
ber of supernovae detected or the number of supernovae simulations, the dust model in the Milky Way integrated
with sufficient observations to have a well-characterized to 100 kpc is re-normalized to match the Schlegel et al.
light curve. (1998) dust maps.
Galaxy catalogs are derived from the Millennium sim-
2.7.2. Catalog Generation ulations of De Lucia et al. (2006). These models ex-
tend pure dark matter N-body simulations to include
The simulated astronomical catalogs (CatSim; Con-
gas cooling, star formation, supernovae and AGN, and
nolly et al. 2014) are stored in an SQL database. This
are designed to reproduce the observed colors, luminosi-
base catalog is queried using sequences of observations
ties, and clustering of galaxies as a function of redshift.
derived from the Operations Simulator. Each simulated
To generate the LSST simulated catalogs, a light cone,
pointing provides a position and time of the observa-
covering redshifts 0 < z < 6, was constructed from 58
tion together with the appropriate sky conditions (e.g.,
simulation snapshots 500 h−1 Mpc on a side. This light
seeing, moon phase and angle, sky brightness and sky
cone extends to a depth of approximately r = 28 and
transparency). Positions of sources are propagated to
covers a 4.5◦ ×4.5◦ footprint on the sky. Replicating this
the time of observation (including proper motions for
catalog across the sky simulates the full LSST footprint.
stars and orbits for Solar System sources). Magnitudes
As with the stellar catalog, an SED is fit to the colors
and source counts are derived using the atmospheric and
of each source using Bruzual & Charlot (2003) spectral
filter response functions appropriate for the airmass of
synthesis models. These fits are undertaken separately
the observation and after applying corrections for source
for the bulge and disk components and, for the disk,
variability. The resulting catalogs are then formatted to
include inclination-dependent reddening. Morphologies
be output to users, or to be fed into an image simulator.
are modeled using two Sérsic profiles. The bulge-to-disk
The current version of the LSST simulation framework
ratio and disk scale lengths are taken from De Lucia
incorporates galaxies derived from an N-body simulation
et al. (2006). Half-light radii for bulges are estimated
of a ΛCDM cosmology, quasars/AGNs, stars that match
using the empirical absolute-magnitude vs. half-light
the observed stellar distributions within our Galaxy, as-
radius relation given by González et al. (2009). Compar-
teroids generated from simulations of our Solar Sys-
isons between the redshift and number-magnitude dis-
tem, and a 3-D model for Galactic extinction. Stel-
tributions of the simulated catalogs with those derived
lar sources are based on the Galactic structure models
from deep imaging and spectroscopic surveys showed
of Jurić et al. (2008) and include thin-disk, thick-disk,
that the De Lucia et al. (2006) models under-predict
and halo star components. The distribution and col-
the density of sources at faint magnitudes and high red-
ors of the stars match those observed by SDSS. Each
shifts. To correct for these effects, sources are cloned in
star in the simulation is matched to a template spectral
magnitude and redshift space until their densities reflect
energy distribution (SED). Kurucz (1993) model spec-
the average observed properties.
tra are used to represent main-sequence F, G, and K
Quasar/AGN catalogs are generated using the Bon-
stars as well as RGB stars, blue horizontal branch stars,
giorno et al. (2007) luminosity function for MB < −15.
and RR Lyrae variables. SEDs for white dwarf stars
Their observed SEDs are generated using a composite
are taken from Bergeron et al. (1995). SEDs for M,
rest-frame spectrum derived from SDSS data by Vanden
L, and T dwarfs are generated from a combination of
Berk et al. (2001). The host galaxy is selected to have
spectral models and stacks of spectra from the SDSS
the closest match to the preferred stellar mass and color
(e.g., Cushing et al. 2005; Bochanski et al. 2007; Burrows
at the AGN’s redshift, following the results from Xue
et al. 2006; Pettersen & Hawley 1989; Kowalski et al.
et al. (2010). Each galaxy hosts at most one AGN, and
2010). The adopted metallicity for each star is based on
no explicit distinction is made between low-luminosity
a model from Ivezić et al. (2008), and proper motions
AGN and quasars that dramatically outshine their host
are based on the kinematic model of Bond et al. (2010).
galaxies. The light curve for each AGN is generated
Light curve templates are assigned to a subset of the
using a damped random walk model and prescriptions
stellar population so that variability may also be simu-
given by MacLeod et al. (2010).
lated. This assignment and variability are matched to
Asteroids are simulated using the Solar System mod-
variability trends observed by the Kepler satellite, and
els of Grav et al. (2007). They include: Near Earth Ob-
augmented by simulated distributions of RR-Lyrae and
jects (NEOs), Main Belt Asteroids, the Trojans of Mars,
Cepheids. For Galactic reddening, a value of E(B−V ) is
Jupiter, Saturn, Uranus, and Neptune, Trans Neptunian niques in a fast ray-tracing algorithm and all optical
Objects, and Centaurs. Spectral energy distributions surfaces include a spectrum of perturbations based on
are assigned using the C and S type asteroids of DeMeo design tolerances. Each optic moves according to its
et al. (2009). Positions for the 11 million asteroids in the six degrees of freedom within tolerances specified by
simulation are stored within the base catalog (sampled the LSST system. Fast techniques for finding intercepts
once per night for the ten year duration of the LSST sur- on the aspheric surface and altering the trajectory of
vey). We generate accurate ephemerides of all asteroids a photon by reflection or wavelength-dependent refrac-
falling within a given LSST point using the OpenOrb tion have been implemented to optimize the efficiency of
software package (Granvik et al. 2009). With typically the simulated images. Wavelength and angle-dependent
8000 sources per LSST field of view, this procedure sig- transmission functions are incorporated within each of
nificantly reduces the computational resources required these techniques, including simulation of the telescope
to simulate asteroid ephemerides. spider.
Both GalSim and PhoSim model the propagation of
2.7.3. Image Simulations photons through the silicon of the detector. The conver-
sion probability, refraction as a function of wavelength
The framework described above provides a parametrized
and temperature, and charge diffusion within the sili-
view of the sky above the atmosphere. Images are sim-
con are modeled for all photons. Photons are pixelated
ulated using two packages: GalSim (Rowe et al. 2015),
and the readout process simulated including blooming,
and Phosim (Peterson et al. 2015). Galsim is a modu-
charge saturation, charge transfer inefficiency, gain and
lar and open-source package that provides a library for
offsets, hot pixels and columns, the dependence of the
simulating stars and galaxies through a range of modern
image size on intensity (a.k.a. the “brighter-fatter” ef-
astronomical telescopes. Point-spread-functions (PSFs)
fect), and QE variations.
are treated as either analytic functions or modeled from
An example of a simulated LSST image using PhoSim
ray-traced optics. Convolutions by the PSF can be
is shown in Fig. 17.
applied to parameterized galaxy profiles (e.g. Sérsic
profiles) or to directly observed images. Operations are 3. ANTICIPATED DATA PRODUCTS AND THEIR
applied in Fourier space to enable an effective trade-off CHARACTERISTICS
between speed of simulation and accuracy. GalSim is
written in C++ with a Python API and is integrated The LSST observing strategy is designed to maximize
within the LSST CatSim framework. the scientific throughput by minimizing slew and other
Phosim is an open-source package that simulates im- downtime and by making appropriate choices of the fil-
ages by drawing photons from the spectral energy dis- ter bands given the real-time weather conditions. Using
tribution of each source (scaled to the appropriate flux simulated surveys produced with the Operations Sim-
density based on the apparent magnitude of a source and ulator described in § 2.7.1, we illustrate predictions of
accounting for the spatial distribution of light for ex- LSST performance with two examples.
tended sources). Each photon is ray-traced through the
atmosphere, telescope and camera to generate a CCD 3.1. The Baseline LSST Surveys
image. The atmosphere is modeled using a Taylor frozen The fundamental basis of the LSST concept is to scan
screen approximation (with the atmosphere described the sky deep, wide, and fast, and to obtain a dataset that
by six layers). The density fluctuations within these simultaneously satisfies the majority of the science goals.
screens are described by a Kolmogorov spectrum with We present here a specific realization, the so-called “uni-
an outer scale (typically 10 m to 200 m). All screens versal cadence”, which yields the main deep-wide-fast
move during an exposure, with velocities derived from survey and meets our core science goals. However, at
NOAA measurements of the wind velocities above the this writing, there is a vigorous discussion of cadence
LSST site in Chile. Typical velocities are on the order plans in the LSST community, exploring variants and
of 20 m s−1 , and are found to have a seasonable de- alternatives that enhance various specific science pro-
pendence that is modeled when generating the screens. grams, while maintaining the science requirements de-
Each photon’s trajectory is altered due to refraction as scribed in the SRD.
it passes through each screen. The main deep-wide-fast survey will use about 90%
After the atmospheric refraction, the photons in of the observing time. The remaining 10% of the ob-
PhoSim are reflected and refracted by the optical sur- serving time will be used to obtain improved coverage
faces within the telescope and camera. The mirrors of parameter space such as very deep (r ∼ 26) obser-
and lenses are simulated using geometric optics tech- vations, observations with very short revisit times (∼1
Figure 17. A simulated image of a single LSST CCD using PhoSim (covering a 13.3 × 13.3 arcmin2 region of the sky). The
image is a color composite (Lupton et al. 2004) from a set of 30 second gri visits.
minute), and observations of “special” regions such as visits separated by 15–60 minutes. This strategy will
the Ecliptic plane, Galactic plane, and the Large and provide motion vectors to link detections of moving ob-
Small Magellanic Clouds. jects in the Solar System, and fine-time sampling for
3.1.1. The Main Deep-Wide-Fast Survey and its measuring short-period variability. The ranking criteria
extensions also ensure that the visits to each field are widely dis-
tributed in position angle on the sky and rotation angle
The observing strategy for the main survey will be
of the camera in order to minimize systematic effects in
optimized for the homogeneity of depth and number of
galaxy shape determination.
visits. In times of good seeing and at low airmass, pref-
The universal cadence provides most of LSST’s power
erence is given to r-band and i-band observations. As
for detecting Near Earth Objects (NEO) and Kuiper
often as possible, each field will be observed twice, with
The resulting sky coverage for the LSST baseline ca-

dence (known internally as minion 1016), based on de-
tailed operations simulations, is shown for the r band
in Fig. 18. The anticipated total number of visits for a
ten-year LSST survey is about 2.45 million (∼4.9 million
15-second long exposures, summing over the six filters).
The per-band allocation of these visits is shown in Ta-
ble 1.
The baseline universal cadence is by no means the
definitive plan for the entire survey. Rather, it repre-
sents a proof of concept that it is indeed possible to
design a observing strategy which addresses a wide vari-
ety of science goals in a nearly optimal way. With input
Figure 18. The distribution of the r band visits on the sky and engagement of the community, we are undertak-
for a simulated realization of the baseline cadence. The sky ing a vigorous and systematic research effort to explore
is shown in the equal-area Mollweide projection in equato- the enormously large parameter space of possible sur-
rial coordinates (the vernal equinoctial point is in the center, veys (see LSST Science Collaboration et al. 2017). The
and the right ascension is increasing from right to left). The
scientific commissioning period will be used to test the
number of visits for a 10-year survey, normalized to the SRD
design value of 184, is color-coded according to the legend. usefulness of various observing modes and to explore al-
The three regions with smaller number of visits than the ternative strategies.
main survey (“mini-surveys”) are the Galactic plane (arc on
the right), the region around the South Celestial Pole (bot-
tom), and the so-called “northern Ecliptic region” (upper 3.1.2. Mini-surveys and Deep Drilling Fields
left; added in order to increase completeness for moving ob-
jects). Deep drilling fields, with a much higher number of Although the uniform treatment of the sky provided
visits than the main survey, are also visible as small circles. by the universal cadence proposal can satisfy the ma-
The fields were dithered on sub-field scales and pixels with jority of LSST scientific goals, roughly 10% of the time
angular resolution of ∼30 arcmin were used to evaluate and will be allocated to other strategies that significantly
display the coverage.
enhance the scientific return. These surveys aim to ex-
tend the parameter space accessible to the main survey
Belt Objects (KBOs) and naturally incorporates the by going deeper or by employing different time/filter
southern half of the ecliptic within its 18,000 square de- sampling. We have already discussed three examples
grees, with a declination cut of about δ = +2◦ . Ad- of such mini-surveys: the Northern Ecliptic Spur to im-
ditional coverage of a crescent within 10 degrees of the prove completeness of the asteroid and KBO population,
Northern ecliptic plane would sample the full azimuthal the Southern Celestial Cap to extend the survey foot-
distribution of KBOs, crucial for understanding the dif- print to the South Pole (thus providing coverage of the
ferent dynamical families in which they fall. Thus, we Magellanic Clouds), and the Galactic Plane survey to
plan to extend the universal cadence to this region using include low Galactic latitude fields.
the r and i filters only, along with more relaxed limits As an additional example of a mini-survey, consider a
on airmass and seeing. Relaxed limits on airmass and program that uses one hour of observing time per night
seeing are also adopted for ∼700 deg2 around the South to observe a single pointing (9.6 deg2 ) to substantially
Celestial Pole, allowing coverage of the Large and Small greater depth in individual visits. Accounting for read-
Magellanic Clouds (Fig. 18). out time and filter changes, it could obtain about 50
Finally, the universal cadence proposal excludes ob- consecutive 15-second exposures in each of four filters
servations at low Galactic latitudes, where the high stel- in an hour. If a field is visited every two days over
lar density leads to a confusion limit at much brighter four months, about 600 deg2 can be observed with this
magnitudes than those attained in the rest of the sur- cadence over 10 years. Taking weather into account,
vey. Within this region, the Galactic plane proposal the selected fields would each have on average about 40
provides 30 observations in each of the six filters, dis- hour-long sequences of 200 exposures each. Each 15-
tributed roughly logarithmically in time (it may not be second exposure in a sequence would have an equivalent
necessary to use the u and g filters for this heavily ex- 5σ depth of r ∼ 24, and each filter subsequence when
tincted region). coadded would be 2 magnitudes deeper than the main
survey visits (r ∼ 26.5). When all 40 sequences and the
main survey visits are coadded, they would extend the Table 2. The Parameters From Eqs. 5 and 6
depth to r ∼ 28.
This data set would be excellent for a wide variety u g r i z y
a
of science programs. The individual sequences would be msky 22.99 22.26 21.20 20.48 19.60 18.61
sensitive to 1% variability on sub-minute time scales, al- θb 0.81 0.77 0.73 0.71 0.69 0.68
lowing discovery of planetary eclipses and of interstellar θef f c 0.92 0.87 0.83 0.80 0.78 0.76
scintillation effects, expected when the light of a back- γd 0.038 0.039 0.039 0.039 0.039 0.039
ground star propagates through a turbulent gas medium km e 0.491 0.213 0.126 0.096 0.069 0.170
(Moniez 2003; Habibi et al. 2011). If these fields were Cm f 23.09 24.42 24.44 24.32 24.16 23.73
selected at Galactic latitudes of |b| ∼ 30 deg, they would m5 g 23.78 24.81 24.35 23.92 23.34 22.45
include about 10 million stars with r < 21 observed with ∆Cm ∞h
0.62 0.18 0.10 0.07 0.05 0.04
signal-to-noise ratio above 100 in each visit. When sub- ∆Cm (2)i 0.23 0.08 0.05 0.03 0.02 0.02
sequences from a given night were coadded, they would ∆m5 j 0.21 0.16 0.14 0.13 0.13 0.14
provide dense time sampling to a faint limit of r ∼ 26.5 a
and would enable deep searches for SN, trans-Neptunian The expected median zenith sky brightness at Cerro
objects, and other faint transient, moving and variable Pachón (AB mag arcsec−2 ).
b
sources. For example, the SN sample would be extended The expected delivered median zenith seeing (FWHM, arc-
to redshifts of z ∼ 1.2, with more densely sampled light sec). The seeing approximately scales with airmass, X, as
curves than obtained from the universal cadence. Such X 0.6 .
c
sequences would also serve as excellent tests of our pho- The effective zenith seeing (arcsec) used for m5 computa-
tometric calibration procedures. tion.
d
The LSST has already selected four distant extra- The band-dependent parameter from Eq. 5.
galactic survey fields14 that the project guarantees to e
Adopted atmospheric extinction.
observe as Deep Drilling Fields with deeper coverage and f
more frequent temporal sampling than provided by the The band-dependent parameter from Eq. 6.
g
standard LSST observing pattern. These fields (Elias The typical 5σ depth for point sources at zenith, assuming
S1, XMM-LSS, Extended Chandra Deep Field-South, exposure time of 2×15 sec, and observing conditions as
and COSMOS) are well-studied survey fields with sub- listed. For larger airmass the 5σ depth is brighter; see the
stantial existing multiwavelength coverage and other bottom row.
h
positive attributes. These four fields are only the first The loss of depth due to instrumental noise (assuming 9 e−
chosen for deep-drilling observations. The project plans per pixel and readout, and two readouts per visit).
a community call for white papers suggesting additional i
Additive correction to Cm when exposure time is doubled
deep drilling fields and other specialized observing ca- from its fiducial value to 60 sec.
dences. j The loss of depth at airmass of X = 1.2 due to seeing
3.2. Detailed Analysis of Simulated Surveys degradation and increased atmospheric extinction.
As examples of analysis enabled by the Operations

Simulator (§ 2.7.1), we describe determination of the
completeness of the LSST NEO sample, and estimation
of errors expected for trigonometric parallax and proper magnitudes (roughly the inverse of the signal-to-noise
motion measurements. In both examples, the conclu- ratio) for a single visit can be written as
sions crucially depend on the assumed accuracy of the
photometry and astrometry, as we now describe. σ12 = σsys
2 2
+ σrand , (4)
3.2.1. Expected Photometric Signal-to-Noise Ratio
where σrand is the random photometric error and σsys
The output of operations simulations is a data stream
is the systematic photometric error (due to, e.g., im-
consisting of a position on the sky and the time of obser-
perfect modeling of the point spread function, but not
vation, together with observing conditions such as seeing
including uncertainties in the absolute photometric ze-
and sky brightness. The expected photometric error in
ropoint). The calibration system and procedures are
designed to maintain σsys < 0.005 mag. Based on SDSS
14 For details, see https://www.lsst.org/News/enews/ experience (Sesar et al. 2007), the random photometric
deep-drilling-201202.html error for point sources, as a function of magnitude, is
well described15 by To predict 5σ depths for exposure time τ times longer

than the fiducial tvis = 30 sec., the following correction
2
σrand = (0.04 − γ) x + γ x2 (mag2 ), (5) should be added to the values of Cm listed in Table 2:
with x ≡ 100.4 (m−m5 ) . Here m5 is the 5σ depth (for ∞
10(0.8 ∆Cm ) − 1

∞
point sources) in a given band, and γ depends on the sky ∆Cm (τ ) = ∆Cm − 1.25 log10 1 + .
τ
brightness, readout noise, etc. Detailed determination (7)
of the system throughput yields the values of γ listed in By definition, ∆Cm (τ = 1) = 0. Again, this effect is
Table 2. The 5σ depth for point sources is determined only substantial in the u band, as demonstrated by the
from values of ∆Cm (τ = 2) listed in Table 2.
The loss of depth at the airmass of X = 1.2 due to
m5 = Cm + 0.50 (msky − 21) + 2.5 log10 (0.7/θef f ) +
seeing degradation and increased atmospheric extinction
+1.25 log10 (tvis /30) − km (X − 1) (6) is listed in the last row in Table 2. Note that the limiting
depth predictions are uncertain by about 0.1–0.2 mag
where msky is the sky brightness (AB mag arcsec−2 ),
due to unpredictable solar activity (which influences the
θef f is the seeing (in arcsec), tvis is the exposure time
night sky brightness, Patat 2008).
(seconds), k is the atmospheric extinction coefficient,
and X is airmass. Here the seeing corresponds to the 3.2.2. The NEO Completeness Analysis
“effective” seeing computed from the seeing FWHM fol-
lowing the procedure described in Angeli et al. (2016). Detailed analyses of the LSST completeness for PHAs
The seeing FWHM in each band is listed in the second and NEOs are described in Jones et al. (2018), Vereš
row of Table 2, and the effective seeing is listed in the & Chesley (2017a,b), and Grav et al. (2016). After ac-
third row of Table 2. counting for differences in their input assumptions and
The constants Cm depend on the overall throughput models, each of these independent works calculates a
of the instrument and are computed using our current completeness value which is consistent within a few per-
best throughput estimates for optical elements and sen- cent. Here we briefly summarize the LSST project anal-
sors. The resulting Cm values are listed in Table 2 and ysis carried out in Jones et al. (2018); this approach
in all six bands they imply single visit depths m5 (also is roughly the same for each of the studies mentioned
listed in Table 2) that lie between the minimum and de- above.
sign specification values from the Science Requirements To assess the LSST completeness for PHAs, the PHA
Document listed in Table 1. The differences in perfor- population is represented by a sample of orbits taken
mance between LSST and, for example, SDSS follow from the Solar System model of Grav et al. (2007).
directly from these relations16 . The simulated baseline survey is used to determine
The structure of eq. 6 nicely illustrates decoupling be- which PHAs are present in each exposure and at what
tween the system sensitivity which is fully absorbed into signal-to-noise ratio they were observed. In addition
Cm and observing conditions specified by msky , θ, tvis , to seeing, atmospheric transparency, and sky back-
km and X. The computation of Cm listed in Table 2 ground effects (see eq. 6), the signal-to-noise compu-
assumed instrumental noise of 9 e− per pixel and per tation takes into account losses due to non-optimal de-
readout, whose effect on m5 is significant only in the tection filters and object trailing. Using mean aster-
u band. This loss of depth due to instrumental noise, oid reflectance spectra (DeMeo et al. 2009), combined
∆Cm ∞
, is listed in Table 2; it also corresponds to an addi- with the LSST bandpasses, we calculate expected mag-
tive correction to Cm when the exposure time tvis → ∞. nitudes and colors, assuming all PHAs are C type as-
teroids, of V − m = (−1.53, −0.28, 0.18, 0.29, 0.30, 0.30)
for m = (u, g, r, i, z, y) to transform standard V band
15 Eq. 5 can be derived from σ
rand = N/S, where N is noise and magnitudes to the magnitudes expected in each filter
S is signal, and by assuming that N 2 = No2 + αS. The constants
No and α can be expressed in terms of a single unknown constant
(Ivezić et al. 2001). Due to very red V − u colors, and
γ by using the condition that σrand = 0.2 for m = m5 . the relatively bright limiting magnitude in the y band,
16 SDSS data typically reach a 5σ depth for point sources of the smallest objects are preferentially detected in the
r = 22.5 with an effective aperture of D = 2.22 m, an exposure griz bands. The correction for trailing is implemented
time of tvis = 54 sec, the median r band sky brightness of rsky =
20.9 mag arcsec−2 , the median seeing of θ = 1.5 arcsec, and the by subtracting from the right-hand side of eq. 6
median airmass of X = 1.3. In comparison, the LSST loses 0.32
∆mtrailing = 1.25 log10 1 + 0.42 x2

mag in depth due to shorter exposures, and gains 1.17 mag due
5 (8)
to larger aperture, 0.83 mag due to better seeing, and 0.20 mag
due to fainter sky, for a net gain of ∼1.9 mag. v tvis
x= , (9)
24 θ
LSST Baseline
where the object’s velocity, v, is expressed in deg. day−1 . 1.0
NEOs
1.0
PHAs
Existing resources Existing resources
For the nominal exposure time of 30 seconds and θ = 0.7 LSST Only
LSST + Existing
LSST Only
LSST + Existing
81%
0.8 0.8
arcsec, the loss of limiting magnitude is 0.04 mag for JPL NEO (D > 140m) 73%
NEO completeness H22
PHA completeness H22

v = 0.25 deg. day−1 , typical for objects in the main as- 0.6 0.6
teroid belt, and 0.46 mag for v = 1.0 deg. day−1 , typical
0.4 0.4
of PHAs passing near Earth. PHAs are characterized by
their “absolute magnitude” H, i.e., their apparent mag- 0.2 0.2
nitude if they were placed 1 AU from both the Sun and 0.0 0.0
the Earth, with a phase angle of 0◦ . For a given albedo, 2015 2020 2025
Year
2030 2035 2015 2020 2025
Year
2030 2035
Extended LSST Survey (12 years, 30 day linking windows)

H scales directly with diameter of the asteroid. The NEOs PHAs
1.0 1.0
PHA orbits are cloned over an H magnitude distribu- Existing resources
LSST Only
Existing resources
LSST Only 86%
tion with dN/dH = 10α H , with α = 0.33, in order to 0.8

LSST + Existing
JPL NEO (D > 140m)
77% 0.8
LSST + Existing
NEO completeness H22
PHA completeness H22

evaluate completeness as a function of H. 0.6 0.6
An object is considered to be discovered if the object
was detected on at least three nights within a window 0.4 0.4
of 15 days, with a minimum of two visits per night. 0.2 0.2
The same criterion has been used in NASA studies,

and is confirmed as reliable by a detailed analysis of 0.0
2015 2020 2025 2030 2035
0.0
2015 2020 2025 2030 2035
Year Year
orbital linking and orbit determination using the Mov-
ing Object Processing System (MOPS) code (Vereš & Figure 19. Cumulative completeness of the LSST survey
Chesley 2017a,b; Jedicke et al. 2005) developed by the for NEOs (left in each panel) and PHAs (right in each panel)
brighter than a given absolute magnitude, H ≤ 22 (related to
Pan-STARRS project (and adopted by LSST in a col-
the size of the object and albedo; H=22 mag is equivalent to
laborative effort with Pan-STARRS). The MOPS soft- a typical 140 m asteroid). The top panel illustrates cumula-
ware system and its algorithms are significantly more tive completeness for the LSST baseline cadence and MOPS
advanced than anything previously fielded for this pur- configuration. In the baseline, LSST alone would discover
pose to date. Realistic MOPS simulations show >99% 66% of the PHAs with H ≤ 22 (61% of NEOs); LSST com-
linking efficiency across all classes of Solar System ob- bined with current and on-going surveys can discover 81%
jects (Denneau et al. 2013), and at least 93% efficiency of PHAs (73% of NEOs). The bottom panel illustrates cu-
for NEOs (Vereš & Chesley 2017a,b). mulative completeness when LSST is operated for 12 years,
with extra visits around the ecliptic, and when the MOPS
The LSST baseline cadence discovers 66% of PHAs
linking window is increased to 30 days from the baseline 15.
and 61% of NEOs with H ≤ 22 (equivalent to D ≥ In this case, LSST alone could discover 74% of the PHAs
140 m) after 10 years of operations (Jones et al. 2018). with H ≤ 22 (69% of NEOs); LSST combined with existing
This cadence spends 6% of the total observing time on resources could discover 86% of PHAs (77% of NEOs).
NEO-optimized observations north of δ = +5◦ , and
MOPS links objects with windows of 15 days. The isting/ongoing surveys would result in a system-wide
baseline survey cumulative completeness as a function cumulative completeness of 86% for PHAs (77% for
of time for objects with H ≤ 22 is shown in the up- NEOs), approaching the 90% required by the Congres-
per panel of Fig. 19, both with and without including sional mandate (see lower panels of Fig. 19).
contributions from current and on-going surveys. These
figures are likely to be uncertain at the level of ±5% 3.2.3. The Expected Accuracy of Trigonometric Parallax
due to uncertainties in the orbital distribution of the and Proper Motion Measurements
true population, the size distribution, uncertain distri-
To model the astrometric errors, we need to consider
butions of shape (and thus light curve variations) and
both random and systematic effects. Random astro-
surface properties (thus colors and albedo), plus varia-
metric errors per visit for a given star are modeled as
tions in survey cadence due to weather, etc.
θ/SN R, with θ = 700 mas and SN R determined using
Various adjustments to the baseline cadence and
eq. 6. Systematic errors of 10 mas are added in quadra-
MOPS can boost the completeness for H ≤ 22 PHAs.
ture, and are assumed to be uncorrelated between differ-
By improving MOPS and increasing the MOPS linking
ent observations of a given object. Systematic and ran-
window from 15 to 30 days we can boost completeness
dom errors become similar at about r = 22, and there
by about 3%. By running the survey for an additional
are about 100 stars per LSST sensor (0.05 deg2 ) to this
two years, we can boost completeness by another 4%.
depth (and fainter than the LSST saturation limit at
Considering this ‘extended’ LSST in the context of ex-
r ∼ 16) even at the Galactic poles.
Table 3. The expected proper motion, par- vatory (Monet et al. 2003). The expected proper motion
allax and accuracy for a 10-year long baseline and parallax errors for a 10-year long baseline survey, as
survey. a function of apparent magnitude, are summarized in
Table 3. Blue stars (e.g., F/G stars) fainter than r ∼ 23
r σxy a σπ b σµ c σ1 d σC e
will have about 50% larger proper motion and parallax
mag mas mas mas/yr mag mag errors than given in the table due to decreased numbers
21 11 0.6 0.2 0.01 0.005 of z and y band detections. The impact on red stars is
22 15 0.8 0.3 0.02 0.005 smaller due to a relatively small number of observations
23 31 1.3 0.5 0.04 0.006 in the u and g bands, but extremely red objects, such
24 74 2.9 1.0 0.10 0.009 as L and T dwarfs, will definitely have larger errors, de-
pending on details of their spectral energy distributions.
a Typical astrometric accuracy (rms per coor-
After the first three years of the survey, the proper mo-
dinate per visit). tion errors will be about five times as large, and parallax
b Parallax accuracy for 10-year long survey. errors will be about twice as large, as the values given in
c Proper motion accuracy for 10-year long Table 3; the errors scale as t−3/2 and t−1/2 , respectively.
survey. This error behavior is a strong independent argument for
d Photometric error for a single visit (two 15- a survey lifetime of at least 10 years (c.f. § 2).
For comparison with Table 3, the SDSS-POSS proper
second exposures).
motion measurements have an accuracy of ∼5 mas yr−1
e Photometric error for coadded observations
per coordinate at r = 20 (Munn et al. 2004). Gaia is
(see Table 1). expected to deliver parallax errors of 0.3 mas and proper
motion errors of 0.2 mas yr−1 at its faint end at r ∼
20 (Perryman et al. 2001). Hence, LSST will smoothly
extend Gaia’s error vs. magnitude curve 4 magnitudes
HSC data from the Subaru telescope reduced with the fainter (for illustration, see fig. 21 in Ivezić et al. 2012).
LSST software stack indicate that systematic errors of
10 mas on spatial scales of several arcminutes are real- 3.3. Data Products and Archive Services
istic even at this stage of maturity of the code; results Data collected by the LSST telescope and camera will
reported by DES (Bernstein et al. 2017) indicate as- be automatically processed to data products – catalogs,
trometric residuals of ∼ 7 mas for 30 s exposures in a alerts, and reduced images – by the LSST Data Manage-
4m, with scope for further improvements from denser ment system (§ 2.6.3). These products are designed to
astrometric standard grids. Even a drift-scanning sur- enable a large majority of LSST science cases, without
vey such as SDSS delivers uncorrelated systematic errors the need to work directly with the raw pixels. We give
(dominated by seeing effects) at the level of 20-30 mas a high-level overview of the LSST data products here;
(measured from repeated scans; Pier et al. 2003); the further details may be found in the LSST Data Products
expected image quality for LSST will be twice as good Definition Document (Jurić et al. 2017b).
as for SDSS. Furthermore, there are close to 1000 galax- Two major categories of data products will be pro-
ies per sensor with r < 22, which will provide exquisite duced and delivered by LSST DM:
control of systematic astrometric errors as a function of
magnitude, color and other parameters, and thus enable • Prompt products17 , designed to support the
absolute proper motion measurements. discovery, characterization, and rapid follow-up of
Given the observing sequence for each sky position in time-dependent phenomena (“transient science”).
the main survey as produced by the Operations Simula- These will be generated continuously every ob-
tor (§ 2.7.1), we generate a time sequence of mock astro- serving night, by detecting and characterizing
metric measurements, with random and statistical errors sources in images differenced against deep tem-
modeled as described above. The astrometric transfor- plates. They will include alerts to objects that
mations for a given CCD and exposure, and proper mo-
tion and parallax for all the stars from a given CCD, are 17 Historically, these have been referred to as “Level 1 Data
simultaneously solved for using an iterative algorithm. Products”, but going forward we prefer to use the more descrip-
The astrometric transformations from pixel to sky co- tive Prompt Products designation. Note that the old terminology
ordinates are modeled using low-order polynomials and is still in use in present-day LSST documents and code; new and
updated documents will gradually transition to the new, descrip-
standard techniques developed at the U.S. Naval Obser- tive, nomenclature used in this paper.
were newly discovered, or have changed bright- on suitable linear combinations of input images
ness or position at a statistically significant level. (with careful propagation of PSFs and noise). An
The alerts to such events will be published within extended source model – a constrained linear com-
60 seconds of observation; we expect several mil- bination of two Sérsic profiles – and a point source
lion alerts per night. model with proper motion – will generally be fit-
ted to each detected object20 .
In addition to transient science, the prompt prod-
ucts will support discovery and follow-up of ob- Secondly, for the extended source model fits, the
jects in the Solar System. Objects with motions LSST will characterize and store the shape of the
sufficient to cause trailing in a single exposure will associated likelihood surface (and the posterior),
be identified and flagged as such when the alerts and not just the maximum likelihood values and
are broadcast. Those that are not trailed will be covariances. The characterization will be accom-
identified and linked based on their motion from plished by sampling, with up to ∼200 (indepen-
observation to observation, over a period of a few dent) likelihood samples retained for each object.
days. Their orbits as derived by MOPS will be For storage cost reasons, these samples may be
published within 24 hours of identification. The retained only for those bands of greatest interest
efficiency of linking (and thus the completeness of for weak lensing studies.
the resulting orbit catalog) will depend on the fi-
nal observing cadence chosen for LSST, as well as As described in § 3.1.2, approximately 10% of the ob-
the performance of the linking algorithm (§ 3.2.2). serving time will be devoted to mini-surveys that do not
follow the LSST baseline cadence. The data products for
• Data release products18 are designed to enable these programs will be generated using the same process-
systematics- and flux-limited science, and will be ing system and will be released on the same timescale
made available in annual Data Releases19 . These as the rest of the survey; any specialized processing that
will include the (reduced and raw) single-epoch these require will be the responsibility of the community.
images, deep coadds of the observed sky, cata- While a large majority of science cases will be ad-
logs of objects detected in LSST data, catalogs of equately served by prompt and data release products,
sources (the detections and measurements of ob- more specialized investigations may benefit from cus-
jects on individual visits), and catalogs of “forced tom, user-created, products derived from the LSST
sources” (measurements of flux on individual vis- data. These could be new catalogs created by sim-
its at locations where objects were detected by ple post-processing of the LSST data release catalogs,
LSST or other surveys). LSST data releases will entirely new data products generated by running cus-
also include fully reprocessed prompt products, tom code on raw LSST imaging data, or something in-
as well as all metadata and software necessary between. We will make it possible for the end-users to
for the end-user to reproduce any aspect of LSST create (or use) such user-generated21 products at the
data release processing. LSST Data Facility, using the services offered within the
LSST Science Platform (§ 3.3.1).
A noteworthy aspect of LSST data release pro-
cessing is that it will largely rely on multi-epoch 3.3.1. The LSST Science Platform
model fitting, or MultiFit, to perform near-
optimal characterization of object properties. The LSST Science Platform (Jurić et al. 2017a) repre-
That is, while the coadds will be used to per- sents LSST’s vision for a large-scale astronomical data
form object detection, the measurement of their archive that can enable effective research with datasets
properties will be performed by simultaneously of LSST size and complexity. It builds on recent trends
fitting (PSF-convolved) models to all single-epoch in remote data analysis, and practical experiences in the
observations. It is not yet clear to what extent we astronomical context gathered by projects such as the
will be able to make some of these measurements JHU SciServer (Raddick et al. 2017), Gaia GAVIP (Vagg
et al. 2016), or NOAO Datalab (Fitzpatrick et al. 2016).
18 These have been referred to as “Level 2 Data Products” in
the past; as with their “Level 1” counterparts, we will use the 20 For performance reasons, it is likely that only the point source
more descriptive nomenclature going forward. model will be fitted in the most crowded regions of the Galactic
19 The first-year data will probably be split into two data re- plane.
leases. 21 Formerly known as “Level 3 Data Products”.
The LSST Science Platform will be a set of web appli- served for the LSST Science Platform needs, and to be
cations (portals) and services through which the users shared by all LSST DAC users. Based on the current
will access the LSST data products and, if desired, con- plans and technology projections, these equate to ap-
duct remote analyses or create user generated products. proximately 2,400 cores, 4 PB of file storage, and 3 PB
The platform makes this possible through three user- of database storage at the beginning of LSST operations
facing aspects: (in 2022).
• The web Portal, designed to provide the essen-
4. EXAMPLES OF LSST SCIENCE PROJECTS
tial data access and visualization services through
a simple-to-use website. It will enable querying The design and optimization of the LSST system
and browsing of the available datasets in ways the leverages its unique capability to scan a large sky area
users are accustomed to at archives such as IRSA, to a faint flux limit in a short amount of time. The
MAST, or the SDSS archive. main product of the LSST system will be a multi-color
ugrizy image of about half the sky to unprecedented
• The JupyterLab aspect, that will provide a
depth (r ∼ 27.5). For a comparison, one of the best
Jupyter22 Notebook-like interface and is geared to-
analogous contemporary datasets is that of SDSS, which
wards enabling next-to-the-data remote analysis.
provides ugriz images of about a quarter of the sky to
A large suite of commonly used astronomical soft-
r ∼ 22.5, with twice as large seeing (see Figs. 20 and 21).
ware, including the LSST software stack (§ 2.6.4),
A major advantage of LSST is the fact that this deep sky
will be made available through this interface. The
map will be produced by taking hundreds of shorter ex-
user experience will be nearly identical to working
posures (see Table 1). Each sky position within the main
with Jupyter notebooks locally, except that com-
survey area will be observed close to 1000 times, with
putation and analysis will occur with resources
time scales spanning seven orders of magnitude (from
provided at the LSST Data Access Center. This
30 sec to 10 years), and produce roughly thirty trillion
is an implementation of the “bringing computa-
photometric measures of celestial sources.
tion to the data” paradigm: rather than imposing
It is not possible to predict all the science that LSST
the burden of downloading, storing, and process-
data will enable. We now briefly discuss a few projects
ing (potentially large) subsets of LSST data at
to give a flavor of anticipated studies, organized by the
their home institutions, we make it possible for
four science themes that drive the LSST design (al-
the users to bring their codes and perform anal-
though some projects span more than one theme). For
yses at the LSST DAC. This reduces the barrier
an in-depth discussion of LSST science cases, we refer
to entry and shortens the path to science for the
the reader to the LSST Science Book, and more special-
LSST science community.
ized documents discussing cosmology (LSST Dark En-
• The Web API aspect will expose the LSST DAC ergy Science Collaboration 2012; Zhan & Tyson 2018),
services to other software tools and services us- galaxy science (Robertson et al. 2017), and synergy with
ing commonly accepted formats and protocols23 . other ground-based and space-based facilities (Najita
This interface will open the possibility for remote et al. 2016; Jain et al. 2015; Rhodes et al. 2017).
access and analysis of the LSST data set using
applications that the users are already comfort- 4.1. Probing Dark Energy and Dark Matter
able with such as TOPCAT (Taylor 2005), or li- A unique aspect of LSST as a probe of dark energy
braries such as Astropy (Astropy Collaboration and dark matter is the use of multiple cross-checking
et al. 2013; Jenness et al. 2016). Furthermore, the probes that reach unprecedented precision (see Fig. 22).
offered APIs will allow for federation with other Any given probe constrains degenerate combinations of
astronomical archives, bringing added value to the cosmological parameters, and each probe is affected by
LSST dataset. different systematics, thus the combination of probes al-
Approximately 10% of the total budget for the LSST lows systematics to be calibrated and for degeneracies
Data Facility compute and storage capacity has been re- to be broken. Dark energy manifests itself in two ways.
The first is the relationship between redshift and dis-
22
tance (the Hubble diagram), or equivalently the expan-
http://jupyter.org/
23
sion rate of the Universe as a function of cosmic time.
For example, industry-standard protocols such as WebDAV
may be used to expose file data, or Virtual Observatory protocols The second is the rate at which matter clusters with
such as TAP and SIAP may be used for access to catalogs and time. Structure formation involves a balance between
images respectively. gravitational attraction of matter over-densities and the
Figure 20. A comparison of ∼ 7.5 × 7.5 arcmin2 images of Figure 21. A comparison of angular resolution for 20 × 20
the same area of sky (centered on α=9h 200 4700 and δ=30◦ arcsec2 images obtained by the SDSS (top, median seeing
80 1200 ) obtained by the SDSS (top, r < 22.5) and the Deep of 1.5 arcsec) and expected from LSST (bottom, seeing of
Lens Survey (bottom, r < 24.5). These are gri composites, 0.7 arcsec). The images show a lensed SDSS quasar (SDSS
colorized following Lupton et al. (2004). The depth gain for J1332+0347, Morokuma et al. 2007); the bottom image was
the bottom image is mostly due to the lower surface bright- taken with Suprime-cam at Subaru. Adapted from Bland-
ness limit, which is also responsible for the apparent increase ford & LSST Strong Lensing Science Collaboration (2007).
of galaxy sizes. LSST will obtain ∼100 gri color images to
the same depth (∼200 for the riz composites) of each point
over half the Celestial sphere (18,000 deg2 , equivalent to 1.15 rapid expansion of the background. Thus, quantifying
million ∼ 7.5 × 7.5 arcmin2 regions), and with better seeing. the rate of growth of structures from early times until
After their coaddition, the final image will be another ∼ 3 the present provides additional tests of the energy con-
mag deeper (a faint limit of r = 27.5 for point sources). tents of the Universe and their interactions.
The joint analysis of LSST weak lensing and galaxy

clustering is particularly powerful in constraining the
dynamical behavior of dark energy, i.e., how it evolves
with cosmic time or redshift (Hu & Jain 2004; Zhan
2006). By simultaneously measuring the growth of large-
scale structure, and luminosity and angular distances as
functions of redshift (via weak lensing, LSS, SN, and
cluster counting), LSST data can reveal whether the re-
cent cosmic acceleration is due to dark energy or mod-
ified gravity (Lue et al. 2004; Knox et al. 2006; Ishak
et al. 2006; Jain & Zhang 2008; Oguri & Takada 2011;
Jain et al. 2013; Weinberg et al. 2013). The Dark En-
ergy Survey (see e.g., DES Collaboration et al. 2017,
and references therein) provides a compelling proof of
concept for this program.
Over a broad range of accessible redshifts, the sim-
ple linear model for the dark energy equation of state
(w = w0 + wa (1 − a)) is a poor representation of more
general dark energy theories. Barnard et al. (2008)
Figure 22. Constraints on the dark energy equation of
showed that in a high-dimensional dark energy model state (w = w0 + wa (1 − a)) from LSST cosmological probes.
space, LSST data could lead to a hundred- to thousand- The various ellipses assume constraints from BAO (dashed
fold increase in precision over its precursor experiments, line), cluster counting (dash-dotted line), supernovae (dot-
thereby confirming its status as a premier Stage IV ex- ted line), WL (solid line), joint BAO and WL (green shaded
periment in the sense of Albrecht et al. (2006). area), and all probes combined (yellow shaded area). The
The power and accuracy of LSST dark energy and BAO and WL results are based on galaxy–galaxy, galaxy–
shear, and shear–shear power spectra only. Adding other
dark matter probes are a result of the enormous sam-
probes such as strong lensing time delay and higher-order
ples that LSST will have, including several billion galax- galaxy and shear statistics will further improve the con-
ies and millions of Type Ia supernovae. At i < 25.3 straints. While the details of the contours will change slightly
(SNR > 20 for point sources), the photometry of galax- as the survey parameters are updated, the key point remains
ies will be of high enough quality to provide photomet- that this combination of dark energy probes results in con-
ric redshifts with an RMS accuracy (σ/(1 + z)) of 2% tours with different degeneracy directions, and hence their
over the range 0.3 < z < 3.0 (only 10% of the sample combination results in tight constraints on the dark energy
equation of state.
will have redshift errors larger than 6%). This number
represents a requirement on the accuracy of the pho-
tometry at delivering photometric redshifts given known the same fields (Newman 2008; Matthews & Newman
templates for the SEDs. The degradation in photomet- 2010; Ménard et al. 2013; Davis et al. 2017).
ric redshift quality associated with requiring more train- The main LSST observables in the context of dark
ing data than currently exists to define the template set energy and matter are described below.
increases the expected σ/(1 + z) to ∼0.05 (e.g. New-
• The joint analysis of shear–shear, galaxy–shear,
man et al. 2015; Graham et al. 2018), which is still well
and galaxy–galaxy correlation functions has be-
within the expected range for a Stage IV dark energy
come standard in analyses of precursor datasets
experiment. The sample to i = 25.3 will include several
(e.g. DES Collaboration et al. 2017; Joudaki et al.
billion galaxies. At a slightly brighter cut, there will be
2018). WL and LSS are highly complementary
around 30 galaxies arcmin−2 with shapes measured well
probes, and the combination of their auto- and
enough for weak lensing measurements (Chang et al.
cross-correlations will constrain the properties of
2013, 2015), with the number realized in practice being
the late-time accelerated expansion while provid-
dependent on the performance of the deblending and
ing internal cross-checks for marginalizing over
shape measurement algorithms. The median redshift
systematic uncertainties (e.g., Mandelbaum 2017).
for this sample will be z ∼1.2, with the third quartile at
These measurements consist of the two-point auto-
z ∼ 2. It will be possible to further improve photometric
and cross-correlations of shear and positions for
redshift calibration by cross-correlating the photometric
billions of galaxies across ∼ 10 redshift bins. As
sample with redshift surveys of galaxies and quasars in
described in the following two items, the galaxy-
horizon at decoupling, which is imprinted on the

mass distribution at all redshifts and calibrated
with the CMB, provides a standard ruler to mea-
sure the angular diameter distance as a function of
redshift (Eisenstein et al. 1998; Cooray et al. 2001;
Blake & Glazebrook 2003; Hu & Haiman 2003;
Linder 2003; Seo & Eisenstein 2003). LSST photo-
z BAO will achieve percent-level precision on the
angular diameter distance at ∼10 redshifts loga-
rithmically spaced between z = 0.4 to 3.6. The
combination with CMB and weak lensing (WL)
shear yields tight constraints on the dynamical
behavior of dark energy (Fig. 23). In particular,
high-redshift BAO data can break the degeneracy
between curvature and dark energy, constraining
Ωk to within 0.001.
• Higher-order shear and galaxy statistics and shear

peak counts can improve dark energy constraints
Figure 23. Marginalized 1σ errors on the comoving distance and provide self-calibration of various systematics
(open triangles) and growth factor (open circles) parameters
(Takada & Jain 2004; Dolney et al. 2006; Huterer
from the joint analysis of LSST LSS and WL (galaxy–galaxy,
galaxy–shear, and shear–shear power spectra) with a conser- et al. 2006; Petri et al. 2016). They are also probes
vative level of systematic uncertainties in the photometric of both primordial non-Gaussianities and those
redshift error distribution and additive and multiplicative caused by non-linear structure.
errors in the shear and galaxy power spectra. The maximum
multipole used for WL is 2000, and that for LSS is 3000 • Primordial non-Gaussianity is also probed by the
[with the additional requirement that ∆2δ (`/DA ; z) < 0.4]. large-scale power of any biased tracer of the matter
The growth parameters are evenly spaced in log(1 + z) be- overdensities (Dalal et al. 2008). Although mea-
tween z = 0 and 5, and the distance parameters start at surements of the galaxy power spectrum on very
z1 = 0.14. The error of each distance (growth) parameter is large scales are challenging due to sky systemat-
marginalized over all the other parameters including growth
ics (Leistedt et al. 2014) and cosmic variance, the
(distance) parameters. The joint constraints on distance are
relatively insensitive to the assumed systematics (Zhan et al.
prospect of using multiple tracers of the same field
2009). could significantly improve the constraining power
for this observable (Seljak 2009). Similar measure-
ments of the large-scale power will also be used to
galaxy and galaxy-shear correlation functions pro-
test phenomenological models of clustering dark
vide additional probes of dark energy and dark
energy (Takada 2006).
matter.
• Similarly, weak lensing magnification tomography
• The galaxy–shear correlation function probes the (Morrison et al. 2012) offers a complementary
growth of dark matter large-scale structure and probe of a mix of cosmic geometry and growth
is a diagnostic of the underlying cosmology. The of dark matter structure.
combination with the galaxy–velocity correlation
function estimated from currently planned spec- • The two LSST observing programs are comple-
troscopic surveys could test General Relativity and mentary in the supernova samples they will pro-
its variants at high redshift (Reyes et al. 2010). vide. The main survey will obtain light curves
in six bands and photometric redshifts of about
• The galaxy–galaxy correlation function is vital to 400,000 photometrically-classified Type Ia super-
constrain the galaxy bias impacting the galaxy- novae that can be used for cosmological distance
shear correlation and is therefore an essential com- measurements, with further spectroscopic follow-
ponent in the joint analysis of LSS and WL. In up of a sub-sample of their host galaxies. Such a
addition, the presence of Baryon Acoustic Oscilla- sample will not only provide larger statistics for
tions in the galaxy angular correlation functions is the study of the Type Ia population in the uni-
a strong cosmological probe on its own. The sound verse, but will also be spread across the full 18,000
deg2 LSST main survey footprint, allowing dif- mass function on small scales (Dalal & Kochanek
ferent probes of the large scale structure of the 2002).
low redshift universe. This sample of supernovae
can be used as a tracer of large scale structure • The abundance of galaxy clusters as a function of
by directly probing the gravitational potential of mass and redshift is sensitive to cosmological pa-
structure through inferences of their peculiar ve- rameters (SciBook, Ch. 13; von der Linden et al.
locities (Gordon et al. 2007; Bhattacharya et al. 2014). LSST will produce a large catalog of clus-
2011; Howlett et al. 2017), weak lensing of super- ters detected through their member galaxy popu-
nova brightnesses (Dodelson & Vallinotto 2006; lation to redshift z ∼ 1.2. In addition, LSST will
Quartin et al. 2014; Macaulay et al. 2017; Scov- identify optical counterparts and provide deep op-
acricchi et al. 2017), and the local bulk flow (Riess tical imaging for clusters detected in other wave-
2000; Dai et al. 2011; Turnbull et al. 2012; Feindt bands (e.g., Staniszewski et al. 2009).
et al. 2013; Huterer et al. 2015), as well as low red-
• The clustering properties of those same galaxy
shift constraints on the isotropy of the universe
clusters will also be used to constrain cosmologi-
(Antoniou & Perivolaropoulos 2010; Colin et al.
cal parameters (Mo et al. 1996; Mana et al. 2013),
2011; Campanelli et al. 2011; Cai et al. 2013; Ja-
to marginalize over uncertainties in the mass-
vanmardi et al. 2015). The rapidly sampled deep
observable relation and photometric redshift un-
drilling fields, possibly coadded over short time
certainties (Oguri & Takada 2011), and to con-
scales, will yield well-sampled light curves of tens
strain the effects of super-sample covariance in
of thousands of supernovae to redshifts peaking
the two-point functions of WL and LSS (Hu &
around z ∼ 0.7 and reaching beyond a redshift of
Kravtsov 2003; Takada & Spergel 2014).
1.0, limited by the systematics related to the limits
of our astrophysical understanding of supernovae • LSST will discover several hundred galaxy clusters
populations and relative photometric calibration. that produce multiple-image lenses of background
In addition to the usual use of Type Ia supernovae objects. Cluster mass reconstruction based on the
to probe the redshift-distance relation to high red- multiple image positions can probe the cluster in-
shift, the luminosities will be magnified by lensing ner mass profile, and can provide a separate test of
from foreground structure, a correlation which can cosmology, especially in cases with strongly lensed
be probed with these data. The ultimate promise background objects at different redshift (Porciani
of such supernova surveys will be linked to the ob- & Madau 2000; Oguri & Kawano 2003).
serving strategy employed by the LSST.
• Time delays of galaxy-scale lensed quasars will al-
• Cosmological analyses can be carried out using SN, low one to measure Hubble’s constant (e.g., Suyu
WL, and LSS in subsets of the LSST data in differ- et al. 2010; Bonvin et al. 2017) in hundreds of sys-
ent regions of the sky, testing fundamental cosmo- tems; sub-percent level precision in H(z) should be
logical assumptions of homogeneity and isotropy achievable (Coe & Moustakas 2009; Treu & Mar-
(e.g., Zhan et al. 2009). shall 2016), providing a further independent dark
• The shape of the power spectrum of dark matter energy probe. LSST will also discover between
fluctuations measured by LSST weak lensing will 500 and 1000 strongly lensed Type Ia supernovae
constrain the sum of neutrino masses with an ac- (Goldstein & Nugent 2017; Goldstein et al. 2017),
curacy of 0.04 eV or better (Cooray 1999; Song which will provide hundreds of additional high-
& Knox 2004; Hannestad et al. 2006). Given the quality time delays. Time delays for quasars mul-
current constraints on neutrino mass mixing, this tiply lensed by clusters as a function of redshift are
is at the level to determine whether there is an an independent test of dark energy (Kundić et al.
inverted neutrino mass hierarchy, a fundamental 1997). The natural timescale (many months to
question in particle physics. years) is well matched to the LSST survey (Oguri
& Marshall 2010).
• Tens of thousands of galaxy-galaxy lenses will pro-
vide the needed statistics to probe dark matter • Standard sirens are a new cosmological probe,
halo profiles and substructure (e.g., Mandelbaum demonstrated by the recent discovery of a binary
et al. 2006; Vegetti et al. 2012). The image fluxes neutron star merger by LIGO with an electromag-
in several thousand well-measured strongly lensed netic counterpart (Abbott et al. 2017a), which was
quasars will enable constraints of the dark matter used to constrain the Hubble parameter to roughly
15% precision (Abbott et al. 2017b). Constraints

from standard sirens are independent of the local
distance ladder, with the primary uncertainties be-
ing the local velocity field and the inclination an-
gle of the system. Scolnic et al. (2018) estimate of
order 70 such systems could be found with LSST.
4.2. Taking an Inventory of the Solar System
The small bodies of the Solar System, such as main-
belt asteroids, the Trojan populations of the giant plan-
ets and the Kuiper Belt objects, offer a unique insight
into its early stages because they provide samples of
the original solid materials of the solar nebula. Under-
standing these populations, both physically and in their
number and size distribution, is a key element in testing
various theories of Solar System formation and evolu-
tion.
The baseline LSST cadence will result in orbital pa-
Figure 24. An example of color-based asteroid taxonomy.
rameters for several million objects; these will be dom-
The figure shows the distribution of asteroids in the proper
inated by main-belt asteroids, with light curves and semi-major axis vs. sin(i) plane for 45,000 asteroids with col-
multi-color photometry for a substantial fraction of de- ors measured by SDSS (Parker et al. 2008). The color of each
tected objects. The LSST sample of asteroids with ac- dot is representative of the object’s color. Note the strong
curate orbits and multi-color light curves will be 10 to correlation between asteroid families (objects with similar
100 larger than currently available sample. LSST will orbital elements) and colors. There are at least five different
make a significant contribution to the Congressional tar- taxonomic types distinguishable with SDSS measurements;
LSST color measurements of asteroids will be more than
get completeness of 90% for PHAs larger than 140 m
twice as accurate and will increase the number of objects
(§ 3.2.2), and will detect over 30,000 TNOs brighter than by roughly two orders of magnitude.
r ∼ 24.5 using its baseline cadence. LSST will be capa-
ble of detecting objects like Sedna to beyond 100 AU,
• Studies of the distribution of orbital elements for
thus enabling in situ exploration far beyond the edge of
about 100,000 NEOs as a function of color and size
the Kuiper belt at ∼50 AU. Because most of these ob-
(Rabinowitz 1993; Dandy et al. 2003); correlations
jects will be observed several hundred times, accurate
with the analogous distributions for main-belt ob-
orbital elements, colors, and variability information will
jects, and studies of object shapes and structure
also be available.
using light curves.
The following are some examples of the LSST science
opportunities in Solar System science: • Studies of the distribution of orbital elements for
• Studies of the distribution of orbital elements for close to 300,000 Jovian Trojan asteroids as a func-
over 5 million main-belt asteroids as a function tion of color and size (Jewitt et al. 2000; Yoshida
of color-based taxonomy (see Fig. 24) and size; & Nakamura 2005; Szabó et al. 2007); the search
size distributions of asteroid families (Parker et al. for dynamical families (Knezevic & Milani 2005);
2008) and their correlations with age (Jedicke et al. studies of shapes and structure using light curves.
2004; Nesvorný et al. 2005); dynamical effects
(Bottke et al. 2001); and studies of object shapes • Studies of the distribution of orbital elements for
and spin states using light curve inversion tech- about 30,000 TNOs (see Fig. 25) as a function of
niques (Pravec & Harris 2000; Durech et al. 2009). color and size; the search for dynamical families
(Marcus et al. 2011); studies of shapes and struc-
• Studies of transient mass loss in asteroids (ac- ture using light curves (Duncan et al. 1995; Tru-
tive asteroids or main belt comets, Hsieh & Je- jillo et al. 2001; Gladman et al. 2001; Bernstein
witt 2006); such objects will appear extended in et al. 2004; Elliot et al. 2005; Jones et al. 2006;
the sensitive LSST images. Only a few such ob- Doressoundiram et al. 2007).
jects are currently known (Jewitt et al. 2011; Je-
witt 2012); LSST will increase the sample of such • An unbiased and complete census of both Jupiter-
objects to ∼100. family and Oort-cloud comets. These comets
2012 VP113 (Trujillo & Sheppard 2014). Study-

ing the distribution of their orbits (in particular
including any clustering in the argument of peri-
helion) will test models predicting the existence of
a planetary-mass object beyond Neptune, a pro-
posed Planet 9 (Trujillo & Sheppard 2014; Batygin
& Brown 2016; Brown & Batygin 2016; Sheppard
& Trujillo 2016; Brown 2017). Depending on the
proposed Planet 9’s on-sky location and bright-
ness, it may be possible for LSST to directly detect
it in the wide survey images (Batygin & Brown
2016; Brown & Batygin 2016; Sheppard & Trujillo
2016; Brown 2017).
• Mapping the propagation of coronal mass ejections

through the Solar System using induced activity in
a large sample of comets at different heliocentric
distances (SciBook Ch. 5).
Figure 25. The LSST detection limits for distant Solar Sys- • Probing the inventory and frequency of interstel-
tem objects as a function of distance. Moving objects with lar asteroids/comets. The recent Pan-STARRS1
diameters as small as 100 m in the main asteroid belt and discovery of the interstellar object 1I/2017 U1
100 km in the Kuiper Belt (TNOs) will be detected in indi- (‘Oumuamua) (Bacci et al. 2017) has shown the
vidual visits, depending on the albedo. Specialized deeper
power of large, complete all-sky surveys to unearth
observations (see § 3.1.2) will detect TNOs as small as 10
km. Adapted from Jones et al. (2007). rare and exciting classes of objects. LSST will be
some three magnitudes more sensitive than cur-
rent NEO surveys (like Pan-STARRS1), and will
will have detailed six-band high-resolution images
cover more sky more often. Therefore, LSST is
extending to low surface brightness, in multiple
likely to find more interstellar objects, and more
points through their orbits, allowing detailed stud-
frequently. Estimates from Cook et al. (2016), En-
ies of activity as a function of distance from the
gelhardt et al. (2017), and Trilling et al. (2017)
Sun (Lowry et al. 1999; A’Hearn 2004). LSST
suggest that LSST will increase the number of
will discover an unprecedentedly large number of
such rare objects by an order of magnitude which,
comets with typically 50 observations per object
among other outcomes, will help constrain the fre-
spread throughout their orbits during the 10-year
quency and properties of planetary system forma-
survey, and will help us to constrain models of
tion in the solar neighborhood.
the origin of comets (Solontoi 2010; Silsbee &
Tremaine 2016). Combining the CN production 4.3. Exploring the Transient Optical Sky
rates determined from observations in the u band-
pass, as a proxy for overall gas activity, with the Time domain science will greatly benefit from LSST’s
non-volatile production rate calculated from the unique capability to simultaneously provide large area
continuum-sensitive r, i, and z bands allows for coverage, dense temporal coverage, accurate color infor-
the determination of the gas-to-dust ratio. The re- mation, good image quality, and rapid data reduction
lationship between the gas-to-dust ratio in comets and classification. Since LSST extends time-volume-
and their dynamical class (and places of forma- color space 50-100 times over current surveys (e.g., Djor-
tion) is a fundamental, and still unresolved, ques- govski et al. 2013) it will facilitate new population and
tion in cometary science (see e.g., A’Hearn et al. statistical studies and also the discovery of new classes of
1995; Bockelée-Morvan & Biver 2017). objects. LSST data products will enable many projects
including:
• Searching for objects with perihelia out to several
hundred AU. For example, an object like Sedna • Discovery and characterization of thousands of hot
(Brown et al. 2004) would be detectable at 130 AU. Jupiters in exoplanetary systems via the transit
This will result in a much larger, well-understood method (Wright et al. 2012). LSST will extend
sample of inner Oort Cloud objects like Sedna and the extrasolar planet census to larger distances
the magnetic accretion column and in the low state

for the majority of the 10-year survey.
• Studies of transients from poorly-constrained

stages of stellar evolution including stellar erup-
tions, luminous blue variable (LBV), stellar merg-
ers, and helium core flashes leading to white dwarf
formation. We will be able to identify the progeni-
tors of eruptive transients in the deep LSST stacks
and even look for faint precursor eruptions. We
will also constrain the rates of individual erup-
tion subclasses (Smith 2014) by detecting them in
galaxies out to tens of Mpc.
• A census of light echoes of historical explosive and

eruptive transients in the Milky Way and Local
Group through high resolution time series.
Figure 26. The phase space of cosmic explosive and erup-
tive transients as represented by their absolute V band peak • Studies of known and unusual SN populations and
brightness and the event timescale, defined as the time taken parameterization of their light curves (e.g., Höflich
to drop one magnitude in V band brightness from peak lu- et al. 1998; Wang et al. 2003; Howell et al. 2007;
minosity (adapted from Kulkarni et al. (2007) and Kasliwal Kowalski et al. 2008; Hicken et al. 2009; Foley
(2011)). The locus of the Classical Novae is as described in 2012; Bianco et al. 2014; Arcavi et al. 2017), in-
della Valle & Livio (1995). LSST will open up large regions cluding late-time observations of rapidly-evolving
of this phase space for systematic exploration by extending
transients to deep limits, critical for ascertaining
time-volume space more than 100 times over existing sur-
veys.
their nature. Measurements of intrinsic rates for
both peculiar transients (e.g., Drout et al. 2014)
and for SN as a function of sub-type and host envi-
within the Galaxy, thus enabling detailed studies
ronment properties (e.g., metallicity; Graur et al.
of planet frequency as a function of stellar metal-
2017).
licity and parent population (e.g., Hartman et al.
2009; Bayliss & Sackett 2011). The out-of-transit • A deep search for new populations of novae and
variability of exoplanet host stars will also provide supernova progenitors (Smartt 2009; Thompson
characterization of the system via flaring behav- et al. 2009; Smith et al. 2011, see Fig.26) both
ior and stellar age via gyrochronology, the latter through direct imaging and through the detection
helping constrain theories of tidal evolution and of SN precursor events (Ofek et al. 2013), charac-
migration in giant planets. terization of pre-SN variability of SN progenitors
and the frequency of pre-SN outbursts.
• Gravitational microlensing in the Milky Way (see
Han 2008) as well as in the Local Group and be- • The discovery of strongly lensed SNe; 500 − 1000
yond (de Jong et al. 2008). multiply imaged SN Ia (Goldstein & Nugent 2017;
Goldstein et al. 2017) and at least several hun-
• Studies of dwarf novae, including their use as dred strongly lensed core-collapse SNe (Oguri &
probes of stellar populations and structure in the Marshall 2010) are expected to be discovered by
Local Group (Neill & Shara 2005; Shara 2006; LSST. Time delays between the multiple images
Shen & Bildsten 2009). Population studies of the of strongly lensed core-collapse SNe can be used
end points of binary evolution, mapping the distri- to observe the elusive shock breakout phase of the
bution and quantifying the demographics of long light curve, providing an unprecedented look at
and short orbital period dwarf novae, and distin- the earliest emission from these transients (Suwa
guishing recurrent from normal novae. Regular 2018).
cadence, long term color observation on a large
sample of galactic sources will enable the identi- • A large, well characterized sample of super lumi-
fication of CVs containing highly magnetic white nous supernovae including object at redshift as
dwarfs, that are red due to cyclotron emission from high as z = 2.5, a sample large enough to be
leveraged for cosmology improving constraints on • Optical identification of transients and variables
w and Ωm (Scovacricchi et al. 2016). detected in other electromagnetic wavebands,
from gamma rays to radio. Examples include
• Studies of optical bursters (those varying faster optical and gamma ray variability in blazars (Ho-
than 1 mag hr−1 ) to r ∼ 25 mag. vatta et al. 2014), radio transients associated with
tidal disruption flares (Giannios & Metzger 2011),
• Detection and measurement of gamma-ray burst and radio counterparts to supernovae and GRBs
afterglows and transients (e.g., Zhang & Mészáros (Gal-Yam et al. 2006). Deep optical observations
2004; Zhang et al. 2006; Kann et al. 2010) to high with LSST may also help illuminate the nature
redshift (∼7.5). of fast radio bursts (FRBs, Lorimer et al. 2007;
Thornton et al. 2013).
• Large scale studies of stellar tidal disruptions by
nuclear supermassive black holes (e.g., Evans &
Kochanek 1989; Gezari et al. 2008; Strubbe & • Optical identification of counterparts to non-
Quataert 2009; Bloom et al. 2011; Gezari 2012; electromagnetic sources, such as gravitational
Komossa 2015), as well as binary supermassive waves (GW) and neutrino events (LIGO24 , ICE-
black holes in the in-spiral phase (e.g., Cuadra CUBE25 ). LSST’s unique ability to characterize
et al. 2009; Coughlin et al. 2017a). Persistent ob- the faint variable sky over large areas will be
servations leading to complete lightcurves (other important for the detection of GW associated
than the seasonal gaps) of long duration events sources, with an estimate of ∼ 7 discoveries per
like TDEs. Measurements of rates as function of year (Scolnic et al. 2018). The power of the Ad-
galaxy type, redshift, and level of nuclear activity. vanced LIGO (aLIGO)/Virgo26 experiment has
An assessment of the diversity of these events in led to the discovery of four GW events in less
terms of total power, effective temperature, and than a year. The binary neutron star merger
jet launching efficiency. event GW170817 was accompanied by emission
detected across the entire electromagnetic spec-
• A study of quasar variability using accurate, mul- trum (Abbott et al. 2017a). The optical/NIR
ticolor light curves for a few million quasars, lead- emission had two distinct components, a blue
ing to constraints on the accretion physics and emission (which peaked and then faded away on a
nuclear environments (de Vries et al. 2003; Van- time scale of a few days) and a redder component
den Berk et al. 2004; MacLeod et al. 2010; Jiang that persisted for ∼ 15 days. This longer-lasting
et al. 2017). Relations between quasar variabil- component arose from the radioactive decay of
ity properties and luminosity, redshift, rest-frame heavy elements synthesized during the NS merger,
wavelength, time scale, color, radio-jet emission, a “kilonova” (AT 2017gfo). While both these
black-hole mass, and Eddington-normalized lumi- components had been predicted (Metzger 2017),
nosity will be defined with massive statistics, in- the ∼ 100 kilonova sample that LSST is expected
cluding the potential to detect rare but important to generate will enable comparative studies of
events such as jet flares and obscuration events. these transients, allowing us to understand how
Microlensing events will also be monitored in the the presence and relative luminosity of the two
∼4000 gravitationally-lensed quasars discovered components varies to the properties of the binary
by LSST and used to measure the spatial struc- system (e.g., mass) and its remnant. Furthermore,
ture of quasar accretion disks. LSST will be important for identifying the optical
transient corresponding to LIGO events in the first
• The superb continuum light curves of AGN will
place, eliminating false positives (Nissanke et al.
enable economical “piggyback” reverberation-
2013; Metzger & Berger 2012; Cowperthwaite &
mapping efforts using spectroscopy of emission
Berger 2015; Coughlin et al. 2017b). At 24th mag,
lines (e.g., Chelouche & Daniel 2012; Shen et al.
rejecting thousands of false positives from other
2015; Grier et al. 2017). These results will
new transients appearing during the imaging of
greatly broaden the luminosity-redshift plane of
reverberation-mapped AGNs with black-hole mass
estimates. For LSST data alone, the inter-band 24 http://www.ligo.caltech.edu
25
continuum lags will provide useful structural in- http://icecube.wisc.edu
26 http://public.virgo-gw.eu/language/en/
formation.
ments (see Figs. 27 and 28) will be available for about

200 million main-sequence F/G stars which will sample
the halo to distances of 100 kpc (Ivezić et al. 2008; An
et al. 2013) over a solid angle of roughly 20,000 deg2 .
No other existing or planned survey will provide such a
powerful dataset to study the outer halo: Gaia is flux
limited at r = 20, and the Dark Energy Survey (Ros-
setto et al. 2011) and Pan-STARRS both lack observa-
tions in the u band, necessary for estimating metallicity.
The LSST in its standard surveying mode will be able
to detect RR Lyrae variables (pulsating stars and stan-
dard candles) and classical novae (exploding stars and
standard candles) at a distance of 400 kpc and hence
explore the extent and structure of our halo out to half
the distance to the Andromeda galaxy. Thus, the LSST
will enable studies of the distribution of main-sequence
stars beyond the presumed edge of the Galaxy’s halo
(see Fig. 29), of their metallicity distribution through-
Figure 27. The g − r vs. u − g color-color diagram for
about a million point sources from the SDSS Stripe 82 area. out most of the halo, and of their kinematics beyond
Accurate multi-color photometry contains information that the thick disk/halo boundary. LSST will also obtain di-
can be used for source classification and determination of rect distance measurements via trigonometric parallax
detailed stellar properties such as effective temperature and below the hydrogen-burning limit for a representative
metallicity. LSST will enable such measurements for billions thin-disk sample.
of stars. In addition to the study of hydrogen burning stars,
LSST will uncover the largest sample of stellar rem-
the GW event area requires a strategy of multiple nants to date. Over 97% of all stars eventually exhaust
passes in different filters. their fuel and cool to become white dwarfs. Given the
4.4. Mapping the Milky Way age of the Galactic halo, a significant fraction of the
mass in this component may reside in these remnant
The LSST will map the Galaxy in unprecedented de-
stars (e.g., Alcock et al. 2000; Tisserand et al. 2007) and
tail, and by doing so revolutionize the fields of Galactic
therefore their discovery directly constrains the Galactic
Astronomy and Near-field Cosmology. The great detail
mass budget. These large populations of disk and halo
with which the Milky Way can be studied complements
white dwarfs will provide unprecedented constraints on
the statistical power of extra-galactic observations. The
the luminosity function of these stars, which will di-
overarching goal of near-field cosmology is to use spa-
rectly yield independent ages for the Galactic disk and
tial, kinematic, and chemical measurements of stars to
halo (e.g., through the initial-final mass relation, Kalirai
reveal the structure and evolution history of the Milky
et al. (2008)).
Way and its environment. LSST will reveal this fossil
The sky coverage of LSST naturally targets both field
record in great detail and provide a Rosetta Stone for
stars and star clusters. To date, no systematic survey
extragalactic astronomy by setting the context within
of the stellar populations of Southern hemisphere clus-
which we interpret these broader datasets. Moreover,
ters has been performed (e.g., such as the CFHT Open
different candidate supersymmetric particle dark matter
Star Cluster Survey, or the WIYN Open Star Cluster
models predict different mass clustering on small scales,
Survey in the North; Kalirai et al. 2001; Mathieu 2000).
and thus different mass functions for subhalos of the
Multiband imaging of these co-eval, co-spatial, and iso-
Milky Way. Thus the LSST census of faint satellites
metallic systems will provide vital insights into funda-
and stellar streams in the halo will offer a unique means
mental stellar evolution. For example, the depth of
to constrain the particle nature of dark matter.
LSST will enable construction of luminosity and mass
The LSST will produce a massive and exquisitely ac-
functions for nearby open clusters down to the hydro-
curate photometric and astrometric dataset for about
gen burning limit and beyond. Variations in the initial
20 billion Milky Way stars. The coverage of the Galac-
mass function will be studied as a function of environ-
tic plane will yield data for numerous star-forming re-
ment (e.g., age and metallicity). The wide-field coverage
gions, and the y band data will penetrate through the
will also allow us to track how the stellar populations in
interstellar dust layer. Photometric metallicity measure-
Predicted LSST spatial map
c
kp
0
20
20 kpc
SDSS spatial map
Figure 28. The median metallicity map for 2.5 mil- Figure 29. A predicted spatial distribution of stars out
lion main-sequence F-type stars within 10 kpc from the Sun to 150 kpc from the center of the Milky Way, from Bul-
(adapted from Ivezić et al. 2008). The metallicity is esti- lock & Johnston (2005). LSST will resolve main sequence
mated using u − g and g − r colors measured by SDSS. The turnoff stars out to 300 kpc, ten times more volume than
position and size of the mapped region, relative to the rest of shown here, enabling a high-fidelity spatial map over the en-
the Milky Way, is illustrated in the top right corner, where tire observed virial volume. (Note that this is significantly
the same map is scaled and overlaid on an image of the An- larger than the 100 kpc probed by metallicity measurements
dromeda galaxy. The gradient of the median metallicity is in Figure 28, which is limited by the depth of the u-band
essentially parallel to the Z axis, except in the Monoceros observations.) Overlaid on this prediction is the observed
stream region, as marked. LSST will extend this map out to SDSS stellar number density map based on main sequence
100 kpc, using a sample of over 100 million main-sequence F stars with 0.10 < r − i < 0.15 (Jurić et al. 2008). This map
stars. extends up to ∼ 20 kpc from the Sun, with the white box
showing a scale of 20 kpc across and the left side aligned with
each cluster vary as a function of radius, from the core to the Galactic center. The revolutionary Galaxy map provided
beyond the tidal radius. Fainter remnant white dwarfs by SDSS is only complete to ∼40 kpc, or only ∼1% of the
virial volume. However, the outermost reaches of the stel-
will be uncovered in both open and globular clusters (the
lar halo are predicted to bear the most unique signatures of
nearest of which are all in the South), thereby providing our Galaxy’s formation (Johnston et al. 2008; Cooper et al.
a crucial link to the properties of the now evolved stars 2010). LSST will be the only survey capable of fully testing
in each system. such predictions.
In summary, the LSST data will revolutionize studies
of the Milky Way and the entire Local Group. We list • Deep and highly accurate color-magnitude dia-
a few specific Galactic science programs that LSST will grams for over half of the known globular clus-
enable: ters, including tangential velocities from proper
motion measurements (An et al. 2008; Casetti-
• High-resolution studies of the distribution of stars
Dinescu et al. 2007).
in the outer halo in the six-dimensional space
spanned by position, metallicity and proper mo- • Mapping the metallicity, kinematics and spatial
tions (e.g., Girard et al. 2006; Bell et al. 2008; profile of the Saggitarius dwarf tidal stream (e.g.,
Jurić et al. 2008; Ivezić et al. 2008; Bond et al. Ibata et al. 2001; Majewski et al. 2003; Law et al.
2010). 2005; Belokurov et al. 2014) and the Magellanic
stream (Zaritsky et al. 2004).
• The most complete search possible for halo
streams, Galaxy satellites and intra-Local Group • The measurement of the internal motions of Milky
stars (e.g. Belokurov et al. 2007a; Walsh et al. Way dwarf galaxies via proper motions, thereby
2009; Bochanski et al. 2014). constraining their density profiles and possibly the
nature of dark matter (e.g., Walker & Peñarrubia

2011).
• Detailed constraints on the formation and evolu-

tion of the populations within the Galactic Bulge,
as traced by the spatial distribution, motion, and
chemistry of ∼107−8 of its stars (e.g. Hill et al.
2011; Ness et al. 2014).
• Studies of the clumpiness of the gravitational po-

tential in the Galaxy using fragile wide-angle bi-
naries selected with the aid of trigonometric and
photometric parallaxes, and common proper mo-
tion (e.g., Yoo et al. 2004; Longhitano & Binggeli
2010).
• Detailed studies of variable star populations; 2%

or better accurate multicolor light curves will be
available for a sample of at least 50 million variable
stars (Sesar et al. 2007), enabling studies of cat-
aclysmic variables, eclipsing binary systems, and
rare types of variables.
• Discovery of rare and faint high proper motion ob-

jects: probing the faint end of the stellar mass
function (Lépine 2008; Finch et al. 2010), and
searching for free-floating planet candidates (Lu-
cas & Roche 2000; Luhman 2014).
Figure 30. A comparison of an SDSS image (2×4 arcmin2
• Direct measurement of the faint end of the stellar gri composite) showing a typical galaxy at a redshift of ∼0.1
(top) with a similar BV R composite image of the same field
luminosity function using trigonometric parallaxes
obtained by the MUSYC survey (bottom; Gawiser et al.
(Reid et al. 2002) and a complete census of the so- 2006). The MUSYC image is about 4 mag deeper than the
lar neighborhood to a distance of 100 pc based SDSS image (and about 1 mag less deep than the anticipated
on trigonometric parallax measurements for ob- LSST 10-year coadded data). Note the rich surface bright-
jects as faint as Mr = 17 (∼L5 brown dwarfs). ness structure seen in the MUSYC image that is undetectable
For example, LSST will deliver 10% or better in the SDSS image.
distances for a sample of about 2,500 stars with
18< Mr <19. stars (Planck Collaboration et al. 2011; Berry et al.
2012; Green et al. 2014).
• The separation of halo M sub-dwarfs from disk M
dwarfs, using the z − y color which is sensitive to • A census of AGB stars in the Galaxy by searching
their rich molecular band structure (West et al. for resolved envelopes and optical identifications of
2011; Bochanski et al. 2013). IR counterparts (e.g., from the WISE survey), and
by using long-term variability and color selection
• Studies of white dwarfs using samples of several (Ivezić 2007).
million objects, including the determination of
the halo white dwarf luminosity function (SciBook • A complete census of faint populations in nearby
Ch. 6). star forming regions using color and variability se-
lection (e.g. Briceño et al. 2005).
• Measurements of physical properties of stars us-
ing large samples of eclipsing binary stars (Stassun
4.5. Additional Science Projects
et al. 2013).
The experience with any large survey (e.g., SDSS,
• High-resolution three-dimensional studies of inter- 2MASS, VISTA, WISE, GALEX, to name but a few)
stellar dust using 5-color SEDs of main sequence is that much of their most interesting science is often
unrelated to the main science drivers, and is often unan-

ticipated at the time the survey is designed. LSST will
enable far more diverse science than encompassed by the
four themes that drive the system design. We list a few
anticipated major programs.
• Detailed studies of galaxy formation and evolu-
tion using their distribution in luminosity-color-
morphology space as a function of redshift. For ex-
ample, LSST will enable studies of the rest-frame
UV emission, similar to those based on GALEX Figure 31. A comparison of the distribution of galaxies
data for local galaxies, to a redshift of ∼2 for an in luminosity–color–density space measured by SDSS (left)
unprecedentedly large number of galaxies. These and a model based on the Millennium simulation (right).
studies project onto many axes: The linearly-spaced contours outline the distribution of a
volume-limited sample of galaxies in the plotted diagram,
– the evolution of the galaxy luminosity func- and the color-coded background shows the median envi-
tion with redshift, as a function of morphol- ronmental density (computed using the ten nearest neigh-
ogy and color; bors) for galaxies with the corresponding luminosity and
color. Such multi-variate distributions encode rich informa-
– the evolution of the galaxy color distribution tion about formation and evolution of galaxies. Galaxies
over a wide range of rest-frame wavelengths, detected by SDSS are representative of the low-redshift Uni-
and as a function of luminosity and morphol- verse (the median redshift is ∼0.1). The LSST will enable
ogy; such studies as a function of redshift, to z ∼2. Adapted from
Cowan & Ivezić (2008).
– bulge-disk decomposition, as a function of
luminosity and color, over a large redshift
range; • The combination of LSST, Euclid and WFIRST
data should allow discovery of at least tens of
– detailed distribution of satellite galaxies in
quasars at z > 7.5 (R. Barnett 2017, priv. comm).
luminosity-color-morphology space as a func-
tion of luminosity, color, and morphology of • LSST data will provide good constraints on AGN
the primary galaxy; lifetimes, or at least the timescales over which
– correlations of luminosity, color and mor- they make distinct accretion-state transitions
phology with local environment from kpc to (MacLeod et al. 2016), due to large sample size and
Mpc scales, and as a function of redshift (see survey lifetime (e.g. Martini & Schneider 2003).
Figs. 30 and 31);
– the properties of galaxy groups and clusters • The first wide field survey of ultra low surface
as a function of cosmic time. brightness galaxies, with photometric redshift in-
formation. The currently available samples (e.g.
• AGN census to very faint luminosity and a large Greco et al. 2018) are highly incomplete, espe-
redshift limit (Ivezić et al. 2014), yielding 20 mil- cially in the Southern Hemisphere (see Fig. 7 in
lion objects from LSST data alone, and the ability Belokurov et al. 2007a).
to identify up to ∼ 100 million objects once multi-
wavelength data are used to aid AGN selection (see • Search for strong gravitational lenses to a faint sur-
Fig. 32). By reaching substantially further down face brightness limit (e.g. Bartelmann et al. 1998;
the AGN luminosity function than has been possi- Tyson et al. 1998; Belokurov et al. 2007b), which
ble before over a very large solid angle, LSST data can be used to explore the dark matter profiles of
will test evolutionary cosmic downsizing scenar- galaxies (e.g., Treu et al. 2006).
ios across the full range of cosmic environments,
and lead to a much clearer understanding of black-
hole growth during the first Gyr. For example, 4.5.1. Synergy with other projects
LSST should discover several thousand AGNs at LSST will not operate in isolation and will greatly
z ∼ 6 − 7.5, representing a dramatic increase over benefit from other precursor and coeval data at a vari-
present samples (Brandt & LSST Active Galax- ety of wavelengths, depths, and timescales. For exam-
ies Science Collaboration 2007, see also SciBook ple, in the visual wavelength range, most of the Celes-
Ch. 10). tial Sphere will be covered to a limit several magnitudes
Dark Energy Survey is scanning ∼5000 deg2 a magni-

tude deeper still in the southern sky. Despite the lack of
the u band data and its relatively shallow imaging, the
Pan-STARRS surveys represents a valuable complement
to LSST in providing Northern sky coverage to a limit
fainter than that of SDSS and SkyMapper. LSST and
Gaia will be highly complementary datasets for study-
ing the Milky Way in the multi-dimensional space of
three-dimensional positions, proper motions and metal-
licity (Ivezić et al. 2012). The Gaia survey will provide
calibration checks at the bright end for proper motions
and trigonometric parallax measurements by LSST, and
LSST will extend the Gaia survey by four magnitudes.
The upcoming Zwicky Transient Facility (e.g., Laher
et al. 2017), with its 600 Megapixel camera and a 47
deg2 large field of view, will generate the largest opti-
cal transient stream prior to LSST (at about one tenth
of LSST rate) and thus provide an early insight into
astrophysical surprises and technical challenges await-
ing LSST. The LSST data stream will invigorate subse-
quent investigations by numerous other telescopes that
will provide additional temporal, spectral and spatial
resolution coverage.
WFIRST and Euclid will carry out wide-field imaging
surveys in the near-infrared, giving highly complemen-
tary photometry to LSST. The resulting galaxy SEDs
should give rise to even better photometric redshifts, as
well as tighter constraints on stellar masses and star
formation histories crucial for galaxy evolution stud-
ies. The weak lensing analyses from space and from
the ground will also be highly complementary, and will
provide crucial cross-checks of one another. LSST also
presents the opportunity to conduct simultaneous obser-
Figure 32. The LSST will deliver AGN sky densities of vations of WFIRST’s Galactic Bulge survey fields, from
1000–4000 deg−2 (top panel); The total LSST AGN yield, which it will be possible to measure the parallax and
selected using colors and variability, should be well over 10 hence the lens masses for most microlensing events, as
million objects, especially once multiwavelength data are also well as providing valuable lightcurve coverage during the
utilized. The bottom panel shows the expected distribution
large data gaps between WFIRST survey ‘seasons’.
of these objects in the absolute magnitude vs. redshift plane,
color-coded by the probability for an object to be detected as
LSST will also enable multi-wavelength studies of faint
variable after 1 year of observations. Note that quasars will optical sources using gamma-ray, X-ray, IR and radio
be detected to their formal luminosity cutoff (M < −23) even data. For example, the SDSS detected only 1/3 of all
at redshifts of ∼5. Adapted from Brandt & LSST Active 20cm FIRST sources (Becker et al. 1995) because it was
Galaxies Science Collaboration (2007). too shallow by ∼4 mag for a complete optical identi-
fication. Similarly, deep optical data are required for
fainter than LSST saturation (r ∼ 16), first by the com- identification of faint X-ray sources (Brandt & Hasinger
bination of SDSS, Pan-STARRS1 (PS1), the Dark En- 2005; Brandt & Vito 2017).
ergy Survey (Dark Energy Survey Collaboration et al. LSST will provide a crucial complementary capabil-
2016) and SkyMapper (Keller et al. 2007), and then by ity to space experiments operating in other wavebands,
the Gaia survey. The SkyMapper survey will obtain such as the NuSTAR Mission (Harrison et al. 2013),
imaging data in the southern sky to similar depths as eROSITA (Merloni et al. 2012), and the Fermi Gamma-
SDSS, the PS1 surveys provides multi-epoch data some- ray Space Telescope (e.g., Atwood et al. 2009). The
what deeper than SDSS in the northern sky, and the Laser Interferometer Gravitational Wave Observatory
(LIGO) has now detected ultracompact binaries and publicly available astronomical data sets. The SDSS
black-hole mergers through the gravitational wave out- has published a series of large incremental data releases
bursts that are emitted during the final stages of such via a sophisticated database, roughly once per year, to-
events. LSST will aid studies of the electromagnetic gether with a paper describing the content of each data
signal that accompanies the gravitational wave emis- release, and extensive on-line documentation giving in-
sion, thereby providing an accurate position on the sky structions on downloading the catalogs and image data
for the system, which is crucial for subsequent obser- (see http://www.sdss.org). The overwhelming majority
vations. LSST will also add new value to the archives of the almost 8000 refereed papers based on SDSS data
for billion-dollar class space missions such as Chandra, to date have been written by scientists from outside the
XMM-Newton, Spitzer, Herschel, Euclid, and WFIRST, project, and include many of the most high-profile re-
because deep optical multi-color data will enable mas- sults that have come from the survey.
sive photometric studies of sources from these missions. Nevertheless, it is clear that many of the highest pri-
All areas of the sky – whether by design or by serendip- ority LSST science investigations will require organized
ity – in which past, present, or future multiwavelength teams of professionals working together to optimize sci-
surveys overlap with LSST sky coverage, will be fur- ence analyses and to assess the importance of system-
ther promoted by LSST investigations to become “opti- atic uncertainties on the derived results. To meet this
cal plus multiwavelength Selected Areas”. Last but not need, a number of science collaborations have been es-
least, the huge samples of various astronomical source tablished in core science areas. For example, the LSST
populations will yield extremely rare objects for inves- Dark Energy Science Collaboration includes members
tigations by powerful facilities such as JWST (Gardner with interests in the study of dark energy and related
et al. 2006) and the next generation of 20–40 meter tele- topics in fundamental physics with LSST data. As of the
scopes. time of this contribution, there are over 800 participants
In summary, the diversity of science enabled by LSST in these collaborations. The science collaborations are
will be unparalleled, extending from the physics of grav- listed on the LSST web page, together with a description
ity and the early Universe to the properties of “killer” of the application process for each one. All those at US
asteroids. While there are other projects that aim to and Chilean institutions, as well as named individuals
address some of the same science goals, no other project from institutions in other countries which have signed
matches this diversity and LSST’s potential impact on Memoranda of Agreement to contribute to LSST opera-
society in general. tions costs are eligible to apply. As described in §§ 2.6.3
and 3.3, LSST will make available substantial computa-
5. COMMUNITY INVOLVEMENT tional resources to the science community to carry out
their analyses; the system has been sized accordingly.
LSST has been conceived as a public facility: the
As we design our observing strategies, we are actively
database that it will produce, and the associated ob-
seeking and implementing input by the LSST science
ject catalogs that are generated from that database,
community. The LSST science collaborations in partic-
will be made available with no proprietary period to
ular have helped develop the LSST science case and con-
the U.S. and Chilean scientific communities, as well as
tinue to provide advice on how to optimize their science
to those international partners who contribute to oper-
with choices in cadence, software, and data systems. A
ations funding. As described in § 6, data will also be
recent example is the publication of a document enti-
made available to the general public for educational and
tled “Science-Driven Optimization of the LSST Observ-
outreach activities. The LSST data management sys-
ing Strategy” LSST Science Collaboration et al. (2017),
tem (§ 3.3) will provide user-friendly tools to access this
a living document that quantifies the science returns in
database, support user-initiated queries and data explo-
different areas for different observing cadence. The ca-
ration, and carry out scientific analyses on the data, us-
dence will continue to be refined, with input from the
ing LSST computers either at the archive facility or at
science collaborations, during the commissioning, and
the data access centers. We expect that many, perhaps
the observing strategy will be regularly reviewed, with
even the majority, of LSST discoveries will come from
flexibility built in, during operations.
research astronomers with no formal affiliation to the
The Science Advisory Committee (SAC), chaired by
project, from students, and from interested amateurs,
Michael Strauss, provides a formal, and two-way, con-
intrigued by the accessibility to the Universe that this
nection to the external science community served by
facility uniquely provides.
LSST. This committee takes responsibility for policy
The SDSS provides a good example for how the scien-
questions facing the project and also deals with tech-
tific community can be effective in working with large,
nical topics of interest to both the science community evant science content. Science notebooks will also ac-
and the LSST Project. The SAC minutes and notes are commodate access to LSST data for lifelong learners and
available publicly. Current members on this committee anyone that visits the portal.
are: T. Anguita (Andrés Bello, Chile), R. Bean (Cor- Anyone around the world will be able to participate
nell), W.N. Brandt (Penn State), J. Kalirai (STScI), in a variety of citizen science projects that use LSST
M. Kasliwal (Caltech), D. Kirkby (UC Irvine), C. Liu data. The EPO Team will work with the Zooniverse to
(Staten Island), A. Mainzer (JPL), R. Malhotra (U develop the Project Builder to include tools specifically
Arizona), N. Padilla (U. Católica de Chile), J. Simon designed to utilize LSST data, allowing LSST principal
(Carnegie), A. Slosar (Brookhaven), M. Strauss (Prince- investigators to create any number of projects to help
ton), L. Walkowicz (Adler), and R. Wechsler (Stanford). them accomplish their science goals. EPO anticipates
that the number of citizen science projects in the as-
6. EDUCATIONAL AND SOCIETAL IMPACTS tronomy field will increase dramatically when LSST is
operational, giving a whole new generation of citizen sci-
The impact and enduring societal significance of
entists the opportunity to deepen their engagement with
LSST will exceed its direct contributions to advances in
astronomy using authentic data from LSST.
physics and astronomy. LSST is uniquely positioned to
LSST EPO will produce and maintain a digital library
have high impact with the interested public, planetar-
of multimedia assets including images, video clips, and
iums and science centers, and citizen science projects,
3D models. Multimedia assets will be aligned to stan-
as well as middle school through university educational
dards such as IMERSA Dome Master and Astronomy
programs. LSST will contribute to the national goals
Visualization Metadata, when applicable, allowing full
of enhancing science literacy and increasing the global
flexibility for adoption by content creators at planetari-
competitiveness of the US science and technology work-
ums and science centers. We will also follow the Inter-
force. Engaging the public in LSST activities has been
national Planetarium Society’s Data2Dome standard, to
part of the project design from the beginning.
maximize the number of platforms that can use our as-
The mission of LSST’s Education and Public Out-
sets.
reach (EPO) program is to provide worldwide access to a
The LSST EPO program will rely on a cloud-based
subset of LSST data through accessible and engaging on-
EPO Data Center (EDC) to handle the unique needs of
line experiences so anyone can explore the universe and
the EPO audiences. These needs include, for example,
be part of the discovery process. To do this, LSST EPO
a fast and smooth browsing experience on mobile de-
will facilitate a pathway from entry-level exploration of
vices, and the need to handle inevitable spikes and lulls
astronomical imagery and information to more sophisti-
in visitor traffic and data transfers. As such, the EDC
cated interaction with LSST data using tools similar to
will follow best practices popularized by cloud comput-
what professional astronomers use for their work.
ing, leveraging on-demand computing and auto-scalable
A dynamic, immersive web portal will enable mem-
architecture. Remaining agile and relevant during the
bers of the public to explore color images of the full
full lifetime of Operations by adjusting to technology
LSST sky, examine objects in more detail, view events
trends and education priorities is an important part of
from the nightly alert stream, learn more about LSST
the LSST EPO design process.
science topics and discoveries, and investigate scientific
LSST EPO is committed to engaging with diverse au-
questions that excite them using real LSST data in on-
diences and is undertaking a multi-faceted approach to
line science notebooks. The portal will also link to nu-
reaching diverse individuals. LSST EPO is planning
merous citizen science projects using LSST data.
to partner with at least five organizations serving 1)
LSST data can become a key part of classrooms em-
women/girls, 2) individuals from traditionally underrep-
phasizing student-centered research in middle school,
resented groups in STEM, and 3) individuals in low so-
high school, and undergraduate settings. Using online
cioeconomic communities. Representatives from these
science notebooks, teachers will be able to bring real
organizations will be key stakeholders of the EPO pro-
LSST telescope data into their classrooms without hav-
gram, helping to shape deliverables and a culturally re-
ing to download, install, and maintain software locally.
sponsive program evaluation. Furthermore, these rela-
Educational investigations will be designed to support
tionships will allow for co-creation of EPO deliverables
key aspects of the Next-Generation Science Standards
to help ensure materials are accessible to, of interest to,
(NGSS) in the USA, and goals of the Explora pro-
and relevant to diverse populations.
gram through CONICYT in Chile. Educators will be
LSST EPO is breaking new ground in bringing astro-
supported through professional development that offers
nomical data to the public in a timely, engaging, and
training on the online notebook technology and also rel-
easy-to-use way. It is not unreasonable to anticipate in Astronomy (AURA) has formal responsibility for the
tens of millions of public users browsing the LSST sky, LSST project since 2011. At this writing, the project is
exploring discoveries as they are broadcast, and mon- near the peak of the construction effort, and is preparing
itoring objects of interest. Results of EPO’s ongoing for the transition to commissioning and operations.
evaluation will be made publicly available, allowing us The construction cost of LSST is being borne by the
to share lessons learned, insights, and program impacts US National Science Foundation, the Department of En-
with the larger science EPO community. ergy, generous contributions from several private foun-
dations and institutions, and the member institutions of
7. SUMMARY AND CONCLUSIONS the LSST Corporation. The LSST budget includes a sig-
nificant Education and Public Outreach program (§ 6).
Until recently, most astronomical investigations have
The U.S. Department of Energy is supporting the cost
focused on small samples of cosmic sources or individ-
of constructing the camera. LSST has high visibility in
ual objects. Over the past few decades, however, ad-
the high-energy physics community, both at universities
vances in technology have made it possible to move be-
and government laboratories. The telescope will see first
yond the traditional observational paradigm and to un-
light with a commissioning camera in late 2019, and the
dertake large-scale sky surveys, such as SDSS, 2MASS,
project is scheduled to begin regular survey operations
GALEX, Gaia, and others. This observational progress,
by 2022.
based on a synergy of advances in telescope construc-
The LSST survey will open a movie-like window on
tion, detectors, and above all, information technology,
objects that change brightness, or move, on timescales
has had a dramatic impact on nearly all fields of astron-
ranging from 10 seconds to 10 years. The survey will
omy, many areas of fundamental physics, and society in
have a raw data rate of about 15 TB per night (about
general. The LSST builds on the experience of these sur-
the same as one complete Sloan Digital Sky Survey per
veys and addresses the broad goals stated in several na-
night), and will collect about 60 PB of data over its life-
tionally endorsed reports by the U.S. National Academy
time, resulting in an incredibly rich and extensive public
of Sciences. The 2010 report “New Worlds, New Hori-
archive that will be a treasure trove for breakthroughs in
zons in Astronomy and Astrophysics” by the Commit-
many areas of astronomy and physics. About 20 billion
tee for a Decadal Survey of Astronomy and Astrophysics
galaxies and a similar number of stars will be detected
ranked LSST as its top priority for large ground-based
– for the first time in history, the number of cataloged
programs. The LSST will be unique: the combination
celestial objects will exceed the number of living people!
of large aperture and large field of view, coupled with
About a thousand observations of each position across
the needed computation power and database technology,
half of the Celestial Sphere will represent the greatest
will enable simultaneously fast and wide and deep imag-
movie of all time.
ing of the sky, addressing in one sky survey the broad
Alerts of transient, variable, and moving objects will
scientific community’s needs in both the time domain
be issued worldwide within 60 seconds of detection. An
and deep universe.
extensive public outreach program will provide a new
The realization of the LSST involves extraordinary en-
view of the sky to curious minds of all ages worldwide.
gineering and technological challenges: the fabrication
We are working with prospective foreign partners to
of large, high-precision optics; construction of a huge,
make all of the LSST science data more broadly available
highly-integrated array of sensitive, wide-band imaging
worldwide. As of 2017, 34 institutions from 23 countries
sensors; and the operation of a data management facility
have signed Memoranda of Agreement to contribute sig-
handling tens of terabytes of data each day. The design,
nificantly to the LSST operating costs, in exchange for
development and construction effort has been underway
participation in the science collaborations and data ac-
since 2006 and will continue through the onset of full
cess. The software which processes the pixels and cre-
survey operations. This work involves hundreds of per-
ates the LSST database is open source. LSST will be
sonnel at institutions all over the US, Chile, and the rest
a significant milestone in the globalization of the infor-
of the world.
mation revolution. The vast relational database of 32
In December 2013, LSST passed the NSF Final De-
trillion observations of 40 billion objects will be mined
sign Review for construction, and in May 2014 the Na-
for the unexpected and used for precision experiments
tional Science Board approved the project. The pri-
in astrophysics. LSST will be in some sense an inter-
mary/tertiary mirror was cast in 2008, and the polished
net telescope: the ultimate network peripheral device
mirror was completed in 2015. In 2014 LSST transi-
to explore the Universe, and a shared resource for all
tioned from the design and development phase to con-
humanity.
struction, and the Associated Universities for Research
This material is based upon work supported in part Laboratory. Additional LSST funding comes from pri-
by the National Science Foundation through Coopera- vate donations, grants to universities, and in-kind sup-
tive Agreement 1258333 managed by the Association of port from LSSTC Institutional Members.
Universities for Research in Astronomy (AURA), and
Facility: LSST
the Department of Energy under Contract No. DE-
AC02-76SF00515 with the SLAC National Accelerator
APPENDIX
A. VERSION HISTORY
Version 1.0 (May 15, 2008): the first posting.

Version 2.0 (June 7, 2011): acknowledged the Decadal Survey 2010 report; updated construction schedule; updated
expected performance in Table 2; added sections on LSST simulations and data mining; updated several figures;
updated references; expanded author list.
Version 3.0 (August 26, 2014): acknowledged the start of federal construction; updated system description and
science examples, updated several figures; refreshed references; expanded author list.
Version 4.0 (May 15, 2018): updated system description and science examples, updated expected performance in
Table 2; updated several figures; refreshed references; expanded author list.
REFERENCES
Abazajian, K. N., Adelman-McCarthy, J. K., Agüeros, Anderson, G. P., Berk, A., Acharya, P. K., et al. 1999, in
M. A., et al. 2009, ApJS, 182, 543, Proc. SPIE, Vol. 3866, Optics in Atmospheric
doi: 10.1088/0067-0049/182/2/543 Propagation and Adaptive Systems III, ed. A. Kohnle &
Abazajian, K. N., Adshead, P., Ahmed, Z., et al. 2016, J. D. Gonglewski, 2–10
ArXiv e-prints. https://arxiv.org/abs/1610.02743 Anderson, S. F., Haggard, D., Homer, L., et al. 2005, AJ,
Abbott, B. P., Abbott, R., Abbott, T. D., et al. 2017a, 130, 2230, doi: 10.1086/491587
ApJL, 848, L12, doi: 10.3847/2041-8213/aa91c9 Angel, R., Lesser, M., Sarlot, R., & Dunham, E. 2000, in
—. 2017b, Nature, 551, 85, doi: 10.1038/nature24471 ASP Conf. Ser., Vol. 195, Imaging the Universe in Three
A’Hearn, M. F. 2004, in Comets II, ed. M. C. Festou, H. U. Dimensions, ed. W. van Breugel & J. Bland-Hawthorn, 81
Keller, & H. A. Weaver (University of Arizona Press), Angeli, G. Z., Xin, B., Claver, C., et al. 2016, in
17–22 Proc. SPIE, Vol. 9911, Modeling, Systems Engineering,
A’Hearn, M. F., Millis, R. C., Schleicher, D. O., Osip, D. J., and Project Management for Astronomy VI, 991118
& Birch, P. V. 1995, Icarus, 118, 223, Antilogus, P., Bailly, P., Barrillon, P., et al. 2017, Journal
doi: 10.1006/icar.1995.1190 of Instrumentation, 12, C03017
Aihara, H., Arimoto, N., Armstrong, R., et al. 2018, PASJ, Antoniou, I., & Perivolaropoulos, L. 2010, JCAP, 12, 012,
70, S4, doi: 10.1093/pasj/psx066 doi: 10.1088/1475-7516/2010/12/012
Albrecht, A., Bernstein, G., Cahn, R., et al. 2006, ArXiv Araujo-Hauck, C., Sebag, J., Liang, M., et al. 2016, in
Astrophysics e-prints. Proc. SPIE, Vol. 9906, Ground-based and Airborne
https://arxiv.org/abs/astro-ph/0609591 Telescopes VI, 99060L
Alcock, C., Allsman, R. A., Alves, D. R., et al. 2000, ApJ, Arcavi, I., Howell, D. A., Kasen, D., et al. 2017, Nature,
542, 281, doi: 10.1086/309512 551, 210, doi: 10.1038/nature24030
—. 2001, ApJL, 550, L169, doi: 10.1086/319636 Arnett, W. D., & Meakin, C. 2011, ApJ, 741, 33,
Amôres, E. B., & Lépine, J. R. D. 2005, AJ, 130, 659, doi: 10.1088/0004-637X/741/1/33
doi: 10.1086/430957 Astropy Collaboration, Robitaille, T. P., Tollerud, E. J.,
An, D., Johnson, J. A., Clem, J. L., et al. 2008, ApJS, 179, et al. 2013, A&A, 558, A33,
326, doi: 10.1086/592090 doi: 10.1051/0004-6361/201322068
An, D., Beers, T. C., Johnson, J. A., et al. 2013, ApJ, 763, Atwood, W. B., Abdo, A. A., Ackermann, M., et al. 2009,
65, doi: 10.1088/0004-637X/763/1/65 ApJ, 697, 1071, doi: 10.1088/0004-637X/697/2/1071
Bacci, P., Maestripieri, M., Tesi, L., et al. 2017, Minor Bhattacharya, S., Kosowsky, A., Newman, J. A., &
Planet Electronic Circulars, 2017-U181 Zentner, A. R. 2011, PhRvD, 83, 043004,
Barnard, M., Abrahamse, A., Albrecht, A., Bozek, B., & doi: 10.1103/PhysRevD.83.043004
Yashar, M. 2008, PhRvD, 78, 043528, Bianco, F. B., Modjaz, M., Hicken, M., et al. 2014, ApJS,
doi: 10.1103/PhysRevD.78.043528 213, 19, doi: 10.1088/0067-0049/213/2/19
Bartelmann, M., Huss, A., Colberg, J. M., Jenkins, A., & Blake, C., & Glazebrook, K. 2003, ApJ, 594, 665,
Pearce, F. R. 1998, A&A, 330, 1 doi: 10.1086/376983
Batista, V., Beaulieu, J.-P., Gould, A., et al. 2014, ApJ, Blandford, R. D., & LSST Strong Lensing Science
780, 54, doi: 10.1088/0004-637X/780/1/54 Collaboration. 2007, in Bulletin of the American
Batygin, K., & Brown, M. E. 2016, AJ, 151, 22, Astronomical Society, Vol. 39, American Astronomical
doi: 10.3847/0004-6256/151/2/22 Society Meeting Abstracts, 979
Bayliss, D. D. R., & Sackett, P. D. 2011, ApJ, 743, 103, Bloom, J. S., Starr, D. L., Butler, N. R., et al. 2008,
doi: 10.1088/0004-637X/743/2/103 Astronomische Nachrichten, 329, 284,
Beaulieu, J.-P., Bennett, D. P., Fouqué, P., et al. 2006, doi: 10.1002/asna.200710957
Nature, 439, 437, doi: 10.1038/nature04441 Bloom, J. S., Giannios, D., Metzger, B. D., et al. 2011,
Becker, A. C., Wittman, D. M., Boeshaar, P. C., et al. Science, 333, 203, doi: 10.1126/science.1207150
2004, ApJ, 611, 418, doi: 10.1086/421994 Bochanski, J. J., Savcheva, A., West, A. A., & Hawley,
Becker, R. H., White, R. L., & Helfand, D. J. 1995, ApJ, S. L. 2013, AJ, 145, 40, doi: 10.1088/0004-6256/145/2/40
450, 559, doi: 10.1086/176166 Bochanski, J. J., West, A. A., Hawley, S. L., & Covey,
Becla, J., & Wang, D. L. 2014, in Exascale Radio K. R. 2007, AJ, 133, 531, doi: 10.1086/510240
Astronomy, Vol. 2 Bochanski, J. J., Willman, B., West, A. A., Strader, J., &
Becla, J., Wang, D., Monkewitz, S., et al. 2017, LSST Data Chomiuk, L. 2014, AJ, 147, 76,
Management Database Design, LSST Document doi: 10.1088/0004-6256/147/4/76
LDM-135, LSST. https://ls.st/LDM-135 Bockelée-Morvan, D., & Biver, N. 2017, Philosophical
Bell, E. F., Zucker, D. B., Belokurov, V., et al. 2008, ApJ, Transactions of the Royal Society of London Series A,
680, 295, doi: 10.1086/588032 375, 20160252, doi: 10.1098/rsta.2016.0252
Bellm, E. 2014, in The Third Hot-wiring the Transient Bond, N. A., Ivezić, Ž., Sesar, B., et al. 2010, ApJ, 716, 1,
Universe Workshop, ed. P. R. Wozniak, M. J. Graham, doi: 10.1088/0004-637X/716/1/1
A. A. Mahabal, & R. Seaman, 27–33 Bongiorno, A., Zamorani, G., Gavignaud, I., et al. 2007,
Belokurov, V., & Koposov, S. E. 2016, MNRAS, 456, 602, A&A, 472, 443, doi: 10.1051/0004-6361:20077611
doi: 10.1093/mnras/stv2688 Bonvin, V., Courbin, F., Suyu, S. H., et al. 2017, MNRAS,
Belokurov, V., Zucker, D. B., Evans, N. W., et al. 2007a, 465, 4914, doi: 10.1093/mnras/stw3006
ApJ, 654, 897, doi: 10.1086/509718 Bosch, J., Armstrong, R., Bickerton, S., et al. 2018, PASJ,
Belokurov, V., Evans, N. W., Moiseev, A., et al. 2007b, 70, S5, doi: 10.1093/pasj/psx080
ApJL, 671, L9, doi: 10.1086/524948 Bottke, W. F., Vokrouhlický, D., Broz, M., Nesvorný, D., &
Belokurov, V., Koposov, S. E., Evans, N. W., et al. 2014, Morbidelli, A. 2001, Science, 294, 1693,
MNRAS, 437, 116, doi: 10.1093/mnras/stt1862 doi: 10.1126/science.1066760
Benson, B. A., Ade, P. A. R., Ahmed, Z., et al. 2014, in Brandt, N., & LSST Active Galaxies Science Collaboration.
Proc. SPIE, Vol. 9153, Millimeter, Submillimeter, and 2007, in Bulletin of the American Astronomical Society,
Far-Infrared Detectors and Instrumentation for Vol. 39, American Astronomical Society Meeting
Astronomy VII, 91531P Abstracts, 980
Bergeron, P., Wesemael, F., & Beauchamp, A. 1995, PASP, Brandt, W. N., & Hasinger, G. 2005, ARA&A, 43, 827,
107, 1047, doi: 10.1086/133661 doi: 10.1146/annurev.astro.43.051804.102213
Bernstein, G. M., Trilling, D. E., Allen, R. L., et al. 2004, Brandt, W. N., & Vito, F. 2017, Astronomische
AJ, 128, 1364, doi: 10.1086/422919 Nachrichten, 338, 241, doi: 10.1002/asna.201713337
Bernstein, G. M., Armstrong, R., Plazas, A. A., et al. 2017, Briceño, C., Calvet, N., Hernández, J., et al. 2005, AJ, 129,
PASP, 129, 074503, doi: 10.1088/1538-3873/aa6c55 907, doi: 10.1086/426911
Berry, M., Ivezić, Ž., Sesar, B., et al. 2012, ApJ, 757, 166, Brown, M. E. 2017, AJ, 154, 65,
doi: 10.1088/0004-637X/757/2/166 doi: 10.3847/1538-3881/aa79f4
Brown, M. E., & Batygin, K. 2016, ApJL, 824, L23, Cooray, A., Hu, W., Huterer, D., & Joffre, M. 2001, ApJL,
doi: 10.3847/2041-8205/824/2/L23 557, L7, doi: 10.1086/323323
Brown, M. E., Trujillo, C., & Rabinowitz, D. 2004, ApJ, Cooray, A. R. 1999, A&A, 348, 31
617, 645, doi: 10.1086/422095 Coughlin, E. R., Armitage, P. J., Nixon, C., & Begelman,
Bruzual, G., & Charlot, S. 2003, MNRAS, 344, 1000, M. C. 2017a, MNRAS, 465, 3840,
doi: 10.1046/j.1365-8711.2003.06897.x doi: 10.1093/mnras/stw2913
Budavári, T., Connolly, A. J., Szalay, A. S., et al. 2003, Coughlin, M., Dietrich, T., Kawaguchi, K., et al. 2017b,
ApJ, 595, 59, doi: 10.1086/377168 ApJ, 849, 12, doi: 10.3847/1538-4357/aa9114
Bullock, J. S., & Johnston, K. V. 2005, ApJ, 635, 931, Cowan, N. B., & Ivezić, Ž. 2008, ApJL, 674, L13,
doi: 10.1086/497422 doi: 10.1086/528986
Burke, D. L., Axelrod, T., Blondin, S., et al. 2010, ApJ, Cowperthwaite, P. S., & Berger, E. 2015, ApJ, 814, 25,
720, 811, doi: 10.1088/0004-637X/720/1/811 doi: 10.1088/0004-637X/814/1/25
Burke, D. L., Rykoff, E. S., Allam, S., et al. 2018, AJ, 155, Cuadra, J., Armitage, P. J., Alexander, R. D., & Begelman,
doi: 10.3847/1538-3881/aa9f22 M. C. 2009, MNRAS, 393, 1423,
Burrows, A., Sudarsky, D., & Hubeny, I. 2006, ApJ, 640, doi: 10.1111/j.1365-2966.2008.14147.x
1063, doi: 10.1086/500293
Cuillandre, J.-C. J., Withington, K., Hudelot, P., et al.
Cai, R.-G., Ma, Y.-Z., Tang, B., & Tuo, Z.-L. 2013, 2012, in Proc. SPIE, Vol. 8448, Observatory Operations:
PhRvD, 87, 123522, doi: 10.1103/PhysRevD.87.123522
Strategies, Processes, and Systems IV, 84480M
Campanelli, L., Cea, P., Fogli, G. L., & Marrone, A. 2011,
Cushing, M. C., Rayner, J. T., & Vacca, W. D. 2005, ApJ,
623, 1115, doi: 10.1086/428040
Carollo, D., Beers, T. C., Lee, Y. S., et al. 2007, Nature,
Dai, D.-C., Kinney, W. H., & Stojkovic, D. 2011, JCAP, 4,
450, 1020, doi: 10.1038/nature06460
015, doi: 10.1088/1475-7516/2011/04/015
Casetti-Dinescu, D. I., Girard, T. M., Herrera, D., et al.
Dalal, N., Doré, O., Huterer, D., & Shirokov, A. 2008,
2007, AJ, 134, 195, doi: 10.1086/518507
Chambers, K. C., Magnier, E. A., Metcalfe, N., et al. 2016,
Dalal, N., & Kochanek, C. S. 2002, ApJ, 572, 25,
ArXiv e-prints. https://arxiv.org/abs/1612.05560
doi: 10.1086/340303
Chang, C., Jarvis, M., Jain, B., et al. 2013, MNRAS, 434,
Dandy, C. L., Fitzsimmons, A., & Collander-Brown, S. J.
2121, doi: 10.1093/mnras/stt1156
2003, Icarus, 163, 363,
—. 2015, MNRAS, 447, 1746, doi: 10.1093/mnras/stu2553
doi: 10.1016/S0019-1035(03)00087-3
Chapline, G. F., & Frampton, P. H. 2016, JCAP, 11, 042,
Dark Energy Survey Collaboration, Abbott, T., Abdalla,
doi: 10.1088/1475-7516/2016/11/042
F. B., et al. 2016, MNRAS, 460, 1270,
Chelouche, D., & Daniel, E. 2012, ApJ, 747, 62,
doi: 10.1093/mnras/stw641
doi: 10.1088/0004-637X/747/1/62
Davis, C., Gatti, M., Vielzeuf, P., et al. 2017, ArXiv
Choi, J.-Y., Han, C., Udalski, A., et al. 2013, ApJ, 768,
129, doi: 10.1088/0004-637X/768/2/129 e-prints. https://arxiv.org/abs/1710.02517
Chou, M.-Y., Majewski, S. R., Cunha, K., et al. 2007, ApJ, Dawson, W. A., Wittman, D., Jee, M. J., et al. 2012, ApJL,
670, 346, doi: 10.1086/522483 747, L42, doi: 10.1088/2041-8205/747/2/L42
Coe, D., & Moustakas, L. A. 2009, ApJ, 706, 45, De Bernardis, F., Stevens, J. R., Hasselfield, M., et al.
doi: 10.1088/0004-637X/706/1/45 2016, in Proc. SPIE, Vol. 9910, Observatory Operations:
Colin, J., Mohayaee, R., Sarkar, S., & Shafieloo, A. 2011, Strategies, Processes, and Systems VI, 991014
MNRAS, 414, 264, doi: 10.1111/j.1365-2966.2011.18402.x de Bruijne, J. H. J. 2012, Ap&SS, 341, 31,
Connolly, A. J., Angeli, G. Z., Chandrasekharan, S., et al. doi: 10.1007/s10509-012-1019-4
2014, in Proc. SPIE, Vol. 9150, Modeling, Systems de Jong, J. T. A., Kuijken, K. H., & Héraudeau, P. 2008,
Engineering, and Project Management for Astronomy VI, A&A, 478, 755, doi: 10.1051/0004-6361:20078378
915014 De Lucia, G., Springel, V., White, S. D. M., Croton, D., &
Cook, N. V., Ragozzine, D., Granvik, M., & Stephens, D. C. Kauffmann, G. 2006, MNRAS, 366, 499,
2016, ApJ, 825, 51, doi: 10.3847/0004-637X/825/1/51 doi: 10.1111/j.1365-2966.2005.09879.x
Cooper, A. P., Cole, S., Frenk, C. S., et al. 2010, MNRAS, de Vries, W. H., Becker, R. H., & White, R. L. 2003, AJ,
406, 744, doi: 10.1111/j.1365-2966.2010.16740.x 126, 1217, doi: 10.1086/377486
Delgado, F., Saha, A., Chandrasekharan, S., et al. 2014, in Fitzpatrick, M. J., Graham, M. J., Mighell, K. J., et al.
Proc. SPIE, Vol. 9150, Modeling, Systems Engineering, 2016, in Proc. SPIE, Vol. 9913, Software and
and Project Management for Astronomy VI, 915015 Cyberinfrastructure for Astronomy IV, 99130L
della Valle, M., & Livio, M. 1995, ApJ, 452, 704, Foley, R. J. 2012, ApJ, 748, 127,
doi: 10.1086/176342 doi: 10.1088/0004-637X/748/2/127
DeMeo, F. E., Binzel, R. P., Slivan, S. M., & Bus, S. J. Fukugita, M., Ichikawa, T., Gunn, J. E., et al. 1996, AJ,
2009, Icarus, 202, 160, doi: 10.1016/j.icarus.2009.02.005 111, 1748, doi: 10.1086/117915
Denneau, L., Jedicke, R., Grav, T., et al. 2013, PASP, 125, Gaia Collaboration, Brown, A. G. A., Vallenari, A., et al.
357, doi: 10.1086/670337 2016, A&A, 595, A2, doi: 10.1051/0004-6361/201629512
DES Collaboration, Abbott, T. M. C., Abdalla, F. B., et al. Gal-Yam, A., Ofek, E. O., Poznanski, D., et al. 2006, ApJ,
2017, ArXiv e-prints. https://arxiv.org/abs/1708.01530 639, 331, doi: 10.1086/499157
Djorgovski, S. G., Mahabal, A., Drake, A., Graham, M., & Gardner, J. P., Mather, J. C., Clampin, M., et al. 2006,
Donalek, C. 2013, in Planets, Stars and Stellar SSRv, 123, 485, doi: 10.1007/s11214-006-8315-7
Systems. Volume 2: Astronomical Techniques, Software Garnavich, P. M., Smith, R. C., Miknaitis, G., et al. 2004,
and Data, ed. T. D. Oswalt & H. E. Bond (Springer), 223 in Bulletin of the American Astronomical Society,
Djorgovski, S. G., Mahabal, A. A., Drake, A. J., et al. 2012, Vol. 36, American Astronomical Society Meeting
in IAU Symposium, Vol. 285, New Horizons in Time Abstracts, 1530
Domain Astronomy, ed. E. Griffin, R. Hanisch, & Gawiser, E., van Dokkum, P. G., Herrera, D., et al. 2006,
R. Seaman, 141–146 ApJS, 162, 1, doi: 10.1086/497644
Gezari, S. 2012, in European Physical Journal Web of
Dodelson, S., & Vallinotto, A. 2006, PhRvD, 74, 063515,
Conferences, Vol. 39, European Physical Journal Web of
doi: 10.1103/PhysRevD.74.063515
Conferences, 03001
Dolney, D., Jain, B., & Takada, M. 2006, MNRAS, 366,
Gezari, S., Basa, S., Martin, D. C., et al. 2008, ApJ, 676,
884, doi: 10.1111/j.1365-2966.2005.09606.x
944, doi: 10.1086/529008
Doressoundiram, A., Peixinho, N., Moullet, A., et al. 2007,
Gezari, S., Chornock, R., Rest, A., et al. 2012, Nature, 485,
AJ, 134, 2186, doi: 10.1086/522783
217, doi: 10.1038/nature10990
Drake, A. J., Djorgovski, S. G., Mahabal, A., et al. 2009,
Giannios, D., & Metzger, B. D. 2011, MNRAS, 416, 2102,
ApJ, 696, 870, doi: 10.1088/0004-637X/696/1/870
doi: 10.1111/j.1365-2966.2011.19188.x
Drake, A. J., Beshore, E., Catelan, M., et al. 2010, ArXiv
Girard, T. M., Korchagin, V. I., Casetti-Dinescu, D. I.,
e-prints. https://arxiv.org/abs/1009.3048
et al. 2006, AJ, 132, 1768, doi: 10.1086/507331
Drlica-Wagner, A., Albert, A., Bechtol, K., et al. 2015,
Gladman, B., Kavelaars, J. J., Petit, J.-M., et al. 2001, AJ,
ApJL, 809, L4, doi: 10.1088/2041-8205/809/1/L4
122, 1051, doi: 10.1086/322080
Drout, M. R., Chornock, R., Soderberg, A. M., et al. 2014,
Goldstein, D. A., & Nugent, P. E. 2017, ApJL, 834, L5,
ApJ, 794, 23, doi: 10.1088/0004-637X/794/1/23
doi: 10.3847/2041-8213/834/1/L5
Duncan, M. J., Levison, H. F., & Budd, S. M. 1995, AJ, Goldstein, D. A., Nugent, P. E., Kasen, D. N., & Collett,
110, 3073, doi: 10.1086/117748 T. E. 2017, ArXiv e-prints.
Durech, J., Kaasalainen, M., Warner, B. D., et al. 2009, https://arxiv.org/abs/1708.00003
A&A, 493, 291, doi: 10.1051/0004-6361:200810393 González, J. E., Lacey, C. G., Baugh, C. M., Frenk, C. S.,
Eisenstein, D. J., Hu, W., & Tegmark, M. 1998, ApJL, 504, & Benson, A. J. 2009, MNRAS, 397, 1254,
L57, doi: 10.1086/311582 doi: 10.1111/j.1365-2966.2009.15057.x
Elliot, J. L., Kern, S. D., Clancy, K. B., et al. 2005, AJ, Gordon, C., Land, K., & Slosar, A. 2007, Physical Review
129, 1117, doi: 10.1086/427395 Letters, 99, 081301, doi: 10.1103/PhysRevLett.99.081301
Engelhardt, T., Jedicke, R., Vereš, P., et al. 2017, AJ, 153, Gould, A. 1992, ApJ, 392, 442, doi: 10.1086/171443
133, doi: 10.3847/1538-3881/aa5c8a Graham, M. J., Djorgovski, S. G., Drake, A. J., et al. 2014,
Evans, C. R., & Kochanek, C. S. 1989, ApJL, 346, L13, MNRAS, 439, 703, doi: 10.1093/mnras/stt2499
doi: 10.1086/185567 Graham, M. L., Connolly, A. J., Ivezić, Ž., et al. 2018, AJ,
Feindt, U., Kerschhaggl, M., Kowalski, M., et al. 2013, 155, 1, doi: 10.3847/1538-3881/aa99d4
A&A, 560, A90, doi: 10.1051/0004-6361/201321880 Granvik, M., Virtanen, J., Oszkiewicz, D., & Muinonen, K.
Finch, C. T., Zacharias, N., & Henry, T. J. 2010, AJ, 140, 2009, Meteoritics and Planetary Science, 44, 1853,
844, doi: 10.1088/0004-6256/140/3/844 doi: 10.1111/j.1945-5100.2009.tb01994.x
Graur, O., Bianco, F. B., Huang, S., et al. 2017, ApJ, 837, Hu, W., & Kravtsov, A. V. 2003, ApJ, 584, 702,
120, doi: 10.3847/1538-4357/aa5eb8 doi: 10.1086/345846
Grav, T., Jedicke, R., Denneau, L., et al. 2007, in Bulletin Hu, W., & Tegmark, M. 1999, ApJL, 514, L65,
of the American Astronomical Society, Vol. 39, American doi: 10.1086/311947
Astronomical Society Meeting Abstracts, 807 Huterer, D., Shafer, D. L., & Schmidt, F. 2015, JCAP, 12,
Grav, T., Mainzer, A. K., & Spahr, T. 2016, AJ, 151, 172, 033, doi: 10.1088/1475-7516/2015/12/033
doi: 10.3847/0004-6256/151/6/172 Huterer, D., Takada, M., Bernstein, G., & Jain, B. 2006,
Greco, J. P., Greene, J. E., Strauss, M. A., et al. 2018, ApJ, MNRAS, 366, 101, doi: 10.1111/j.1365-2966.2005.09782.x
857, doi: 10.3847/1538-4357/aab842 Ibata, R., Irwin, M., Lewis, G. F., & Stolte, A. 2001, ApJL,
Green, G. M., Schlafly, E. F., Finkbeiner, D. P., et al. 2014, 547, L133, doi: 10.1086/318894
ApJ, 783, 114, doi: 10.1088/0004-637X/783/2/114 Ishak, M., Upadhye, A., & Spergel, D. N. 2006, PhRvD, 74,
Grier, C. J., Trump, J. R., Shen, Y., et al. 2017, ApJ, 851, 043513, doi: 10.1103/PhysRevD.74.043513
21, doi: 10.3847/1538-4357/aa98dc Ivezić, Ž. 2007, in ASP Conf. Ser., Vol. 378, Why Galaxies
Gunn, J. E., Carr, M., Rockosi, C., et al. 1998, AJ, 116, Care About AGB Stars: Their Importance as Actors and
3040, doi: 10.1086/300645 Probes, ed. F. Kerschbaum, C. Charbonnel, & R. F.
Gunn, J. E., Siegmund, W. A., Mannery, E. J., et al. 2006, Wing, 485
AJ, 131, 2332, doi: 10.1086/500975 Ivezić, Ž., Beers, T. C., & Jurić, M. 2012, ARA&A, 50, 251,
Habibi, F., Moniez, M., Ansari, R., & Rahvar, S. 2011, doi: 10.1146/annurev-astro-081811-125504
A&A, 525, A108, doi: 10.1051/0004-6361/201015260 Ivezić, Ž., Brandt, W. N., Fan, X., et al. 2014, in IAU
Han, C. 2008, ApJ, 681, 806, doi: 10.1086/588083 Symposium, Vol. 304, Multiwavelength AGN Surveys and
Hannestad, S., Tu, H., & Wong, Y. Y. 2006, JCAP, 6, 025, Studies, ed. A. M. Mickaelian & D. B. Sanders, 11–17
doi: 10.1088/1475-7516/2006/06/025 Ivezić, Ž., & The LSST Science Collaboration. 2011, LSST
Hargis, J. R., Willman, B., & Peter, A. H. G. 2014, ApJL, Science Requirements Document, LSST Document
795, L13, doi: 10.1088/2041-8205/795/1/L13 LPM-17, LSST. https://ls.st/LPM-17
Harrison, F. A., Craig, W. W., Christensen, F. E., et al. Ivezić, Ž., Tabachnik, S., Rafikov, R., et al. 2001, AJ, 122,
2013, ApJ, 770, 103, doi: 10.1088/0004-637X/770/2/103 2749, doi: 10.1086/323452
Hartman, J. D., Gaudi, B. S., Holman, M. J., et al. 2009, Ivezić, Ž., Lupton, R., Schlegel, D., et al. 2004, in ASP
ApJ, 695, 336, doi: 10.1088/0004-637X/695/1/336 Conf. Ser., Vol. 327, Satellites and Tidal Streams, ed.
Helmi, A., Ivezić, Ž., Prada, F., et al. 2003, ApJ, 586, 195, F. Prada, D. Martinez Delgado, & T. J. Mahoney, 104
doi: 10.1086/367536 Ivezić, Ž., Tyson, J. A., Jurić, M., et al. 2007a, in IAU
Hicken, M., Wood-Vasey, W. M., Blondin, S., et al. 2009, Symposium, Vol. 236, Near Earth Objects, our Celestial
ApJ, 700, 1097, doi: 10.1088/0004-637X/700/2/1097 Neighbors: Opportunity and Risk, ed. G. B. Valsecchi,
Hill, V., Lecureur, A., Gómez, A., et al. 2011, A&A, 534, D. Vokrouhlický, & A. Milani, 353–362
A80, doi: 10.1051/0004-6361/200913757 Ivezić, Ž., Smith, J. A., Miknaitis, G., et al. 2007b, AJ, 134,
Höflich, P., Wheeler, J. C., & Thielemann, F. K. 1998, ApJ, 973, doi: 10.1086/519976
495, 617, doi: 10.1086/305327 Ivezić, Ž., Sesar, B., Jurić, M., et al. 2008, ApJ, 684, 287,
Hovatta, T., Pavlidou, V., King, O. G., et al. 2014, doi: 10.1086/589678
MNRAS, 439, 690, doi: 10.1093/mnras/stt2494 Jain, B., Vikram, V., & Sakstein, J. 2013, ApJ, 779, 39,
Howell, D. A., Sullivan, M., Conley, A., & Carlberg, R. doi: 10.1088/0004-637X/779/1/39
2007, ApJL, 667, L37, doi: 10.1086/522030 Jain, B., & Zhang, P. 2008, PhRvD, 78, 063503,
Howlett, C., Robotham, A. S. G., Lagos, C. D. P., & Kim, doi: 10.1103/PhysRevD.78.063503
A. G. 2017, ApJ, 847, 128, Jain, B., Spergel, D., Bean, R., et al. 2015, ArXiv e-prints.
doi: 10.3847/1538-4357/aa88c8 https://arxiv.org/abs/1501.07897
Hsieh, H. H., & Jewitt, D. 2006, Science, 312, 561, Javanmardi, B., Porciani, C., Kroupa, P., &
doi: 10.1126/science.1125150 Pflamm-Altenburg, J. 2015, ApJ, 810, 47,
Hu, W., & Haiman, Z. 2003, PhRvD, 68, 063004, doi: 10.1088/0004-637X/810/1/47
doi: 10.1103/PhysRevD.68.063004 Jedicke, R., Denneau, L., Grav, T., et al. 2005, in Bulletin
Hu, W., & Jain, B. 2004, PhRvD, 70, 043009, of the American Astronomical Society, Vol. 37, American
doi: 10.1103/PhysRevD.70.043009 Astronomical Society Meeting Abstracts, 1363
Jedicke, R., Nesvorný, D., Whiteley, R., Ivezić, Ž., & Jurić, Kaiser, N., Aussel, H., Burke, B. E., et al. 2002, in
M. 2004, Nature, 429, 275, doi: 10.1038/nature02578 Proc. SPIE, Vol. 4836, Survey and Other Telescope
Jee, M. J., & Tyson, J. A. 2011, PASP, 123, 596, Technologies and Discoveries, ed. J. A. Tyson & S. Wolff,
doi: 10.1086/660137 154–164
Jenness, T., Bosch, J., Owen, R., et al. 2016, in Proc. SPIE, Kaiser, N., Burgett, W., Chambers, K., et al. 2010, in
Vol. 9913, Software and Cyberinfrastructure for Proc. SPIE, Vol. 7733, Ground-based and Airborne
Astronomy IV, 99130G Telescopes III, 77330E
Kalirai, J. S., Hansen, B. M. S., Kelson, D. D., et al. 2008,
Jenness, T., Economou, F., Findeisen, K., et al. 2018, in
ApJ, 676, 594, doi: 10.1086/527028
Proc. SPIE, Vol. 10707, Software and
Kalirai, J. S., Richer, H. B., Fahlman, G. G., et al. 2001,
Cyberinfrastructure for Astronomy V, in press.
AJ, 122, 257, doi: 10.1086/321140
https://ls.st/Document-28452
Kann, D. A., Klose, S., Zhang, B., et al. 2010, ApJ, 720,
Jewitt, D. 2012, AJ, 143, 66,
1513, doi: 10.1088/0004-637X/720/2/1513
doi: 10.1088/0004-6256/143/3/66
Kasliwal, M. M. 2011, PhD thesis, California Institute of
Jewitt, D., Stuart, J. S., & Li, J. 2011, AJ, 142, 28,
Technology
doi: 10.1088/0004-6256/142/1/28
Kaspi, S., Brandt, W. N., Maoz, D., et al. 2007, ApJ, 659,
Jewitt, D. C., Trujillo, C. A., & Luu, J. X. 2000, AJ, 120, 997, doi: 10.1086/512094
1140, doi: 10.1086/301453 Keller, S. C., Schmidt, B. P., Bessell, M. S., et al. 2007,
Jiang, Y.-F., Green, P. J., Greene, J. E., et al. 2017, ApJ, PASA, 24, 1, doi: 10.1071/AS07001
836, 186, doi: 10.3847/1538-4357/aa5b91 Kilbinger, M. 2015, Reports on Progress in Physics, 78,
Johnston, K. V., Bullock, J. S., Sharma, S., et al. 2008, 086901, doi: 10.1088/0034-4885/78/8/086901
ApJ, 689, 936, doi: 10.1086/592228 Knezevic, Z., & Milani, A. 2005, Highlights of Astronomy,
Jones, R. L., Gladman, B., Petit, J.-M., et al. 2006, Icarus, 13, 758
185, 508, doi: 10.1016/j.icarus.2006.07.024 Knox, L., Song, Y.-S., & Tyson, J. A. 2006, PhRvD, 74,
Jones, R. L., Chesley, S. R., Connolly, A. J., et al. 2007, in 023512, doi: 10.1103/PhysRevD.74.023512
Bulletin of the American Astronomical Society, Vol. 39, Komatsu, E., Dunkley, J., Nolta, M. R., et al. 2009, ApJS,
American Astronomical Society Meeting Abstracts, 981 180, 330, doi: 10.1088/0067-0049/180/2/330
Jones, R. L., Yoachim, P., Chandrasekharan, S., et al. 2014, Komossa, S. 2015, Journal of High Energy Astrophysics, 7,
in Proc. SPIE, Vol. 9149, Observatory Operations: 148, doi: 10.1016/j.jheap.2015.04.006
Strategies, Processes, and Systems V, 91490B Kowalski, A. F., Hawley, S. L., Holtzman, J. A.,
Jones, R. L., Slater, C. T., Moeyens, J., et al. 2018, Icarus, Wisniewski, J. P., & Hilton, E. J. 2010, ApJL, 714, L98,
303, 181, doi: 10.1016/j.icarus.2017.11.033 doi: 10.1088/2041-8205/714/1/L98
Kowalski, M., Rubin, D., Aldering, G., et al. 2008, ApJ,
Joudaki, S., Blake, C., Johnson, A., et al. 2018, MNRAS,
686, 749, doi: 10.1086/589937
474, 4894, doi: 10.1093/mnras/stx2820
Krause, E., & Eifler, T. 2017, MNRAS, 470, 2100,
Jurić, M., Becker, A., Shaw, R., Krughoff, K. S., & Kantor,
doi: 10.1093/mnras/stx1261
J. 2013, Winter 2013 LSST DM Data Challenge Release
Kulkarni, S. R., Ofek, E. O., Rau, A., et al. 2007, Nature,
Notes, DM Technical Note DMTN-035, LSST Data
447, 458, doi: 10.1038/nature05822
Management. https://dmtn-035.lsst.io
Kundić, T., Turner, E. L., Colley, W. N., et al. 1997, ApJ,
Jurić, M., Ciardi, D., & Dubois-Felsmann, G. 2017a, LSST
482, 75, doi: 10.1086/304147
Science Platform Vision Document, LSST Document
Kurucz, R. L. 1979, ApJS, 40, 1, doi: 10.1086/190589
LSE-319, LSST. https://ls.st/LSE-319
—. 1993, SYNTHE spectrum synthesis programs and line
Jurić, M., Ivezić, Ž., Brooks, A., et al. 2008, ApJ, 673, 864, data (Smithsonian Astrophysical Observatory)
doi: 10.1086/523619 Laher, R. R., Masci, F. J., Groom, S., et al. 2017, ArXiv
Jurić, M., Kantor, J., Lim, K., et al. 2015, ArXiv e-prints. e-prints. https://arxiv.org/abs/1708.01584
https://arxiv.org/abs/1512.07914 Laureijs, R., Amiaux, J., Arduini, S., et al. 2011, ArXiv
Jurić, M., et al. 2017b, LSST Data Products Definition e-prints. https://arxiv.org/abs/1110.3193
Document, LSST Document LSE-163, LSST. Law, D. R., Johnston, K. V., & Majewski, S. R. 2005, ApJ,
https://ls.st/LSE-163 619, 807, doi: 10.1086/426779
Law, N. M., Kulkarni, S. R., Dekany, R. G., et al. 2009, Mamajek, E. E. 2012, ArXiv e-prints.
PASP, 121, 1395, doi: 10.1086/648598 https://arxiv.org/abs/1210.1616
Leistedt, B., Peiris, H. V., & Roth, N. 2014, Physical Mana, A., Giannantonio, T., Weller, J., et al. 2013,
Review Letters, 113, 221301, MNRAS, 434, 684, doi: 10.1093/mnras/stt1062
doi: 10.1103/PhysRevLett.113.221301 Mandelbaum, R. 2017, ArXiv e-prints.
Lenz, D. D., Newberg, J., Rosner, R., Richards, G. T., & https://arxiv.org/abs/1710.03235
Stoughton, C. 1998, ApJS, 119, 121, doi: 10.1086/313155 Mandelbaum, R., Seljak, U., Kauffmann, G., Hirata, C. M.,
Lépine, S. 2008, AJ, 135, 2177, & Brinkmann, J. 2006, MNRAS, 368, 715,
doi: 10.1088/0004-6256/135/6/2177 doi: 10.1111/j.1365-2966.2006.10156.x
Levi, M., Bebek, C., Beers, T., et al. 2013, ArXiv e-prints.
Marcus, R. A., Ragozzine, D., Murray-Clay, R. A., &
https://arxiv.org/abs/1308.0847
Holman, M. J. 2011, ApJ, 733, 40,
Linder, E. V. 2003, PhRvD, 68, 083504,
doi: 10.1088/0004-637X/733/1/40
doi: 10.1103/PhysRevD.68.083504
Martin, D. C., Fanson, J., Schiminovich, D., et al. 2005,
Lintott, C., Schawinski, K., Bamford, S., et al. 2011,
ApJL, 619, L1, doi: 10.1086/426387
MNRAS, 410, 166, doi: 10.1111/j.1365-2966.2010.17432.x
Martini, P., & Schneider, D. P. 2003, ApJL, 597, L109,
Longhitano, M., & Binggeli, B. 2010, A&A, 509, A46,
doi: 10.1086/379888
doi: 10.1051/0004-6361/200913109
Lorimer, D. R., Bailes, M., McLaughlin, M. A., Narkevic, Mathieu, R. D. 2000, in ASP Conf. Ser., Vol. 198, Stellar
D. J., & Crawford, F. 2007, Science, 318, 777, Clusters and Associations: Convection, Rotation, and
doi: 10.1126/science.1147532 Dynamos, ed. R. Pallavicini, G. Micela, & S. Sciortino,
Lowry, S. C., Fitzsimmons, A., Cartwright, I. M., & 517
Williams, I. P. 1999, A&A, 349, 649 Matthews, D. J., & Newman, J. A. 2010, ApJ, 721, 456,
LSST Dark Energy Science Collaboration. 2012, ArXiv doi: 10.1088/0004-637X/721/1/456
e-prints. https://arxiv.org/abs/1211.0310 Ménard, B., Scranton, R., Schmidt, S., et al. 2013, ArXiv
LSST Science Collaboration, Abell, P. A., Allison, J., et al. e-prints. https://arxiv.org/abs/1303.4722
2009, ArXiv e-prints. https://arxiv.org/abs/0912.0201 Merloni, A., Predehl, P., Becker, W., et al. 2012, ArXiv
LSST Science Collaboration, Marshall, P., Anguita, T., e-prints. https://arxiv.org/abs/1209.3114
et al. 2017, ArXiv e-prints. Metzger, B. D. 2017, Living Reviews in Relativity, 20, 3,
https://arxiv.org/abs/1708.04058 doi: 10.1007/s41114-017-0006-z
Lucas, P. W., & Roche, P. F. 2000, MNRAS, 314, 858, Metzger, B. D., & Berger, E. 2012, ApJ, 746, 48,
doi: 10.1046/j.1365-8711.2000.03515.x doi: 10.1088/0004-637X/746/1/48
Lue, A., Scoccimarro, R., & Starkman, G. 2004, PhRvD, Miyazaki, S., Komiyama, Y., Kawanomoto, S., et al. 2018,
69, 044005, doi: 10.1103/PhysRevD.69.044005 PASJ, 70, S1, doi: 10.1093/pasj/psx063
Luhman, K. L. 2014, ApJL, 786, L18,
Mo, H. J., Jing, Y. P., & White, S. D. M. 1996, MNRAS,
doi: 10.1088/2041-8205/786/2/L18
282, 1096, doi: 10.1093/mnras/282.3.1096
Lupton, R., Blanton, M. R., Fekete, G., et al. 2004, PASP,
Monet, D. G., Levine, S. E., Canzian, B., et al. 2003, AJ,
116, 133, doi: 10.1086/382245
125, 984, doi: 10.1086/345888
Macaulay, E., Davis, T. M., Scovacricchi, D., et al. 2017,
Moniez, M. 2003, A&A, 412, 105,
MNRAS, 467, 259, doi: 10.1093/mnras/stw3339
doi: 10.1051/0004-6361:20031478
MacLeod, C. L., Ivezić, Ž., Kochanek, C. S., et al. 2010,
Morokuma, T., Inada, N., Oguri, M., et al. 2007, AJ, 133,
ApJ, 721, 1014, doi: 10.1088/0004-637X/721/2/1014
214, doi: 10.1086/509701
MacLeod, C. L., Ross, N. P., Lawrence, A., et al. 2016,
MNRAS, 457, 389, doi: 10.1093/mnras/stv2997 Morrison, C. B., Scranton, R., Ménard, B., et al. 2012,
Mahabal, A. A., Djorgovski, S. G., Drake, A. J., et al. 2011, MNRAS, 426, 2489,
Bulletin of the Astronomical Society of India, 39, 387. doi: 10.1111/j.1365-2966.2012.21826.x
https://arxiv.org/abs/1111.0313 Munn, J. A., Monet, D. G., Levine, S. E., et al. 2004, AJ,
Majewski, S. R., Skrutskie, M. F., Weinberg, M. D., & 127, 3034, doi: 10.1086/383292
Ostheimer, J. C. 2003, ApJ, 599, 1082, Najita, J., Willman, B., Finkbeiner, D. P., et al. 2016,
doi: 10.1086/379504 ArXiv e-prints. https://arxiv.org/abs/1610.01661
National Research Council. 2001, Astronomy and Patat, F. 2008, A&A, 481, 575,
Astrophysics in the New Millennium (Washington, DC: doi: 10.1051/0004-6361:20079279
The National Academies Press), doi: 10.17226/9839. Percival, W. J., Reid, B. A., Eisenstein, D. J., et al. 2010,
https://www.nap.edu/catalog/9839/ MNRAS, 401, 2148,
—. 2003a, Connecting Quarks with the Cosmos: Eleven doi: 10.1111/j.1365-2966.2009.15812.x
Science Questions for the New Century (Washington, DC: Perlmutter, S., Aldering, G., Goldhaber, G., et al. 1999,
The National Academies Press), doi: 10.17226/10079. ApJ, 517, 565, doi: 10.1086/307221
https://www.nap.edu/catalog/10079/ Perryman, M. A. C., de Boer, K. S., Gilmore, G., et al.
—. 2003b, New Frontiers in the Solar System: An 2001, A&A, 369, 339, doi: 10.1051/0004-6361:20010085
Integrated Exploration Strategy (Washington, DC: The Peterson, J. R., Jernigan, J. G., Kahn, S. M., et al. 2015,
National Academies Press), doi: 10.17226/10432. ApJS, 218, 14, doi: 10.1088/0067-0049/218/1/14
https://www.nap.edu/catalog/10432/ Petri, A., May, M., & Haiman, Z. 2016, PhRvD, 94,
—. 2010, New Worlds, New Horizons in Astronomy and 063534, doi: 10.1103/PhysRevD.94.063534
Astrophysics (Washington, DC: The National Academies Pettersen, B. R., & Hawley, S. L. 1989, A&A, 217, 187
Press), doi: 10.17226/12951. Pier, J. R., Munn, J. A., Hindsley, R. B., et al. 2003, AJ,
https://www.nap.edu/catalog/12951/ 125, 1559, doi: 10.1086/346138
Neill, J. D., & Shara, M. M. 2005, AJ, 129, 1873, Pinto, P. A., Smith, C. R., & Garnavich, P. M. 2004, in
doi: 10.1086/428482 Bulletin of the American Astronomical Society, Vol. 36,
Nemiroff, R. J. 2003, AJ, 125, 2740, doi: 10.1086/374571 American Astronomical Society Meeting Abstracts, 1530
Planck Collaboration, Ade, P. A. R., Aghanim, N., et al.
Ness, M., Debattista, V. P., Bensby, T., et al. 2014, ApJL,
2011, A&A, 536, A23, doi: 10.1051/0004-6361/201116472
787, L19, doi: 10.1088/2041-8205/787/2/L19
Porciani, C., & Madau, P. 2000, ApJ, 532, 679,
Nesvorný, D., Jedicke, R., Whiteley, R. J., & Ivezić, Ž.
doi: 10.1086/308587
2005, Icarus, 173, 132, doi: 10.1016/j.icarus.2004.07.026
Pravec, P., & Harris, A. W. 2000, Icarus, 148, 12,
Newburgh, L. B., Addison, G. E., Amiri, M., et al. 2014, in
doi: 10.1006/icar.2000.6482
Proc. SPIE, Vol. 9145, Ground-based and Airborne
Quartin, M., Marra, V., & Amendola, L. 2014, PhRvD, 89,
Telescopes V, 91454V
023009, doi: 10.1103/PhysRevD.89.023009
Newburgh, L. B., Bandura, K., Bucher, M. A., et al. 2016,
Rabinowitz, D. L. 1993, ApJ, 407, 412, doi: 10.1086/172524
in Proc. SPIE, Vol. 9906, Ground-based and Airborne
Raddick, J., Souter, B., Lemson, G., & Taghizadeh-Popp,
Telescopes VI, 99065X
M. 2017, in American Astronomical Society Meeting
Newman, A. B., Treu, T., Ellis, R. S., et al. 2013, ApJ, 765,
Abstracts, Vol. 229, American Astronomical Society
24, doi: 10.1088/0004-637X/765/1/24
Meeting Abstracts, 236.15
Newman, J. A. 2008, ApJ, 684, 88, doi: 10.1086/589982
Reid, I. N., Gizis, J. E., & Hawley, S. L. 2002, AJ, 124,
Newman, J. A., Abate, A., Abdalla, F. B., et al. 2015,
2721, doi: 10.1086/343777
Astroparticle Physics, 63, 81,
Reyes, R., Mandelbaum, R., Seljak, U., et al. 2010, Nature,
doi: 10.1016/j.astropartphys.2014.06.007 464, 256, doi: 10.1038/nature08857
Nissanke, S., Kasliwal, M., & Georgieva, A. 2013, ApJ, 767, Rhodes, J., Nichol, R. C., Aubourg, É., et al. 2017, ApJS,
124, doi: 10.1088/0004-637X/767/2/124 233, 21, doi: 10.3847/1538-4365/aa96b0
Ofek, E. O., Sullivan, M., Cenko, S. B., et al. 2013, Nature, Richards, G. T., Fan, X., Newberg, H. J., et al. 2002, AJ,
494, 65, doi: 10.1038/nature11877 123, 2945, doi: 10.1086/340187
Oguri, M., & Kawano, Y. 2003, MNRAS, 338, L25, Riess, A. G. 2000, in ASP Conf. Ser., Vol. 201, Cosmic
doi: 10.1046/j.1365-8711.2003.06290.x Flows Workshop, ed. S. Courteau & J. Willick, 80
Oguri, M., & Marshall, P. J. 2010, MNRAS, 405, 2579, Riess, A. G., Filippenko, A. V., Challis, P., et al. 1998, AJ,
doi: 10.1111/j.1365-2966.2010.16639.x 116, 1009, doi: 10.1086/300499
Oguri, M., & Takada, M. 2011, PhRvD, 83, 023008, Riess, A. G., Strolger, L.-G., Casertano, S., et al. 2007,
doi: 10.1103/PhysRevD.83.023008 ApJ, 659, 98, doi: 10.1086/510378
Padmanabhan, N., Schlegel, D. J., Finkbeiner, D. P., et al. Robertson, B. E., Banerji, M., Cooper, M. C., et al. 2017,
2008, ApJ, 674, 1217, doi: 10.1086/524677 ArXiv e-prints. https://arxiv.org/abs/1708.01617
Parker, A., Ivezić, Ž., Jurić, M., et al. 2008, Icarus, 198, Rocha, M., Peter, A. H. G., Bullock, J. S., et al. 2013,
138, doi: 10.1016/j.icarus.2008.07.002 MNRAS, 430, 81, doi: 10.1093/mnras/sts514
Ross, N. P., Myers, A. D., Sheldon, E. S., et al. 2012, ApJS, Smartt, S. J. 2009, ARA&A, 47, 63,
199, 3, doi: 10.1088/0067-0049/199/1/3 doi: 10.1146/annurev-astro-082708-101737
Rossetto, B. M., Santiago, B. X., Girardi, L., et al. 2011, Smith, N. 2014, ARA&A, 52, 487,
AJ, 141, 185, doi: 10.1088/0004-6256/141/6/185 doi: 10.1146/annurev-astro-081913-040025
Rowe, B. T. P., Jarvis, M., Mandelbaum, R., et al. 2015, Smith, N., Li, W., Silverman, J. M., Ganeshalingam, M., &
Astronomy and Computing, 10, 121, Filippenko, A. V. 2011, MNRAS, 415, 773,
doi: 10.1016/j.ascom.2015.02.002 doi: 10.1111/j.1365-2966.2011.18763.x
Rowe, J. F., Bryson, S. T., Marcy, G. W., et al. 2014, ApJ, Solontoi, M. 2010, PhD thesis, University of Washington
784, 45, doi: 10.1088/0004-637X/784/1/45 Song, Y.-S., & Knox, L. 2004, PhRvD, 70, 063510,
Schlafly, E. F., Finkbeiner, D. P., Jurić, M., et al. 2012, doi: 10.1103/PhysRevD.70.063510
ApJ, 756, 158, doi: 10.1088/0004-637X/756/2/158 Spergel, D., Gehrels, N., Baltay, C., et al. 2015, ArXiv
Schlegel, D. J., Finkbeiner, D. P., & Davis, M. 1998, ApJ, e-prints. https://arxiv.org/abs/1503.03757
500, 525, doi: 10.1086/305772 Staniszewski, Z., Ade, P. A. R., Aird, K. A., et al. 2009,
Science Working Group of the LSST, & Strauss, M. A. ApJ, 701, 32, doi: 10.1088/0004-637X/701/1/32
2004, Towards a Design Reference Mission for the Large Stassun, K., Paegert, M., De Lee, N. M., & Cargile, P.
Synoptic Survey Telescope, Document-26952, LSST. 2013, in American Astronomical Society Meeting
https://ls.st/Document-26952 Abstracts, Vol. 221, American Astronomical Society
Scolnic, D., Kessler, R., Brout, D., et al. 2018, ApJ, 852, Meeting Abstracts #221, 116.01
doi: 10.3847/2041-8213/aa9d82
Strubbe, L. E., & Quataert, E. 2009, MNRAS, 400, 2070,
Scovacricchi, D., Nichol, R. C., Bacon, D., Sullivan, M., &
doi: 10.1111/j.1365-2966.2009.15599.x
Prajs, S. 2016, MNRAS, 456, 1700,
Stubbs, C. W., & Tonry, J. L. 2006, ApJ, 646, 1436,
doi: 10.1093/mnras/stv2752
doi: 10.1086/505138
Scovacricchi, D., Nichol, R. C., Macaulay, E., & Bacon, D.
Stubbs, C. W., High, F. W., George, M. R., et al. 2007,
2017, MNRAS, 465, 2862, doi: 10.1093/mnras/stw2878
PASP, 119, 1163, doi: 10.1086/522208
Scranton, R., Connolly, A., Krughoff, S., et al. 2007, ArXiv
Suwa, Y. 2018, MNRAS, 474, 2612,
e-prints. https://arxiv.org/abs/0709.0752
doi: 10.1093/mnras/stx2953
Seljak, U. 2009, Physical Review Letters, 102, 021302,
Suyu, S. H., Marshall, P. J., Auger, M. W., et al. 2010,
doi: 10.1103/PhysRevLett.102.021302
ApJ, 711, 201, doi: 10.1088/0004-637X/711/1/201
Seo, H.-J., & Eisenstein, D. J. 2003, ApJ, 598, 720,
Szabó, G. M., Ivezić, Ž., Jurić, M., & Lupton, R. 2007,
doi: 10.1086/379122
MNRAS, 377, 1393,
Sesar, B., Ivezić, Ž., Lupton, R. H., et al. 2007, AJ, 134,
doi: 10.1111/j.1365-2966.2007.11687.x
2236, doi: 10.1086/521819
Takada, M. 2006, PhRvD, 74, 043505,
Sesar, B., Ivezić, Ž., Grammer, S. H., et al. 2010, ApJ, 708,
doi: 10.1103/PhysRevD.74.043505
717, doi: 10.1088/0004-637X/708/1/717
Shara, M. M. 2006, AJ, 131, 2980, doi: 10.1086/501502 Takada, M., & Jain, B. 2004, MNRAS, 348, 897,
Sharma, S., Johnston, K. V., Majewski, S. R., Bullock, J., doi: 10.1111/j.1365-2966.2004.07410.x
& Muñoz, R. R. 2011, ApJ, 728, 106, Takada, M., & Spergel, D. N. 2014, MNRAS, 441, 2456,
doi: 10.1088/0004-637X/728/2/106 doi: 10.1093/mnras/stu759
Shen, K. J., & Bildsten, L. 2009, ApJ, 692, 324, Takada, M., Ellis, R. S., Chiba, M., et al. 2014, PASJ, 66,
doi: 10.1088/0004-637X/692/1/324 R1, doi: 10.1093/pasj/pst019
Shen, Y., Brandt, W. N., Dawson, K. S., et al. 2015, ApJS, Taylor, M. B. 2005, in ASP Conf. Ser., Vol. 347,
216, 4, doi: 10.1088/0067-0049/216/1/4 Astronomical Data Analysis Software and Systems XIV,
Sheppard, S. S., & Trujillo, C. 2016, AJ, 152, 221, ed. P. Shopbell, M. Britton, & R. Ebert, 29
doi: 10.3847/1538-3881/152/6/221 Thompson, T. A., Prieto, J. L., Stanek, K. Z., et al. 2009,
Shields, G. A., & Bonning, E. W. 2008, ApJ, 682, 758, ApJ, 705, 1364, doi: 10.1088/0004-637X/705/2/1364
doi: 10.1086/589427 Thornton, D., Stappers, B., Bailes, M., et al. 2013, Science,
Silsbee, K., & Tremaine, S. 2016, AJ, 152, 103, 341, 53, doi: 10.1126/science.1236789
doi: 10.3847/0004-6256/152/4/103 Tisserand, P., Le Guillou, L., Afonso, C., et al. 2007, A&A,
Skrutskie, M. F., Cutri, R. M., Stiening, R., et al. 2006, AJ, 469, 387, doi: 10.1051/0004-6361:20066017
131, 1163, doi: 10.1086/498708 Tokovinin, A. 2002, PASP, 114, 1156, doi: 10.1086/342683
Treu, T., Koopmans, L. V., Bolton, A. S., Burles, S., & Wang, L., Goldhaber, G., Aldering, G., & Perlmutter, S.
Moustakas, L. A. 2006, ApJ, 640, 662, 2003, ApJ, 590, 944, doi: 10.1086/375020
doi: 10.1086/500124 Weinberg, D. H., Bullock, J. S., Governato, F., Kuzio de
Treu, T., & Marshall, P. J. 2016, A&A Rv, 24, 11, Naray, R., & Peter, A. H. G. 2015, Proceedings of the
doi: 10.1007/s00159-016-0096-8 National Academy of Science, 112, 12249,
Trilling, D. E., Robinson, T., Roegge, A., et al. 2017, ApJL, doi: 10.1073/pnas.1308716112
850, L38, doi: 10.3847/2041-8213/aa9989 Weinberg, D. H., Mortonson, M. J., Eisenstein, D. J., et al.
Trujillo, C. A., Jewitt, D. C., & Luu, J. X. 2001, AJ, 122, 2013, PhR, 530, 87, doi: 10.1016/j.physrep.2013.05.001
457, doi: 10.1086/321117 West, A. A., Bochanski, J. J., Bowler, B. P., et al. 2011, in
Trujillo, C. A., & Sheppard, S. S. 2014, Nature, 507, 471, ASP Conf. Ser., Vol. 448, 16th Cambridge Workshop on
doi: 10.1038/nature13156 Cool Stars, Stellar Systems, and the Sun, ed.
Turnbull, S. J., Hudson, M. J., Feldman, H. A., et al. 2012, C. Johns-Krull, M. K. Browning, & A. A. West, 531
MNRAS, 420, 447, doi: 10.1111/j.1365-2966.2011.20050.x Wood-Vasey, W. M., Miknaitis, G., Stubbs, C. W., et al.
Tyson, J. A., Kochanski, G. P., & Dell’Antonio, I. P. 1998, 2007, ApJ, 666, 694, doi: 10.1086/518642
ApJL, 498, L107, doi: 10.1086/311314 Wright, J. T., Marcy, G. W., Howard, A. W., et al. 2012,
Tyson, J. A., Roat, C., Bosch, J., & Wittman, D. 2008, in ApJ, 753, 160, doi: 10.1088/0004-637X/753/2/160
ASP Conf. Ser., Vol. 394, Astronomical Data Analysis Xue, Y. Q., Brandt, W. N., Luo, B., et al. 2010, ApJ, 720,
Software and Systems XVII, ed. R. W. Argyle, P. S. 368, doi: 10.1088/0004-637X/720/1/368
Bunclark, & J. R. Lewis, 107 Yoachim, P., Coughlin, M., Angeli, G. Z., et al. 2016, in
Vagg, D., O’Callaghan, D., O’Hógáin, F., et al. 2016, in Proc. SPIE, Vol. 9910, Observatory Operations:
Proc. SPIE, Vol. 9913, Software and Cyberinfrastructure Strategies, Processes, and Systems VI, 99101A
for Astronomy IV, 99131V Yoo, J., Chanamé, J., & Gould, A. 2004, ApJ, 601, 311,
Vanden Berk, D. E., Richards, G. T., Bauer, A., et al. 2001, doi: 10.1086/380562
AJ, 122, 549, doi: 10.1086/321167 York, D. G., Adelman, J., Anderson, Jr., J. E., et al. 2000,
Vanden Berk, D. E., Wilhite, B. C., Kron, R. G., et al. AJ, 120, 1579, doi: 10.1086/301513
2004, ApJ, 601, 692, doi: 10.1086/380563 Yoshida, F., & Nakamura, T. 2005, AJ, 130, 2900,
Vegetti, S., Lagattuta, D. J., McKean, J. P., et al. 2012, doi: 10.1086/497571
Nature, 481, 341, doi: 10.1038/nature10669 Zaritsky, D., Harris, J., Thompson, I. B., & Grebel, E. K.
Vereš, P., & Chesley, S. R. 2017a, AJ, 154, 12, 2004, AJ, 128, 1606, doi: 10.1086/423910
doi: 10.3847/1538-3881/aa73d1 Zhan, H. 2006, JCAP, 8, 008,
—. 2017b, AJ, 154, 13, doi: 10.3847/1538-3881/aa73d0 doi: 10.1088/1475-7516/2006/08/008
von der Linden, A., Mantz, A., Allen, S. W., et al. 2014, Zhan, H., Knox, L., & Tyson, J. A. 2009, ApJ, 690, 923,
MNRAS, 443, 1973, doi: 10.1093/mnras/stu1423 doi: 10.1088/0004-637X/690/1/923
Walker, M. G., & Peñarrubia, J. 2011, ApJ, 742, 20, Zhan, H., & Tyson, J. A. 2018, Reports on Progress in
doi: 10.1088/0004-637X/742/1/20 Physics, 81, 066901, doi: 10.1088/1361-6633/aab1bd
Walsh, S. M., Willman, B., & Jerjen, H. 2009, AJ, 137, 450, Zhang, B., Fan, Y. Z., Dyks, J., et al. 2006, ApJ, 642, 354,
doi: 10.1088/0004-6256/137/1/450 doi: 10.1086/500723
Wang, D. L., Monkewitz, S. M., Lim, K.-T., & Becla, J. Zhang, B., & Mészáros, P. 2004, International Journal of
2011, in State of the Practice Reports, SC ’11 (New
Modern Physics A, 19, 2385,
York, NY, USA: ACM), 12:1–12:11.
doi: 10.1142/S0217751X0401746X
http://doi.acm.org/10.1145/2063348.2063364

LSST: From Science Drivers To Reference Design and Anticipated Data Products

Uploaded by

Copyright:

Available Formats

LSST: From Science Drivers To Reference Design and Anticipated Data Products

Uploaded by

Document Information

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

LSST: From Science Drivers To Reference Design and Anticipated Data Products

Uploaded by

Copyright:

Available Formats

Draft version August 13, 2019

Typeset using LATEX twocolumn style in AASTeX62

Dominique Boutigny,27 Craig A. Brackett,13 Andrew Bradshaw,4 William Nielsen Brandt,28

place Jussieu, 75005 Paris, France

Alice Domon et Léonie Duquet, Paris Cedex 13, France

Chemin de Bellevue – BP 110, 74940 Annecy-le-Vieux Cedex, France

48 Lawrence Berkeley National Laboratory, 1 Cyclotron Road, Berkeley, CA 94720, USA

Chaoyang District, Beijing 100012, China

1. INTRODUCTION nearly all fields of astronomy – and several areas of fun-

5 For details see http://neo.jpl.nasa.gov/neo/report2007.html

ulations that make up the Milky Way can be gained

visits, and each of these quantities has its own science

• The single-visit exposure time should not be longer

• The mean revisit time (assuming uniform cadence)

Pan-STARRS survey (PS4). With an étendue about 6

9 More information about Pan-STARRS is available from http:

Airmass 1.2 5.0

Figure 9. The baseline design for the LSST telescope. The

ror was cast and polished by the Richard F. Caris Mirror

Figure 11. A cutaway view of LSST camera. Not shown

in an optics structure at the front of the camera body,

The detailed outputs of the LSST Data Management

• Process each night’s data during the day and de-

• Periodically process the accumulated survey data

• Make all LSST data available through an inter-

The DM system will span four key facilities on three

scientific algorithms, public. Our primary goals in pub-

The resulting sky coverage for the LSST baseline ca-

As examples of analysis enabled by the Operations

well described15 by To predict 5σ depths for exposure time τ times longer

NEO completeness H22

PHA completeness H22

Extended LSST Survey (12 years, 30 day linking windows)

tion with dN/dH = 10α H , with α = 0.33, in order to 0.8

NEO completeness H22

PHA completeness H22

of 15 days, with a minimum of two visits per night. 0.2 0.2

The same criterion has been used in NASA studies,

The joint analysis of LSST weak lensing and galaxy

horizon at decoupling, which is imprinted on the

• Higher-order shear and galaxy statistics and shear

15% precision (Abbott et al. 2017b). Constraints

2012 VP113 (Trujillo & Sheppard 2014). Study-

• Mapping the propagation of coronal mass ejections

the magnetic accretion column and in the low state

• Studies of transients from poorly-constrained

• A census of light echoes of historical explosive and

ments (see Figs. 27 and 28) will be available for about

Predicted LSST spatial map

nature of dark matter (e.g., Walker & Peñarrubia

• Detailed constraints on the formation and evolu-

• Studies of the clumpiness of the gravitational po-

• Detailed studies of variable star populations; 2%

• Discovery of rare and faint high proper motion ob-

unrelated to the main science drivers, and is often unan-

Dark Energy Survey is scanning ∼5000 deg2 a magni-

Version 1.0 (May 15, 2008): the first posting.

You might also like