The Dark Energy Survey From Scientific Goals to Science Quality Data Brenna Flaugher Fermilab April 2006 P5 Meeting B.
Download
Report
Transcript The Dark Energy Survey From Scientific Goals to Science Quality Data Brenna Flaugher Fermilab April 2006 P5 Meeting B.
The Dark Energy Survey
From Scientific Goals to Science Quality Data
Brenna Flaugher
Fermilab
April 2006 P5 Meeting
B. Flaugher P5 April 2006
1
DES Science and Technical
Requirements
The Science Requirements
flow to Technical Requirements
• A large camera, on the Blanco 4m
•
5000 deg2 of the So. Galactic
Cap in 525 nights (5 yrs)
– 3 deg2 camera with ≥ 2.2 deg FOV
• Data Management system
– 300GB/night, automated processing
– Publicly available data archive after 1 yr
•
•
photometric-redshifts to z=1.3
with dz < 0.02.
•
A small and stable point spread •
function (PSF) < 0.9'' FWHM
median
Filters, CCDs, Read noise
– SDSS g,r,i,z filters; 400 - 1100nm
– QE > 50% in the z band (825-1100nm)
– Read noise <10 eOptical Corrector with excellent images
– Pixel size <0.3” /pixel
– < 0.4” FWHM in the i and z bands
B. Flaugher P5 April 2006
2
The DES Instrument: DECam
DECam will replace the prime focus cage on the Blanco
F8 Mirror
Filters
Shutter
3556 mm
CCD
Read out
Hexapod
Optical
Lenses
1575 mm
Prime Focus Instrument
-in optical path
-space and thermal constraints
B. Flaugher P5 April 2006
3
DES: DECam and Data Management
DECam will be larger than any existing CCD camera
Each image:
DES Focal Plane
~ 20 Galaxy clusters
~ 200,000 Galaxies
Each night ~ 300 GB
Entire survey ~ 1 PB
John Peoples is the DES project director
Fermilab leads the DECam project
UIUC/NCSA leads the DM project
OUTLINE of this talk
•DECam
•project description
•cost and schedule
62 2kx4k Image CCDs: 520 MPix
8 2kx2k Guide, focus, alignment
•Data Management
•project description
•cost and schedule
B. Flaugher P5 April 2006
4
DES CCDs
LBNL Design: fully depleted 2kx4k CCDs
– QE> 50% at 1000 nm, 250 microns thick
– 15 m pixels, 0.27”/pixel
– readout 250 kpix/sec, readout time ~17sec
LBNL CCDs are much
more efficient than the SITE
CCDs in Mosaic II at high
wavelengths
LBNL CCDs in use on WIYN
telescope. From S. Holland et
al, LBNL-49992 IEEE Trans.
Elec. Dev. Vol.50, No 1, 225338, Jan. 2003
DECam / Mosaic II QE comparison
100
90
80
DES is the 1st
production quantity
application for LBNL
CCDs
z band
70
60
To reach redshifts of ~1.3
DES will spend 46% of survey
time in z –band
QE, LBNL (%)
QE, SITe (%)
50
40
30
20
10
0
300
400
500
600
700
800
900
1000
1100
Wavelength (nm)
DES CCD design has already been used on telescopes in small numbers (3)
SNAP CCDs are the next generation, optimized for space
B. Flaugher P5 April 2006
5
CCD Fabrication and Packaging
Follow LBNL business model developed for SNAP:
•
Foundry delivers partially processed wafers to LBNL
(~650 microns thick)
•
LBNL finishes wafers (250 microns thick), tests, dices
(production rate 5 wafers/month)
DES Wafers – June 2005!
FNAL builds up the CCD packages and tests CCD – will
match CCD delivery rate
Preconceptual R&D:
•
36 Eng. grade 2kx4k CCDs in hand
•
used to develop focal plane packages, characterize
CCD performance, test CCD readout electronics
•
Expect 20 more in July 06
Potential Science grade devices expected in Oct. 06
FY07: establish CCD processing and packaging yield
– preliminary est. 25% yield (SNAP devices)
– implies 18 months and $1.6M for 70 good devices
– CCD yield is a cost and schedule driver
B. Flaugher P5 April 2006
6
Front End Electronics: CCD Readout
• FNAL, Barcelona, Madrid, UIUC
• Opportunity for large international
contribution: Spanish consortium has
submitted a proposal to their funding
agencies (~$2M). Would provide
~$0.5M for the production FEE.
Filter and
shutter controls
3 operational CCD
testing setups
LN2 Dewars
Readout racks
• Status:
– UIUC funds used to purchase
prototype readout systems
– have already achieved 6.5e noise
at 160kpix/sec,
– have a design that fits in 3 temp.
controlled crates in PF cage
– need to test readout of multiple
CCDs
Part of Fermilab Team in the testing lab
7
B. Flaugher P5 April 2006
Camera Vessel Prototype
10 slot thermally controlled
crate for CCD readout electronics
Cryo and Vacuum controls
Feed-through board for CCD signals
Focal plane
Full size prototype is being
built by U. Chicago. It will be
ready for CCDs this summer
and will be used to test multiCCD readout
B. Flaugher P5 April 2006
8
Survey Image System
Process Integration (SISPI)
U Illinois-HEP (J. Thaler) is leading the SISPI development
- similar to HEP-DAQ systems
CTIO will upgrade the
Telescope Control
System (TCS)
Data Management (DM):
U. Illinois-Astro/NCSA
B. Flaugher P5 April 2006
9
Optical Corrector Design
• Preliminary Design complete (UMich, FNAL, UCL)
– Image quality fwhm: ~ 0.33” (<0.4” required)
• Opportunity now for international collaboration:
– March 05, the UK Consortium submitted a proposal
to PPARC to lead the procurement of the optics
5 elements, fused silica
Dewar
window
C4
filter
– 1.47 M pound proposal to cover cost of polishing,
mounting, and alignment of the lenses in the barrel
– P. Doel at U. College London Optical Science Lab
will manage the procurement and fabrication
C3
C2
• March 2006, PPARC Council announced that it
“will seek participation in DES”
• Additional UK funding ($0.5M ) available through
Portsmouth (SRIF3): ~60% of the blanks
• US University funding could cover the rest.
• Procurement of the optics is ~2 years
• CRITICAL PATH
C1 has
940 mm
diameter
B. Flaugher P5 April 2006
10
DES Filters
Dark Energy Camera Filters
100.0000
90.0000
80.0000
%Transmittance
• U. Michigan will
– handle procurement and testing of
the filters
– match SDSS – g,r,i,z and
introduce a well defined cut-off at
high wavelength
– design and fabricate or procure a
combined filter changer and
shutter
70.0000
60.0000
50.0000
40.0000
30.0000
20.0000
10.0000
0.0000
300
400
500
600
700
800
900
1000
1100
1200
Wavelength
925nm
775nm
635nm
475nm
Filter changer will be a cartridge
system similar to PanStarrs design
B. Flaugher P5 April 2006
11
=
Image Quality:
Elliptical PSF
DES
optics
*
Circular PSF
Primary Mirror Support cell
Blanco Primary mirror
– excellent figure: D80 <0.25”
– radial position controlled by
counterweights
– Upgrades in Oct. 05
reduced the mirror motion
from ~1.5mm to < 0.4mm,
further reduction expected
3 Hard Points
33 Pressure Pads
An observed PSF pattern
Ray tracing model of existing
mirror, camera and corrector
reproduces PSF patterns seen
in data
whisker
24 Radial Supports
optical model
In this model the primary was
misaligned by 0.2 mm x and 0.7 mm y
Primary Mirror
B. Flaugher P5 April 2006
12
CTIO and the Blanco Telescope
• Have produced some of the best Weak Lensing measurements
• Cerro Tololo Image Quality Sept-Feb
– site median PSF 0.65” FWHM
– Prime Focus (PF) delivered median PSF 0.9” FWHM (used in all DES constraint
forecasts)
• DES/CTIO upgrades will stabilize the PSF and should improve
the median FWHM:
• DES
– focus and alignment sensors on focal plane: generate focus and lateral
alignment information with each image
– active control of camera position: hexapods will provide focus and lateral
adjustments of corrector + camera system
– improved thermal environment: heat dissipation will be actively controlled
• CTIO is planning additional upgrades to the radial supports
B. Flaugher P5 April 2006
13
DES Simulations Feed DM Challenges
• 2004 Level 0 Image Simulations → DM Challenge 0: Done!
– Reformatted SDSS data used to simulate DES images
• 2005-06 Level 1 Catalog &Image Sim. → DM Chal. 1: Done!
– 500 sq. deg. catalog; 500 GB of images; FNAL and UChicago computing used
• 2006-07 Level 2 Catalog and Image Sim. In progress
– 5000 sq. deg. catalog; 5 TB of images
– FermiGrid & MareNostrum SuperComputer (Barcelona)
– Higher resolution N-body simulation, more realistic galaxy properties, and more
sophisticated atmosphere and instrument models (noise, ghosts)
– Recover input cosmology from catalogs using 4 DES key project methods
• 2007-8 Level 3 Catalog and Image Simulations
– Suite of full-DES catalogs (i.e., different input cosmologies)
– Synergy with DOE SciDAC proposal (with many DES collaborators) to produce
large cosmological simulations for dark energy studies
– 1 year of DES imaging data
– Recovery of input cosmologies from catalogs and images
– Stress test of full data processing system
B. Flaugher P5 April 2006
14
DECam critical paths: CCDs & Optics
CCDs:
•
LBNL can deliver CCDs at a rate of 20/month after 3 month startup
•
We need 70 CCDs for the FP including spares
•
Preliminary yield estimate of 25% implies ~18 months
•
Cost is ~$23k/wafer, 25% yield implies $1.6M
•
Construction start of Oct. 07 implies last CCD is finished March ’09
• Install last CCD and test full camera ~ 2 months
• Ready to ship to Chile ~ May 09
Optics:
•
Blanks ~ $0.8M , 8 month delivery
•
Polishing ~ $1.5M, 18 month delivery
•
Assembly and alignment into corrector ~ 6 months
• Ready to ship to Chile ~ 2.75 yrs after procurement begins (Oct. 06 → May 09)
B. Flaugher P5 April 2006
15
DECam Schedule
Dependent on DOE funding and international participation
Opportunity exists to capitalize on international interest in DES
• UK consortium would lead and fund procurement of optical elements
• Spanish Consortium would lead and fund procurement of production FEE
• Important for DOE to provide timely assurance that it intends to proceed
with DES (we will be ready technically to proceed with optics procurement
by Aug. 06)
A positive recommendation from P5 would help in this process
DOE Critical Decision Process schedule
• FY06 Preconceptual R&D; CD1 Paper review Sept.06
• FY07 R&D, CD2 Review March 07, CD3 Review Sept. 07
• FY08 MIE Construction start
• FY09-10: Assemble and test camera vessel and corrector
– Ship to Chile and install on Blanco , first DES observations Dec. 09
B. Flaugher P5 April 2006
16
DECam Funding Need Profile
(then yr $, Overhead included)
$M
ACTIVITY
R&D FUNDS (DOE HEP)
DOE HEP MIE FUNDS
EXTERNAL (Foreign + Univ. Non DOE)
TOTAL
FY07
FY08
FY09
FY10
TOTAL
R&D
CONST CONST CONST
4.1
4.1
5.8
4.9
1.9
12.6
1.0
2.0
0.0
0.0
3.0*
5.1
7.8
4.9
1.9
19.7
*Additional $1M in External funds in FY06,Total External Funds = $4M
• Total Project Cost: (FY07-FY10) = $19.7 M
• Total DOE Project cost: $16.7 M
• DOE Major Item of Equipment (MIE) total of $12.6 M
includes M&S equipment ($7M), technical labor ($5.6M)
and a total of ~35% contingency
B. Flaugher P5 April 2006
17
DES Data Management Project
U Illinois/NCSA DES DM Team
• U. Illinois and NCSA lead the DM project
– Joe Mohr (U. Illinois) is the project leader
– Cristina Beldica (NCSA) is the project manager
• DM System Requirements
– Reliably transfer ~300GB/night for 525 nights from CTIO to
U.Illinois/National Center for Supercomputing Applications (NCSA)
– Automatically process data with built-in quality assurance
– Archive the data products and serve the processed data to collaboration
– Provide community access to the archive 1 year after images were collected
• DM Team
– U Illinois/NCSA, Fermilab and NOAO
– Additional DES collaborators
• Deliverables to DES and astronomical community
– DM System (High Performance Computing platforms and workstations)
Pipeline middleware
Astronomy modules
Catalog database
Image Archive
– Archived science ready DES data
B. Flaugher P5 April 2006
18
This grid-based, modular and flexible data management system
was deployed and tested in Data Challenge 1 (Oct ‘05-Jan ‘06) 19
B. Flaugher P5 April 2006
DM Schedule and Status
• Pursuing iterative development strategy ‘04-’09
– Yearly data challenges Oct-Jan ‘05-’08
– Development targets full delivery in 2009
DC1: base level system in place
DC2: data quality, stress test
DC3: deploy and test outside NCSA
DC4: final validation and stress test
Reduced, pseudo-color
DC1 Image
• Data Challenge 1 Results (Oct 1 ‘05-Jan 31 ‘06)
– DM system deployed and tested
– Automated reduction (500GB raw reduced
into 5TB)
– Catalogued and calibrated 50 million objects
– Confirmed photometry and astrometry
B. Flaugher P5 April 2006
20
DES DM Funding Sources
(then yr $, OH included)
DM System Construction (‘04-’09)
$5.1M
U Illinois/NCSA (seed funding, incl. 2.5FTE/yr)
$1.40M
NOAO (NOAO Science Archive, 1.5FTE/yr)
$1.00M
Fermilab* (1.5FTE/yr)
$1.00M
Total resources identified to date
$3.40M
Pending NSF DES CyberInfrastructure proposal
($1.70M)
* Contributions by Fermilab scientists (DOE supported)
• Model is for NSF to fund proposed Data Management system
B. Flaugher P5 April 2006
21
Conclusions
DES provides the next logical step in both
technology and science
– Builds on existing technology and infrastructure, and capitalizes on
collaboration’s experience with large DAQ systems, silicon vertex
detectors, and data handling
– 3 deg2 camera: x7 larger area and x7 faster readout than existing
Mosaic camera on the Blanco
– 1PB total processed images available to the public; data released 1
year after images taken
– Development and implementation of data analysis techniques for
photo-z’s, cluster masses, weak lensing, baryon oscillations, and
supernovae are the next steps toward the science of the Stage IV
projects of the future (LSST, SNAP)
B. Flaugher P5 April 2006
22
extras
B. Flaugher P5 April 2006
23
CCD Requirements
LBNL CCD performance
Pixel array
2048 4096 pixels
Pixel size
15 m 15 m
<QE (400-700 nm)>
~70%
<QE (700-900 nm)>
~90%
<QE (900-1000 nm)>
~60%
Full well capacity
170,000 eDark current
2 e-/hr/pixel at –150oC
Persistence
Erase mechanism
Read noise
7 e- @ 250 kpixel/s
Charge Transfer Inefficiency
< 10-6
Charge diffusion
8 m
Linearity
Better than 1%
DECam requirements/
Reference Design
2048 4096 pixels
15 m 15 m (nominal)
>60%
>80%
>50% at 1000 nm
>130,000 e<~25 e-/hr/pixel
Erase mechanism
< 10 e<10-5
< 10 m
1%
B. Flaugher P5 April 2006
24
CCD testing results
Horizontal Charge Transfer Inefficiency for different voltages
•
Good news: with 5 thin CCDs
tested it looks likely that they can
be grouped together
( need more statistics)
•
Have all the testing infrastructure
running – now need more
analysis!
B. Flaugher P5 April 2006
25
Side view
•
B. Flaugher P5 April 2006
26
Front view
•
B. Flaugher P5 April 2006
27
Isometric view camera end
•
B. Flaugher P5 April 2006
28