Large Surveys

1
Large Surveys the VO

• Konrad Kuijken
• Leiden Observatory

2
Why me?

• Dont know much about the VO

• OUTLINE
• Describe survey
• What can survey do for VO?
• What can VO do for surveys?

3
The KIDS/VIKING survey
LEIDEN GRONINGEN MUNCHEN PARIS NAPLES BONN
EDINBURGH CAMBRIDGE IMPERIAL
4

• 1500 sq.deg. in ugriZYJHK
• 2000 sq.deg. in i (UKIDSS YJHK)

5
(No Transcript)
6
Lensing / photo-z survey

• VST/OmegaCAM 1 sq deg, 2.6m telescope
• VISTA/VISTACAM 0.6 sqdeg, 4m telescope
• 1500 sq.deg. of ugri (400n VST) ZYJHK
  (200n VISTA)
• Deeper in r, with good seeing
• VST 2m deeper than SDSS (1m shallower than
  CFHTLS)
• VISTA 1.5m deeper than UKIDSS

7
KIDS vs. SDSS, CFHTLS
(M.Neeser)
IR coverage!
8
KIDS Science drivers

• Statistical astronomy
• Large-scale structure - power spectrum c.
• Gravitational lensing - dark matter distribution
• Dark energy evolution
• Galaxy properties as function of environment
• Star counts in the Milky Way
• New satellites
• Halo streams
• Requires high fidelity statistics incl. selection
  function

• Needles (diamonds) in haystacks
• High-z quasars, EROs
• High proper-motion objects (nearby faint stars)
• Gravitational lenses
• Dark masses
• Cosmic strings
• Require high fidelity in survey products
• Dont want candidate lists of 1,000,000 artefacts
  !

9
Astro-WISE

• Astronomical Wide-field Imaging System for Europe
• Software / database environment for KIDS
  processing
• Multi-site, federated database
• Query-driven (re)-processing of calibration and
  image data
• Traceable products
• See talk by Verdoes-Kleijn Monday 1200

10
Exploiting large surveys in the VO

• Grab-bag archives
• Gold-mines, especially if interested in a
  particular source or class of sources
• Really should save MUCH telescope time
• VO data browser is great! (eg astrogrid
  astroscope)
• Large surveys deliver huge, homogeneous data sets
• Much of the hard work is done by survey teams
• VO user browses and searches
• E.g., Sloan, IRAS, ROSAT, FIRST, WENSS, UKIDSS,
• Selection functions of large surveys clear(er),
  but !

11
Focal plane geometry of OmegaCAM (typical of
large CCD mosaics) Plus smaller effects bad
pixels, QE variations, Leaves imprint!
12
1 degree
100
60
80
13
Upper limits

• Big problem of catalogue searches non-detections
• E.g. find optical drop-outs (zgt7 QSOs)
• Set threshold ZY lt 110
• 20-? significance limit in Y catalogue
• Many real (e.g., 3-?) sources not in Z catalogue
• Need to search Z image around each candidate
• Automated list-driven photometry query
• Typically done within a multicolour survey, but
  in VO, combining surveys is the goal

14
Useful Meta-data?

• Obviously useful coordinate system, filter
  profile, sensitivity limit,
• Obviously useless(?) CCD temperature, telescope
  operator name,
• Part of quality control of individual surveys
• Useful but hard to describe
• selection algorithm of the pointings
• reason for filter choice
• Delicate
• Magnitude type, corrections for aperture, PSF,
  colour terms, atmosphere,

15
Random catalogues

• Field selection
• eg parallel exposures with HST/WFPC2 tend to be
  20 from galaxy (or star) cluster cores. Need to
  know redshift of cluster and excise it for
  unbiased source samples
• Filter selection
• eg VLT/FORS narrow-band exposures to catch Ly-?
  emitters around radio galaxies biased
• Epoch selection
• eg supernova frequency measurement excise SN
  follow-up observations
• Only belgian-proof remedy area-complete surveys

Bias!!
Bias!!
Bias!!
16
KIDS VO

• Obvious links
• Tying together optical and near-IR
• Linking spectroscopy with images
• Complex queries that test completeness
• Quality control filtering out data with
  different seeing, extinction,
• Dissemination of data products

17
What can KIDS do for the VO?

• Provide good data!

18
Accurate colours

• Photometric redshifts, CMDs, variability,
• All require accurate flux comparisons
• Point sources
• Total, PSF-fitted fluxes can be compared directly
• Extended sources
• Total fluxes ill-defined, noisy
• Aperture fluxes affected by PSF
• Colours can be aperture-dependent

19
Extended sources

• Aperture colours
• Need to correct for PSF differences
• Usual approach
• Reconvolve pixels to poorest seeing, then
• Compute fluxes in similar apertures
• Would like to be able to do this in VO!

Intrinsic Good seeing Bad seeing
20
PSF compensation

• Simplest approach
• Find, pull out images
• Measure PSF on each
• Construct kernel, convolve
• Perform matched-aperture photometry
• Laborious within scope of VO !?

21
GaaP fluxes
(Kuijken 2006)

• Weighted flux that would be observed with a
  Gaussian aperture and PSF
• Can be catalogued (for range of aperture radii)
• Avoids the need for pixel processing by VO user
• Can be implemented as list-driven photometry
• To check the non-detections !!!
• Algorithm based on shapelets

22
Shapelets

• Gaussian x polynomials
• Orthonormal basis Gauss-hermite series
• Nice transformation properties (little mixing)
  under
• Translation
• Rotation
• Shear
• Magnification
• Reasonable approximations to PSF galaxies (?)
• Closed-form expressions for convolutions

(Refregier 2003)
23
PSF mapping
24
(No Transcript)
25
GaaP fluxes with shapelets

• Express sources and PSF as shapelet series
• Express PSF convolution as a matrix operator on
  shapelet coefs
• Deconvolve source by PSF ? S(r)
• Compute Fq (1/2) ? S(r) exp (-r2/4q2)
• Include correction terms based on residuals to
  shapelet fits of source and PSF.

26
Many simulations with different PSF galaxy type
size
Simple shapelets-based recipe
Shapelets pixel - based recipe
Systematic error lt 1
Aperture radius / gaussian radius
27
GaaP flux Fq

• PSF-independent photometric quantity
• Ratios give real colours!
• Can be computed accurately over a range of x 2 in
  PSF size, from image and PSF alone
• Photon statistics propagate straightforwardly
• No need for PSF matching between exposures to get
  accurate colours

28
What can the VO do for KIDS?

• (Scientifically and operationally)

29
Design of the survey

• Field choice
• Other maps
• Spectral surveys
• Extinction, cirrus
• Was done by hand for KIDS (basically follows 2dF
  Galaxy redshift survey), but nice role for VO?
  (Google sky?)

30
Quality control

• Comparison with other surveys astrometry,
  photometry, completeness,
• But whose quality are you controlling?
• Calibration with spectroscopic surveys
• Extinction corrections from colour-colour
  diagrams,

31
Merging of surveys

• KIDS (optical) VIKING (near-IR) 2dfGRS
  (spectra) UKIDSS (near-IR) SDSS
  (opticalspec)
• Data processed and archived in different places
• Visualization of combined survey
• Combined analysis
• eg, detect mass overdensities from lensing in
  optical, then check the IR data for red sequence
  galaxies
• Check whether objects move, vary
• Many tools in hand, but VO addresses similar
  questions and will be very useful

32
KIDS Science drivers

• Statistical astronomy
• Large-scale structure - power spectrum c.
• Gravitational lensing - dark matter distribution
• Dark energy evolution
• Galaxy properties as function of environment
• Star counts in the Milky Way
• New satellites
• Halo streams
• Requires high fidelity statistics incl. selection
  function

• Needles (diamonds) in haystacks
• High-z quasars, EROs
• High proper-motion objects (nearby faint stars)
• Gravitational lenses
• Dark masses
• Cosmic strings
• Require high fidelity in survey products
• Dont want candidate lists of 1,000,000 artefacts
  !

33
Summary

• KIDS/VIKING survey is coming only the beginning
• Large homogeneous surveys are great for VO to be
  useful

• In context of planning large surveys
• VO strength huge variety of data
• Needs list-driven photometry
• Useful to implement standard aperture fluxes
  (GaaP)

• In context of analysing large surveys
• Visualization of existing data
• Naturally work with distributed data
• Calibration (check) tool
• VO weakness g-i-g-o problem
• Meta-data complex