Outline

1
Outline

• Definition of supported science areas
• Basic Data Products
• Figures of Merit
• Implications for System Design

2
Science areas

• Pan-STARRS will enable many science goals
• Difficult to predict what will be most
  interesting 2007-2017
  
• However, science objectives can be classified as
• Study of the static, stationary sky.
• Study of the time variation of the sky.
• Transient science - things that go bump in the
  night - may have temporal structure but have
  finite lifetime.
  
• Variability science --- persistent objects whose
  brightness varies with time.
  
• Moving object science --- minor planets,
  parallax and proper motion measurements.
  
• All of these science areas are to be enabled by
  Pan-STARRS.

3
Sub-division of science goals

• Area coverage
• All-sky
• Medium deep
• Ultra-Deep
• Cadence
• Bright vs Faint (compared to sky)
• Requires different data processing, storage

4
Data Storage Costs

• PS will make 1016-1017 measurements
• These constrain P(intensity)
• Integrated over windows in wavelength, time
• Convolved with the PSF
• Storing all the bits is prohibitively expensive
• On the order of 40M for hardware alone
• Instead we will
• Process the data on the fly for transient, moving
  object science
  
• Store accumulated images of the static sky
• Store copious amounts of catalog data

5
Basic Data Products

• Instrumental catalogs
• Objects detected and photometered in the
  sky-subtracted and calibrated OTA images.
  
• Positions, flux densities and covariances in
  instrumental units
  
• Best estimate of sky coords, true magnitudes.
• Brighter objects will have more attributes.
• A stream of difference images and catalogs
• Images to be kept for a limited time (1 month)
  and then be discarded.
  
• Catalogs of detections down to 3-sigma
  significance level to be preserved.
  
• Accumulated images.
• Sufficient statistic for static sky science.

6
Basic data products (cont)

• Bright object catalogs
• Postage stamps
• Allow de-aliasing and de-convolution etc.
• Faint object catalogs
• Photometric properties of objects detected in
  static sky images
  
• Supports time variability studies

7
Requirements vs Figures of Merit

• Engineers want specific requirements
• Artificial - all or nothing - view of science
  output
  
• Scientists want cost as a function of all design
  parameters
  
• Also unrealistic
• Middle road figure(s) of merit
• Inverse of time taken to achieve some goal
• Each science objective has well defined FOM
• Design should maximize (weighted) FOM

8
Point Source Detection FOM

• The collecting area of the telescope.
• The field of view.
• The width of the filter function.
• Duty cycle
• Fraction of on-target integration time.
• The inverse of the sky intensity plus an
  effective sky intensity from read noise.
  
• The zeroth moment of the square of the optical
  transfer function (OTF)

9
Factors in the OTF

• The seeing (this is coupled to the OT guiding).
• The pupil.
• Aberrations and fine-scale mirror roughness.
• The finite source pixel size.
• Charge diffusion.
• For measurements made from images mapped to the
  sky, add
  
• The pixel mapping algorithm.
• The pixel size for sampling the sky.
• Should really multiply all factors
• Simple approach when seeing dominates
• Add second central moments, or
• Compute effective "Fried length"

10
Point Source Astrometry FOM

• Similar to point source detection
• Contains second moment of squared OTF
• Places premium on image quality
• But we still need to detect objects so this
  should
  
• Should not be the primary FOM

11
Weak Lensing FOM

• WL FOM is complicated
• See POI-Book simulations
• WL requires good image quality
• For detection of faint/small objects
• Measurement of moments in high SB regions
• FOM might seem to prefer broad filter, but
• Systematics worse (as square of bandwidth)
• Desire for crude photo-z
• WL is likely systematics limited
• Favors lots of images rather than long
  integrations

12
Transient Science FOM

• Usually point sources detected in difference
  images
  
• Point source FOM applies
• Assumes photon counting limited sky subtraction.
• Transient searches favor broad/shallow over
  narrow/deep.
  
• This implies lots of images/high compute load
• Requires multiple quasi-simultaneous images to
  exclude false detections
  
• Multiple exposures should be dithered to fill
  gaps.

13
Moving Object Detection

• Point source detection FOM applies for slow
  moving objects
  
• Trailing losses become important for faster
  objects.
  
• Objects should not move many resolution elements
  in one exposure time
  
• Severity has been exaggerated by LSST
• For PHAs that Pan-STARRS will detect, trailing
  loss is 20 for nominal 30s integration
  
• PHA searches are not truly etendue limited

14
Moving object FOM

• True FOM for PHA searches is the amount by which
  the next observation reduces the future fatality
  rate from collisions.
  
• Highly inhomogeneous
• Concentrated toward the ecliptic plane
• Sweet spots
• Color information is not important for PHA
  detection (or KBO science)
  
• Very broad "solar system" filter to be used for
  efficiency

15
Image Multiplicity

• FOMs are somewhat naive
• Assume photon counting is the limit
• Many applications are really limited by
  systematic errors.
  
• Favors multiple short integrations rather than
  long.
  
• With many observations one can hope to model and
  correct for systematics.
  
• Alternatively, if effects are really
  unpredictable, then they average out like
  1/root(N)
  
• For static sky, naive FOM favors long
  integrations, but with highly diminishing
  returns
  
• If any weight is given to averaging systematics
  then much shorter integrations are favored.

16
Uniformity of Data

• Figures of merit do not account for desirability
  of uniform data
  
• Proposed requirements
• Vignetting
• Detector QE variation
• Detector color term variation
• Plate scale variation
• Could be relaxed considerably, but benchmark
• design easily meets this
• Observation scheduler to be motivated to give
  uniform coverage of the sky for each survey
  region.

17
Summary of FOM requirements

• The scientific requirements for Pan-STARRS
  physical design are
  
• It should maximize the point source detection FOM
  subject to the constraints of budget and
  schedule, except that
  
• Individual exposures should not be longer than
  the time needed to become sky noise dominated (by
  a factor 5 in variance) and the read-out rate
  should be adjusted according to the filter in use.

18
Design Implications

• Filter choice
• Site selection
• Telescope design
• Sky coverage
• Stray light, baffling and ghost images
• Control of mirror surfaces
• Focal plane design
• Image Processing Pipeline

19
Filter Choice

• The choice of filter is very much science
  application dependent
  
• Most proposed applications require standard broad
  band filters.
  
• Compatibility with other recent and ongoing
  surveys favors
  
• mimicking SDSS g, r, i, z passbands.
• There is also demand for a y-band filter.
• A u-band filter may be offered but design of
  telescope and optics is to be optimized for g and
  redder.
  
• Leaks outside of the pass bands to be 1 max.
• The exception is solar system work such as PHA
  searches that do not benefit from color
  information and for which it is proposed to use a
  wider filter extending from 450-820nm.

20
Site Selection

• Key factors
• Good seeing
• Low sky background
• Fraction of clear nights
• Trivially factored into FOM budget
• Thickness of the boundary layer is also
  interesting
  
• Evidence that a good deal of the image
  degradation on MK arises very close to the
  ground.
  
• The phase errors from lowest 5-10 meters can be
  taken out by active control of the primary at a
  few Hz.
  
• FOM gains are somewhat speculative, but will
  become known from WFS data.
  
• Requirements
• Analysis of sky brightness, seeing and weather
  data.
  
• See also Site permitting requirements

21
Telescope Design

• Type of telescope
• Various factors favor distributed apertures
• OT guiding
• Removal of diffraction spikes, ghosts
• Single aperture design has overheads for
  transient science.
  
• Equatorial marginally preferred.
• Telescope Aperture
• To be maximized, subject to cost and schedule
  constraints.
  
• Benchmark design meets the Decadal Review spec.
• FOV
• Enters linearly in FOM, but very hard to go
  beyond 3 degree diameter without problems with
  image quality.
  
• Slewing, settling and acquisition time
• Should not exceed 6s.

22
Telescope Design (cont)

• Focal Length
• Critical factor linking the telescope design and
  the CCD design.
  
• Beletic panel recommended delaying decision on
  telescope design until performance of CCDs is
  established
  
• Would extend construction time by 1 year
• However, while the performance of the telescope
  depends critically on the charge diffusion (pixel
  size is sub-dominant), the penalty for making the
  wrong' choice of focal length is very small.
• Requirement
• Focal length 8m
• Can proceed with telescope procurement decoupled
  from the CCD development.

23
Telescope Design (cont)

• Sky Coverage
• We need up to 70 degree zenith distance
• Sweet spots
• Coverage of N-pole.
• Stray Light, Baffling and Ghost Images
• Fully baffled design is needed.
• Ghosting analysis required so that ghost images
  can be excised in the pipeline.
  
• Monitoring for sources of scattered light is
  highly desirable.
  
• Control of Mirror Surfaces
• Required
• Design should allow (possible future upgrade to)
  active control at 3Hz to allow correction for
  boundary layer seeing.

24
Focal Plane Design

• Degradation of FOM by pixel size and charge
  diffusion to be less that 25.
  
• For 0".6 seeing, and focal length of 8m, this is
  achievable with
  
• 12 micron pixel size (8 degradation) and
• 5 micron charge diffusion (17 degradation).
• Good QE in g-y bands.
• QE uniformity (
• Color term variation (
• Variation of PSF width across the focal plane
  (
• Discontinuities across device boundaries to be
  minimized by careful metrology.

25
Focal plane design (cont)

• Wave front sensing that samples across the focal
  plane is desirable.
  
• Pin-hole camera to detect sources of stray light
  is desirable.
  
• See Detector System Requirements

26
Image Processing Pipeline

• To keep up with images at a rate of up to 1 per
  minimum exposure time of 30s.
  
• To deliver transient results within several times
  minimum exposure time.
  
• To support
• Calibration program.
• Pre-Survey.
• Generation of astrometric reference catalog.
• Generation of all-sky photometric standards in
  all bands.
  
• Survey(s) proper.

27
Image Processing Pipeline (cont)

• To provide data products listed above
• Instrumental catalogs required for
• Operation of pipeline
• Applications with premium on precise
  photometry/astrometry
  
• To provide observation/processing history
  meta-data archive.
  
• Variation of pixel scale across tangent plane to
  be
• Basic science client support (access to image and
  basic catalog data).