1 / 33

PISCO Redshifts: Simulations

PISCO Redshifts: Simulations. Will High Harvard University December 2006. The LPPC Contingent. Chris Stubbs (Harvard Physics, Astronomy) Tony Stark (SAO) James Battat (Harvard Astronomy) Passbands Will High (Harvard Physics) high@physics.harvard.edu Photometric redshifts.

xannon
Télécharger la présentation

PISCO Redshifts: Simulations

An Image/Link below is provided (as is) to download presentation Download Policy: Content on the Website is provided to you AS IS for your information and personal use and may not be sold / licensed / shared on other websites without getting consent from its author. Content is provided to you AS IS for your information and personal use only. Download presentation by click this link. While downloading, if for some reason you are not able to download a presentation, the publisher may have deleted the file from their server. During download, if you can't get a presentation, the file might be deleted by the publisher.

E N D

Presentation Transcript


  1. PISCO Redshifts:Simulations Will High Harvard University December 2006

  2. The LPPC Contingent Chris Stubbs (Harvard Physics, Astronomy) Tony Stark (SAO) James Battat (Harvard Astronomy) Passbands Will High (Harvard Physics) high@physics.harvard.edu Photometric redshifts

  3. The Problem • Measure the distance to as many galaxy clusters as possible, as efficiently as possible • Redshift == distance given knowledge (read: assumptions) of geometry of universe • Solution already exists: • The photometric redshift principle: Use • broadband, differential photometry • presumed knowledge of convenient source spectra • and knowledge of filter transmission curves • … to arrive at redshifts z to |dz| < 0.1 • Spectra of Luminous Red Galaxies increase ~monotonically in well understood way… 4000A break • Our problem: we know our filters’ bandwidths and central wavelengths change as a function of field position! How does this affect redshift recovery? • Our tool: simulated observations of LRGs using expected filter curves

  4. PISCO Filters Light Using Dichroic Beamsplitters Instead of Glass D3 D2 D1

  5. PISCO Filters Light Using Dichroic Beamsplitters Instead of Glass D3 D2 D1

  6. Dichroic Curves Solid line = transmit Dashed line = reflect

  7. Rule of thumb: Dichroics pass red

  8. No Photons Here!

  9. No Photons Here!

  10. PISCO Bandpassesat the Focal Planes Multiply those curves appropriately

  11. With CCD Quantum Efficiency • Not yet accounted for: • Glass & other optics • Atmosphere • Possible y band at 1 micron Multiply by MIT/LL high resistivity CCD transmission curves Transmission r i g z Wavelength (Angstrom)

  12. Dichroic Info • Company: Barr Associates • Cost: • BS I $12k • BS II $11k • BS III $11k • Delivery: 12-14 weeks (including substrate procurement time of 8-10 weeks) • Barr can make effective index of refraction as high as 2.0 to minimize the effect of angle of incidence on the transmission/reflection properties (see next slide)

  13. CWL = cut-on wavelength Ratio of cwl at some angle  to cwl at 0 degrees. Increasing n (the effective index of refraction) decreases the effect of AOI (shallower slope)

  14. Position Dependence • Dichroics in highly collimated beam • Angle at dichroic → position in focal plane • But different angles have different cut-on wavelengths! Therefore… • Bandpasses depend on position in focal plane • Effective wavelength λeff changes • Bandwidth changes • Both ill behaved across field • Study how well we recover redshift assuming • Naïve bandpasses = transmission at center of field, assumed to hold everywhere • Correct bandpasses = true transmission, which changes with field position

  15. How We Measure Redshift • Observe redshifted Luminous Red Galaxy (LRG) spectrum with correct bandpasses • Estimate AB mags using naïve and correct bandpasses • Fit colors: LRG color ↔ LRG redshift • kcorrect • Use both naïve and correct bandpasses true LRG observed Photon Flux Density r i g z Wavelength (Angstrom)

  16. We Measure Photo-z’s:Naïve z in = 0.4 z out = 0.438 z in = 0.8 z out = 0.835 Black = output spectrum z in = 1.1 z out = 1.174 z in = 1.4 z out = 1.412 r i Gray = input spectrum z g

  17. We Measure Photo-z’s:Correct z in = 0.4 z out = 0.403 z in = 0.8 z out = 0.804 Black = output spectrum z in = 1.1 z out = 1.128 z in = 1.4 z out = 1.408 r i Gray = input spectrum z g

  18. Photo-z Errors:Naïve Underestimated z in = 0.4 err = 3.9e-2 z in = 0.8 err = 3.7e-2 Overestimated z in = 1.1 err = 1.3e-1 z in = 1.4 err = 1.1e-1

  19. Photo-z Errors:Correct Underestimated z in = 0.4 err = 3.1e-3 z in = 0.8 err = 6.2e-3 Overestimated z in = 1.1 err = 5.0e-2 z in = 1.4 err = 9.5e-2

  20. Worst Photo-z Errors i → z g → r r → i z → … Naive passband photo-z error (max absolute deviation) Correct passbands z in

  21. g AB Mag Errors:Naïve Brighter z in = 0.4 err = 8.3e-2 z in = 0.8 err = 1.3e-1 Dimmer z in = 1.1 err = 1.5e-1 z in = 1.4 err = 8.4e-2

  22. g AB Mag Errors:Correct Brighter z in = 0.4 err = 4.4e-2 z in = 0.8 err = 8.8e-2 Dimmer z in = 1.1 err = 1.0e-1 z in = 1.4 err = 4.0e-2

  23. Worst g AB Mag Errors i → z g → r r → i z → … g AB mag error (max absolute deviation) Naive passband Correct passbands z in

  24. g – r Color Errors:Naïve Bluer z in = 0.4 err = 3.7e-2 z in = 0.8 err = 8.9e-2 Redder z in = 1.1 err = 5.8e-2 z in = 1.4 err = 5.7e-2

  25. g – r Color Errors:Correct Bluer z in = 0.4 err = 1.2e-2 z in = 0.8 err = 4.9e-2 Redder z in = 1.1 err = 2.8e-2 z in = 1.4 err = 8.8e-2

  26. Worst g – r Color Errors i → z g → r r → i z → … g – r error (max absolute deviation) Naive passband Correct passbands z in

  27. g – i Color Errors:Naïve Bluer z in = 0.4 err = 1.3e-1 z in = 0.8 err = 1.4e-1 Redder z in = 1.1 err = 1.8e-1 z in = 1.4 err = 9.3e-2

  28. g – i Color Errors:Correct Bluer z in = 0.4 err = 3.8e-2 z in = 0.8 err = 4.9e-2 Redder z in = 1.1 err = 9.4e-2 z in = 1.4 err = 3.6e-2

  29. Worst g – i Color Errors i → z g → r r → i z → … g – i error (max absolute deviation) Naive passband Correct passbands z in

  30. Worst Color Errors NOTE THE SCATTER Naive passband g – r error Correct passband Wake et al, 2006 astro-ph/0607629 z in

  31. What’s Next? • Add y band? 1 micron, a la Pan-STARRS • Increase sophistication of photo-z module • More templates • Priors

  32. Summary • PISCO achieves broad optical bandpasses in an original way that is more efficient • Downside: bandpasses become weakly position dependent • Outlook for photo-z’s: • Hurts us most at z < 1.0, but in a clean way • At z > 1.0, 4000A break leaving bands, situation messier • Incorporate these facts into analysis (= calibrate) • Contact Will at high@physics.harvard.edu

More Related