USM Photometric Redshifts for Astro-wise

1
USM Photometric Redshiftsfor Astro-wise
R. Bender, A. Gabasch, M. Neeser, R. Saglia, J.
Snigula
Universitätssternwarte München Ludwig-Maximillians
-Universität
2
Introduction

• Photometric Redshifts
• deducing redshifts from multiple-band optical
  and near
• infrared imaging (poor mans spectroscopy)
• Scientific drivers
• Source identifications and redshifts
• Luminosity functions
• Star formation histories
• Large scale structures
• Cluster searches

• An obvious scientific product for the database
  catalogues

3
Spectral Energy Distributions (model input)
Galaxies
Stars
20 SEDs
4
Method

• Filter curves
• convolved with
• detectors

• Observed flux
• for each source

• SEDs
• convolved with
• filters
• stepped in redshift

5
Assigning a redshift and SED to each source
6
Final SED/redshift fit
FDF 2893
7
FDF 2367
8
FDF 914
9
Comparison with zspec
200 FDF spectra
10
Limitations of this method

• Requires adequate spectral coverage (ie. at least
  4 filters)
• Existence of degeneracies in SEDs at some
  redshifts
• SED input library inadequate to accurately map
  the coolest stars
• Ids and redshifts for AGNs must be done
  separately from
• galaxies

11
FDF 4940
12
FDF 2497
13
Integration into Astro-Wise Pipeline

• Envision two modes of operation
• automatic redshifts and source identification
  from catalogue
• colours assuming given default settings (filters,
  SEDs) and
• with output zphot, SED, probability, and
  errors.
• interactive mode with user defined parameters
  (SEDs,
• zrange, Mrange ) with simple plotting facilities
  and filter
• convolution routines.

14
Integration into Astro-Wise Pipeline
Class Photred
Persistent class PhotredConfig()
persistent SED models
model errors filter
convolution seeing factors
filter weight (SED error in given
filter / bad filter value) gt each object
assigned z1, z2, MB (persistent) Dz1,
Dz2 P1, P2 C1, C2 model1,
model2
15
Integration into Astro-Wise Pipeline
Open crucial issues 1/ class definitions 2/
reliable, consistent photometric redshifts can
only be achieved with photometric and PSF
uniformity across filter sets. (ie. PSF
homogenization across all filters).
16
Present Implementation of Photometric Redshift
Routine

• fortran routines to compute chi-square
  minimization and
• redshift probability function
• super mongo routines to display output, with a
  large
• number of user defined parameters
• Munics interactive source selection