Title: Redshift estimation Photometry
1Photometric redshift estimation
2Standard Methods for photometric-redshift
estimation
Template-fitting method Empirical
method Hybrid methods
3Template-fitting method
- Begins with library of template spectra
- (empirical or modeled spectra should cover all
spectral type) - Template spectra are corrected for redshift and
Galactic extinction - Template colors are calculated
- (convolving template spectra and filter
function) - Template colors are compared to observed one by
minimizing -
4Empirical method
- Starts with galaxy sample with known zsp and
- magnitude m (so called training set)
- Relationship zsp zsp(m) is determined
- having m, zph is determined using zsp zsp(m)
- as a calibration curve
- example (Connolly et al., 1995)
5Advantages Disadvantages
Beside, Template-fitting method provide both, z
and T compared to Empirical method that
provides only z
6Hybrid methods(Csabai, Budavari)
Takes advantage of positive sides of the
Empirical method and Template Fitting method
use training set to optimize for the shape of
spectral templates to better match the SEDs of
the galaxy
7Typical Results Errors(Csabai et al 2003)
Empirical
Template fit
Hybrid
8Main Sources of uncertainties
- Color/redshift, age, dust, morphology degeneracy
- Lyman-break/Balmer-break degeneracy for high z
- Errors in photometric measurements
- Spectral template incompleteness
9Accuracy Improvement
- Generally, we expect improvement when
- Breaking (reduce) degeneracies
- Working with better spectral template
- (more reliable in the UV and more complete)
10This Work
In this work we try to include morphological infor
mation to break the color/morphology degeneracy
using Empirical method Strategy 1. Find some
photometrical parameter (combination) from the
SDSS database that best correlates with
morphology (morphological parameter) 2. To use
it for photometrical redshift estimation
improvement
11Morphological Classification (Fukugita et al.
2003)
- For morphological parameter we used parameter T
visually obtained by Fukugita et al. - Visual Classification of 2253 galaxies using
CCD images in g band - 3 researchers were involved in visual
classification - dispersion ?T 0.4 ( ?T(RC3) 1. 8
photo-plates used ) - rp lt 16 - visual inspection limit
- 1866 galaxies have been spectroscopically
observed - 0.001 lt z lt 0.14
- CLASSES
- T 0 1 2 3 4 5 6
- Type E S0 Sa Sb Sc Sd
Irr
12Properties
Redshift distribution
Petrosian magnitude distribution
13Apparent Sizes vs. Redshift
14Correlation
We correlate morphological parameter of Fukugita
et al. (T) with 1. All photo parameters from
SDSS 2. Photo parameters already used for morph.
classification (concentration index, color
index) 3. Combination of them
15Illustrative results
Generally, we have non linear dependences between
photometrical rarameters and morphological
parameter T (nonlinear regression) Best
correlations are find for CI, colors, fracDev,
eClass, tph
16Results
Results of using photo-parameter that best
correlates with morphology in Empirical method
for z estimation
Magnitudes are used in The Empirical Method
Instead of magnitudes, colors are used in the
Empirical Method
17- We have also tried to improve photometric
redshift estimation by - Using directly morphological parameter T (2)
- To increase the number of photometrical
parameters in the empirical relation (4) - Leaving out photometrical parameters in u and z
(larger measurement errors) (2)
18Conclusion
- All photometrical parameters from SDSS are
correlated with - Fukugitas morphological parameters T
- - CI, colors, fracDev correlate best
- Including morphological information in the
photometrical - redshift estimation by empirical method improves
z up to 4-5 - (exeption is eClass with 14 - 15 but it is not
photometrical parameter) - Living out photometrical parameters with larger
measurement error - (u, z bands) slightly improver z.
- Using more than one photometrical parameter we
have improvement - of several
- next step Including morphology in
Template-fitting method