R evised SWIRE photometric redshift

1. Revised SWIRE photometric redshift Author: Michael Rowan-Robinson et al.2012 Present: XuechenZheng Date: 04-17-2013

2. Introduction • SWIRE survey consisted of 49 deg2 of sky surveyed by Spitzer at 3.6, 4.5, 5.8, 8.0, 24.0, 70.0 and 160.0 μm. • Rowan-Robinson et al.(2008) reported the SWIRE Photometric Redshift Catalogue(SPRC) • New optical and near-infrared data from several surveys make it worthwhile revisiting SWIRE photometric redshifts • Additional bands together with the improved photometry resulted in a reduction in the number of catastrophic outliers and improved rmsvalues • The new catalogues comprise 1 009 607 redshifts, out of a total of 1 066 879 in the original SPRC.

3. Aperture corrections • Difficulty: the differences in distant keep us from applying same template for galaxies( integrated SED required) • Solutions: • A.use curve of growth fitted to photometry derived in a series of apertures of different sizes (example: Kron and Petrosianintegrated magnitudes, mag-auto provided in sextractor) • Results: uncertainty of the estimation(mainly comes from different contributions of sky photon noise) cause poor results for photo-z • B. start from photometry derived in a single small aperture in each band, and then apply an aperture correction derived in a single chosen band to all the bands.

4. Optical data • Define: (r is measured in a 2 arcsecaperture, WFC ‘aper2’, and have been PSF aperture corrected) • -5.0<delmag<-0.10, apply aperture correction • -0.10<delmag, considered to be PSs within uncertainties • -5.0>delmag, considered to be erroneous determinations of integrated magnitude • In the 2nd and 3rd conditions, aperture correction was set to zero.

5. Red symbols: Galaxies; Blue sybols: QSOs

6. For SDSS optical data: • no sign of any correlation of r(SDSS, model) − r(WFC, mag-auto) with • SDSS model magnitude is estimating approximately the same total magnitude as the WFC mag-auto • However, wide dispersion exists, which harms photo-z estimation • use WFC mag-auto, because WFC provides more data alone with more accurate magnitudes.

7. Other option for aperture correction • led to slightly worse photo-z result Stick to unless mag-auto is not available

8. Near infrared data • For 2MASS data: • For UKIDSS data: • They are well correlated to , but result in worse photo-z solution • Therefore, use as aperture correction.

10. Result: use as aperture work well in removing the correlation of color with

11. IRAC data (3.6μm)

12. UKIDSS-IRAC comparison shows problem that harms the photo-z estimation • Reason? • IRAC fluxes give good consistency with the aperture-corrected optical magnitudes • Therefore, problem lies with overestimation of the J, H and especially K brightness for extended galaxies • Solution: change the aperture correction • New problem: worse photo-z results • Suggestion: revise the templates in the near-infrared

13. Photometric Redshifts • Photometric redshifts method: • Two-pass template based on six galaxy (11 in the second pass) and three AGN templates. • First pass: empirical galaxy templates + star formation history • Second pass: four infrared SED types + dust extinction • Torus component prevents the use of near-infrared data before new approach developed

14. Comparison of SDSS and present sample

15. Comparison of SPRC and revised SPRC

16. Comparison of new catalogue and LePhare method(2006) in XMM

17. Results of QSOs • Fact 1: SPRC required an object to be flagged as stellar to consider a QSO template, which led to the missing of some QSOs • Fact 2: allow a QSO template option to all galaxies will cause mistakenly high redshifts to too many galaxies • Solution: allow the SDSS stellar flag to override the WFC flag where they disagree and this allows a few more quasars through

18. Salvato et al. introduced two innovations: • 1. track the variability of quasars and apply an appropriate correction to the photometry • 2.they use templates that include a range of contributions from AGN dust tori • Lack of information to track QSO variability, they explored the idea of adding a range of AGN dust tori strengths to our QSO templates during the second pass, and then using the 1.25–8 μm data in the redshift solution. • Comparison of the results: • For at least 11 photometric bands, reduced χ2 <3, r<21.5, rms=9.3%, outlier=9.3% • Salvato, with 30 bands, rms=1.2% for I<22.5, outlier=6.3%

19. Comparison of SPRC and revised SPRC for QSOs

20. Reasons for poor χ2 • Photometric redshift estimates get worse as χ2increases and some sources have very large χ2 • Main reasons: • 1. contamination by stars • 2. poor photometry • 3. QSOs misclassified as galaxies • 4. QSOs with too high dust extinction

21. Recognition and influence of stars

22. After the removal of stars, use color-color diagrams and SEDs plot to exam the rest sample with χ2>7 • Conclusions for QSOs: • 1. most of the χ2>7 objects are stars with slightly higher 3.6 μm fluxes than the stellar sequence • 2. A few are QSOs with too high dust extinction • 3. others are objects with poor photometry • Conclusions for galaxies: • 1. 10% are mistaken QSOs • 2. some are stars with slightly higher 3.6 μm fluxes than the stellar sequence • 3. rest are the problem of photometry

23. Color-color diagrams of QSOs and Galaxies

24. Color-color diagram and histogram for galaxies in EN1

25. Revised SPRC • Object amount: 818 555 • 3% of the SPRC objects did not find a match in fusion catalogue mainly because the latter omitted 24μm only source • Further 1% failed to achieved a redshift solution because of the lack of photometric bands or large χ2 • Along with objects in the area with no new data, the total number of redshifts is 1 009 607 • New catalogue delivers photometric redshift for 26 288 quasars

26. Left-hand: histogram of number of bands used in estimating photometric redshifts in revised catalogue (solid line) compared with SPRC (broken line). Right: histogram of redshifts in revised SPRC (red: galaxies, blue: QSOs, x 10)

27. Analysis of outliers

28. Discussion of the improvement Revised SPRC should be useful for improved studies of the infrared extragalactic population and surveys carried out in all SWIRE fields.

29. Reference • Rowan-Robinson et al., 2013, MNRAS, 428, 1958–1967 • Thanks!