Early Data Release

This PAUS Early Data Release (EDR) corresponds to PAUS data obtained in the COSMOS field, described and used in

The PAU Survey: Early demonstration of photometric redshifts (M.Eriksen, A.Alarcon et al. ) 2018, MNRAS 484, 4200–4215 , https://arxiv.org/abs/1809.0437

Please read this paper first before using the data. The EDR release table is in CosmoHub.

New to CosmoHub?
Register in 1 minute here:

Already in CosmoHub?
Go directly to the public catalogue:

DIRECT LINK

As part of this EDR, we also release, through CosmoHub, the PAU MillGas Lightcone galaxy catalog constructed from the semi-analytic galaxy formation model Galform, which was presented in Stothert et al. (2018). Find more information in the release page of the mock.

The EDR is a JOIN of 3 tables: COSMOS (or reference catalog), PAUDM_FORCED_APERTURE_COADD (this include narrow band fluxes and errors; v821) and BCNZ (this is the photo-z estimation described in Eriksen etal 2018; v866).

Here is a bit more detail about the columns in each table.

COSMOS: is the reference catalog that we use to do forced Narrow Band photometry. Is a join of 3 public catalogs:

1. 1. 1. zCOSMOS-30 catalog: Ilbert, Capak, Salvato et al., ApJ, 2008, astro-ph/0809.2101
    2. Zurich Structure&Morphology catalog – version v1.0 (README_cosmos_morph_zurich_1.0.txt): Scarlata, Carollo, Lilly et al 2007, ApJS, 172, 406, Sargent, Carollo, Lilly et al 2007, ApJS, 172, 434
    3. zCOSMOS public redshift catalog (ADP.2015-10-15T12_55_10.777_t2.txt)

PAUDM_FORCED_APERTURE_COADD: which adds Narrow Band photometry in 40 bands (each from co-adds of 3-10 exposures). These are the last few columns:

flux_NB455 flux in Narrow Band 455nm units such that AB magnitudes are m_AB= 26 -2.5 log10(flux)

flux_error_NB455 error of the flux of Narrow Band 455nm units such that AB magnitudes are m_AB= 26 -2.5 log10(flux)

flux_NB465 flux in Narrow Band 465nm units such that AB magnitudes are m_AB= 26 -2.5 log10(flux)

flux_error_NB465 error of the flux of Narrow Band 465nm units such that AB magnitudes are m_AB= 26 -2.5 log10(flux)

… etc etc for a total on 40 NB in steps on 10nm steps till the last one:

flux_NB845 flux in Narrow Band 845nm units such that AB magnitudes are m_AB= 26 -2.5 log10(flux)

flux_error_NB845 error of the flux of Narrow Band 845nm units such that AB magnitudes are m_AB= 26 -2.5 log10(flux)

BCNZ photo-z as described in the Eriksen etal paper linked above. These are the first few columns, starting with column ‘zb’, which is the photo-z:

zb Photometric redshift (peak of p(z))

odds ODDS quality parameter

pz_width pz_width quality parameter

zb_mean Photometric redshift (mean of p(z))

chi2 Minimum chi2

ebv E(B-V) extinction value

qz Photo-z quality parameter: the lower the better (only qz<4 in this release)

best_run best SED subset

The EDR table contains all the above columns for data with i_auto<22.5 and qz<4 (refer to Eriksen etal 2018 for details). Only good quality photo-z are included in this release as we are still finding the origen of outliers and bad quality data.

You can read here tips to process it as a python panda data frame. There are 6493 separate galaxies with spectra of difference confidence classes. For the Eriksen etal paper we only used 3<= conf < 5. Each galaxy entry has information from 200 (40×5) independent PAUcam exposures.

This plot compares spectroscopic redshifts (column “zspec”) vs photometric redshifts (column “zb”) from the EDR table.

Warning As explained in the above paper, these results are preliminary and we are still working in several points to improve them:

- Working on optimal force photometry subtraction. Current flux values in above files correspond to 62.5% apertures (using HST morphological information when available) without weights. Current PSF convolution uses a single PSF value for all the CCD field. Working on GaaP weighted gaussian PSF homogenisation.
- Zero-points are being revised to account to variation across FoV.
- Improved flat field and scatter light corrections.
- Improved background subtraction.
- Photo-z estimation is based on template min-Chi2 fit on SED subsets and does not currently account for Bayesian priors in the subset of models. We have noticed that this produces global zero point dependence on wavelength which affect ebv values.