______________________________________________________________________ ______________________________________________________________________ K--detected UU38BVRIzJHK catalogue of the Extended Chandra Deep Field South version 3.0 Friday, December 10, 2008 (Constitution Day, Russia) ______________________________________________________________________ ______________________________________________________________________ Generated by Edward N Taylor -- ent@strw.leidenuniv.nl ______________________________________________________________________ ______________________________________________________________________ The catalogue is fully described in Taylor et al. (2009), "A Public, K-Selected, Optical--to--Near-Infrared Catalog of the Extended Chandra Deep Field South (ECDFS) from the MUltiwavelength Survey by Yale-Chile (MUSYC)". Please cite this work when using the catalogue. This README contains a brief description of the catalogue; for further information and explanation please see the data paper. Both photometric redshift determinations from EAZY (Brammer et al. 2008) and interpolated restframe colors are available in companion catalogues; seperate restframe color catalogues based on the photometric and spectroscopic redshifts are given. ______________________________________________________________________ ______________________________________________________________________ RECOMMENDATIONS FOR USE: * Use Kw > 0.75 -- this cut is the lowest possible to cut out a number of spurious detections in the Western and North-Western tiles, where the use of a regular stepped dither pattern led to a number of 'holes' in the exposure map, without losing a significant imaging area. * Use zw == 1 -- the z band data has some funny saturation issues around very bright sources; affected regions have been masked by hand. (Since no exposure map is available for the z band data, this mask is used instead.) This cut also eliminates some spurious sources around bright stars, and serves to cull objects whose photometry may be contaminated. * Use the H photometry with caution -- no exposure maps were available for the H band data; they were mosaicked using 'mock' exposure maps based on the per-pixel RMS in rows and columns. This leaves a lot of noise along the edges of individual tiles, which affects both background estimation, and the photometry itself. These effects are accounted for in a statistical sense in the errors, but in reality the effects are probably strongly non-Gaussian. * Use only those spectroscopic redshift determinations with spec_flag == 1. These objects either have high confidence quality flags from their original sources, or have been confirmed from multiple determinations by independent teams. ______________________________________________________________________ ______________________________________________________________________ COLUMN DESCRIPTIONS: This catalogue follows the agreed-upon standard for MUSYC IR--selected catalogues. Columns are as follows: id Running detection number ra, dec Celestial coordinates field MUSYC field id number (E-CDFS=8) x, y Pixel position on the E-CDFS.ISPI astrometric grid (See note 1 below.) ap_col Effective diameter of the 'color aperture' [arcsec] X_colf, X_colfe 'Color aperture' flux in band X, with associated error (Measured on 1.5" image. See notes 2,3,5 below.) ap_tot Effective diameter of the 'total aperture' [arcsec] K_totf, K_totfe 'Total' K fluxes with associated error (Measured on 1.0" image. See notes 2,4,5 below.) K_autof, Sextractor's AUTO flux, uncorrected K_4autofe (Measured on 1.0" image. See notes 2,6 below.) K_4arcsecf, 4" aperture flux with associated error K_4arcsecfe (Measured on 1.0" image. See notes 2,6 below.) Kr50 K band half-light radius [arcsec] Keps K band ellipticity Kposang K band position angle Xw Normalised exposure time (See note 6 below.) id_SEx SEx's running NUMBER from detection f_deblend_1 SEx's deblending flag (flag 1) f_deblend_2 SEx's nearby neighbour flag (flag 2) z_spec Literature spectroscopic redshift determination source Source code for z_spec qf_spec Quality flag for z_spec (from source, if available) spec_type Spectral classification (from source, if available) (See note 7 below.) Qz_spec Figure of merit for z_spec, from photometry (See note 8 below.) nsources Number of corroborating spec-z determinations spec_flag Flag indicating wheter z_spec is considered 'robust' ______________________________________________________________________ NOTES ON CATALOGUE PARAMETERS: 1. The NIR imaging is slightly offset from the optical imaging. To account for this, the data have been reregistered to a slightly larger frame for the K--selected catalogue photometry than for the BVR selected photometry. The transformation from K--selected position to BVR--selected position is: (x_BVR, y_BVR) = (x_K + 501, y_K+301) . See under 'data verification' for notes about the accuracy of the MUSYC astrometry. 2. All fluxes are normalised to an AB magnitude zeropoint of 25.0, i.e. MAG_X_AB = -2.5 * log_10( FLUX_X ) + 25.0 . To convert these fluxes into uJy, multiply by 10^(-0.44) ~ 0.363. Note that this was wrong in earlier READMEs. The converstion to Vega is given by: MAG_X_VEGA = MAG_X_AB - MAG_VEGA,X_AB , where MAG_VEGA,X_AB is the apparent AB magnitude of Vega in the X band. These are as follows: band U U38 B V R I z J H K MAG_VEGA 1.01 .816 -.119 -.009 .194 .507 .543 .934 1.40 1.83 3. Color fluxes are measured from images which have been PSF matched to 1.5" FWHM. The effective diameter (i.e. sqrt(area / pi)) of the color aperture is given by 'ap_col'. Most often, color fluxes are measured using SExtractor's FLUX_ISO. Where the effective diameter of the ISO aperture was less than 2.5", however, a fixed aperture of diameter 2.5" was used instead. 4. Total fluxes are measured from a K image with a uniform 1.0" FWHM PSF. The effective diameter (i.e. sqrt(area * pi)) of the total aperture is given by 'ap_tot'. Most often, total fluxes are based on SExtractor's FLUX_AUTO. Where the effective diameter of the ISO aperture was less than 2.5", however, a fixed aperture of diameter 2.5" was used instead. For each object, a minimal correction has been applied to account for light laying beyond the aperture, treating each object as if it were a point source. 5. The measurement uncertainties for all of fluxes were determined as follows: A large number of empty apertures of varying sizes (3000 with D < 13" on the 1.0" K image for total fluxes, and 10000 with D < 4" on the 1.5" images for color and 4" fluxes) were placed on the science images, isolated from each other as well as from detected objects (based on the 1.0" K detection segmentation map only for total fluxes, and including the BVR detection segmentation map for color and 4" fluxes). These were used to determine the flux variance as a function of (effective) aperture diameter. For J and K, we also use the RMS maps output by xdimsum in the mosaicking phase to track spatial variations in the noise. These maps are not accurate in a quantitative sense, so we have normalised them using the 'empty aperture' values. The AB magnitudes corresponding to 5 sigma flux in the minimum 2.5" apertures are: band U U38 B V R I z J H K 5sig mag 26.8 26.1 27.0 26.6 26.4 24.7 24.0 23.0 21.6 22.3 6. The given weights are normalised to the nominal exposure time in each band. These times are: band U U38 B V R I z J H K time (hrs) 21.91 ??? 19.29 29.06 24.35 9.60 20.?? 1.33 1.00 1.00 7. Spectroscopic redshifts have been collected from a number of public, published works. In collating these redshifts, where multiple (consistent) redshift determinations are available for a given object, we have chosen to adopt the first published determination, except where a later determination includes spectral classification data not given previously. We also choose Xray selected catalogues (viz. Szokoly et al. and Treister et al.) in preference to others, considering Xray selection as an additional piece of classification information. Where there is no consensus (e.g. two different redshifts from two different sources), we have chosen according to the quality flags. The codes for spectroscopic redshift sources are as follows: K20 - Cimatti et al. (2002), Mignoli et al. (2005); Xray - Szokoly et al. (2004); VVDS - Le Fevre et al. (2004); GDS-F - Vanzella et al. (2005, 2006, 2007); GDS-V - Popesso et al. (2008); IMAGES - Ravikumar et al. (2007); MUS-I, MUS-V - Treister et al. (2008); Kopsv - Koposov et al. (in prep.); KX - Croom et al. (2001); SNe - Strolger et al. (2004); vdWel - Van der Wel et al. (2004, 2005); Daddi - Daddi et al. (2005); LCIRS - Doherty et al. (2005); Kriek - Kriek et al. (2006). Where the original authors have provided a quality flag, this is given in the catalogue; similarly, we given spectral classifications where available. The number of consistent redshift determinations is given in the column 'nsources'. 8. For each spectroscopic redshift determination, we have evaluated a 'figure of merit', characterising the consistency of the photometry with that redshift, using the recipe described by Brammer et al. (2008). This is quantity is given as Qz_spec. ______________________________________________________________________ ______________________________________________________________________ ESSENTIAL CHANGES WITH RESPECT TO v2.0: * Detection and total flux measurement is done on an image with a uniform, 1.0" FWHM PSF. Color fluxes are still done on the 1.5" FWHM PSF-matched images. * Zeropoint issues have finally been resolved. * Total fluxes are measured in AUTO apertures which are, on average, half as large, and thus have smaller errors. * Color fluxes are measured in ISO apertures (rather than fixed apertures) to make sure that color gradients are covered for the largest and brightest objects. * The minimum aperture size for both total and color fluxes is 2.5". * A problem with the NIR photometric errors has been resolved. * Hendrik Hildebrandt has re-reduced the U38 data; the U38 photometry now appears to be reliable. * Fixed 4" aperture fluxes are given (measured on the 1.5" image). * SExtractor detected objects with Kw < 0.2 have been culled from the catalogue. The column id_SEx gives SExtractor's NUMBER from the detection phase. * Spectroscopic redshift determinations and associated information are now included in the photometric catalogue. ______________________________________________________________________ ______________________________________________________________________ DESCRIPTION OF THE DATA: The MUSYC E-CDFS dataset is a conglomeration of existing public and original observations. To summarise the various sources of the imaging, and their original photometric calibrations (NOTE: We have refined these calibrations to optimize the cross--calibration across all ten bands; see next section): -- U U38 BVRI (WFI@MPG/ESO 2.2m) -- These data are a combination of the COMBO-17 (Wolf et al., 2004, A&A 421, 913) and EIS DPS (Arnouts et al., 2001, A&A 379, 740) WFI imaging, which were re-reduced as part of GaBoDS (Erben et al., 2005, AN 326, 432; Hildebrandt et al., 2006, A&A 452, 1121). -- z (MOSAIC II@CTIO Blanco 4m) -- These are original data, taken in January 2005, and reduced in--house. Photometric calibration was done with reference to standard stars observed by Smith et al. (2003, AJ 126, 2037), using the CTIO 0.9m. -- H (SOFI@NTT 3.58m) -- This is a mosaic of 32 seperate SOFI pointings presented by Moy et al. (2003, A&A 403, 493), covering approximately 80% of the 30" square ECDFS region. The data were supplied by Pauline Barmby, fully reduced and calibrated, but unmosaicked. Mosaicking has been techinically challenging, since the data did not come with exposure/weight maps; this remains a potential area for improvement, but there are no plans to address this. -- JK (ISPI@CTIO Blanco 4m) -- These are original data, taken over several observing campaigns, and reduced by Edward Taylor. The reduction and calibration will be discussed in a paper in preparation, but some of the many data verification procedures are summarised below. ______________________________________________________________________ DATA VERIFICATION: Registration -- The relative (band--to--band) registration has been validated to subpixel levels. The absolute astrometry has been tested by Maaike Damen and Ivo Labbe, based on a YSO astrograph catalogue supplied by Terry Girards. They found a simple offset of 0.3" (~1.1 pix), and a slight (< 0.1") shear in the central regions. PSF matching -- The quality of the PSF matching has been validated using on the growth curves of ~ 100 bright, unsaturated, isolated stars (based on COMBO-17 SED classification). For the smallest apertures (2.5"), band-to-band variations in the normalised growth curves are at worst 3 %, and typically less than 1 % Zeropoints -- SO much time has been spent on this. With the new reduction of the WFI data, the GaBoDS/THELI supplied zeropoints now seem fine: the source of the problems seems to have been an issue in the original WFI data reduction. (This was known to the GaBoDS/ THELI team for some time, and took us a while to catch up on.) The B--K photometric zeropoints have been verified by comparing the colours of stars in the GOODS ACS/ISAAC catalogue to the colour transforms predicted using BPGS star spectra. In all cases, the zeropoints agree to < ~0.06 mag, which is less than twice the formal (RMS/sqrt(N+1)) uncertainties. SED photometry -- We have done detailed comparisons with COMBO-17 broadband colors, which showed that earlier catalogues suffered systematic effects due to color gradients for the brightest and biggest sources. This has been mitigated by using the ISO apertures where it is larger in area than a 2.5" fixed aperture. Total magnitudes -- We have done detailed simulations to determine the extent to which SExtractor's AUTO apertures miss flux, as a function of size and brightness. For a K = 22 de Vaucouleurs profile, the point--source like correction reduces the systematic flux underestimate to < 5 % for r_e < 0.5" (4 kpc @ z=1). Completeness and reliability -- Similar simulations have also been done to investigate the catalogue completeness as a function of size and magnitude. Point source completeness is 50 %, 90 %, and 99 % at K = 22.4, 22.2, and 21.0, respectively, and drops for larger sizes. Completeness is quite uniform across the field, although it does drop by a small amount in the Eastern tile. These limits are completely consistent with an empirical determination of completeness and reliability in comparison to the GOODS ACS/ISAAC K-selected catalogue (which covers mostly the central tile). For K < 22, the catalogue is 97 % complete, and 99 % reliable; for the 21.8 < K < 22.0 bin, these numbers are 87.5 % and 97 %. ______________________________________________________________________ ______________________________________________________________________