Chapter 5a - Deep Imaging Photometry


As part of the All-wavelength Extended Groth strip International Survey (AEGIS), the Galaxy Evolution Explorer (GALEX) has obtained deep imaging and spectroscopic observations of the Extended Groth Strip. The GALEX field is centered on 14:19:58.0, +52:46:54 (J2000), and the circle with useful data has a radius of 0.58 deg. We are continuing to observe the field for time-variability studies (e.g., Gezari et al. (2006) ApJL, 653, L25), but this analysis was performed on the GR3 dataset and contains observations through Spring, 2006, with a total exposure time of 120160 sec for the FUV imaging channel, 237066 sec for the NUV imaging channel, and 281714 sec for spectroscopy.

Figure 1 The GALEX NUV image of the Extended Groth Strip. The yellow line shows the perimeter of the region that was deblended using the CFHTLS catalog. The red line shows the perimeter of the region for which there are DEEP2 redshifts.

Standard Pipeline Products

Currently, the best description of the standard GALEX pipeline and calibration is Morrissey et al. (2007) ApJS, 173, 682. Click here to download the GR3 imaging data and here to download the spectroscopic data from MAST. These data products are already flat-fielded and flux-calibrated as described in the Morrissey et al. paper.


The FWHM of the FUV and NUV images are 4.5 and 5.4 arcsec, respectively. The images suffer from significant confusion. In order to obtain reliable flux estimates for sources in the field, we have deblended the images using the CFHTLS catalog to specify the centroids of the sources. As shown in Figure 1, we have limited the deblending to within a radius of 0.58 deg; beyond this radius, the geometric distortions in the GALEX images become severe. We assume that every source in the GALEX image is detected in the CFHTLS image, which is a plausible assumption since the CFHTLS image is deeper than the GALEX image and sources that are bluer than flat spectrum (ƒν) are very rare. We model each source as a point source. We further assume that the photon counts follow a Gaussian distribution. Since the background counts are ~25 in the FUV and ~425 in NUV, this is a safe assumption. The image is deblended in 50 pixel x 50 pixel patches, holding the source locations fixed and simultaneously fitting the source amplitudes and a uniform background level. Error estimates for the sources fluxes are obtained from the covariance matrix in the standard manner.

We provide here two catalogs.

GALEX_20070818_simple.fits.gz - A no-frills "simple" AEGIS format. For each object, this catalog contains a sequence ID, celestial coordinates (and position error estimate), and the deblended fluxes and errors in Janskys for the FUV and NUV bands.

matchGalexCfhtDeep2.fits.gz - A more comprehensive collection of data, as described in Table 1 below.

Table 1 - Columns in matchGalexCfhtDeep2
Column Meaning Units
RA CFHTLS right ascension (J2000) degrees
Dec CFHTLS declination (J2000) degrees
umega CFHTLS u flux magnitudes
umega_err CFHTLS u flux err magnitudes
gmega CFHTLS g flux magnitudes
gmega_err CFHTLS g flux err magnitudes
rmega CFHTLS r flux magnitudes
rmega_err CFHTLS r flux err magnitudes
imega CFHTLS i flux magnitudes
imega_err CFHTLS i flux err magnitudes
zmega CFHTLS z flux magnitudes
zmega_err CFHTLS z flux err magnitudes
half_light_radius CFHTLS half light radius arcseconds
alpha_j2000 GALEX right ascension (J2000) degrees
delta_j2000 GALEX declination (J2000) degrees
E_BV Selective extinction Schlegel et al
fov_radius GALEX field-of-view radius degrees
nuv_mag GALEX pipeline NUV flux magnitudes
nuv_magerr GALEX pipeline NUV flux error magnitudes
fuv_mag GALEX pipeline FUV flux magnitudes
fuv_magerr GALEX pipeline FUV flux error magnitudes
nuv_artifact GALEX NUV artifact flag Definition
fuv_artifact GALEX FUV artifact flag Definition
nuv_mag_aper_4 GALEX NUV 12" diameter flux magnitudes
nuv_mag_aper_7 GALEX NUV 34.5" diameter flux magnitudes
nuv_magerr_aper_4 GALEX NUV 12" diameter flux error magnitudes
nuv_magerr_aper_7 GALEX NUV 34.5" diameter flux error magnitudes
nuv_x_image GALEX NUV X position pixels
nuv_y_image GALEX NUV Y position pixels
fuv_mag_aper_4 GALEX FUV 12" diameter flux magnitudes
fuv_mag_aper_7 GALEX FUV 34.5" diameter flux magnitudes
fuv_magerr_aper_4 GALEX FUV 12" diameter flux error magnitudes
fuv_magerr_aper_7 GALEX FUV 34.5" diameter flux error magnitudes
fuv_x_image GALEX FUV X position pixels
fuv_y_image GALEX FUV Y position pixels
ggo_id GALEX global object ID Definition
cfht_nuv_x Corrected NUV X position pixels
cfht_nuv_y Corrected NUV Y position pixels
cfht_fuv_x Corrected FUV X position pixels
cfht_fuv_y Corrected FUV Y position pixels
separation GALEX/CFHT separation arcseconds
nn10 CFHT distances of 10 nearest neighbors arcseconds
nuv_db Deblended NUV flux magnitudes
nuv_db_err Deblended NUV flux error magnitudes
fuv_db Deblended FUV flux magnitudes
fuv_db_err Deblended FUV flux error magnitudes
z DEEP2 redshift Details
z_err DEEP2 redshift error Details
zquality DEEP2 redshift quality Details

Table 1 Description of the columns in matchGalexCfhtDeep2 catalog. Galactic extinction corrections have not been applied to any of the fluxes.

The following series of plots and captions describe the results of the deblending. (You can click on each of the images to see a larger/legible version.)

Figure 2 Distribution of residual count rate (image - model) normalized by the expected Poisson noise. The solid red curve is Gaussian fit to the distribution; the dashed red curve is the expected Poisson distribution. The residuals are biased slightly positive (μ = 0.0479) and are slightly wider (σ = 1.0554) than the ideal (μ = 0., σ = 1.).

Figure 3 Comparison of NUV deblended magnitudes with standard pipeline magnitudes. The standard pipeline magnitudes are measured in circular apertures with a diameter of 34.5 arcsec ("mag_aper_7"). The sources plotted in red are those without a neighbor within 6 arcsec in the CFHTLS catalog. In other words, the sources in red are those for which the standard pipeline magnitudes ought to be the most reliable.

Figure 4 Normalized distribution of the differences between the deblended and standard pipeline magnitudes (34.5 arcsec diameter circular aperture) for standard pipeline magnitudes brighter than 25 mag. The black histogram is for all sources, while the red histogram is for isolated sources without a neighbor within 6 arcsec. The black histogram for all sources has a large tail of sources in which the standard pipeline flux is too bright. The red histogram for isolated sources does not have this tail, which implies that the standard pipeline overestimates the flux of blended sources.

Figure 5 Signal-to-noise ratio versus deblended NUV magnitude. The sources plotted in red are isolated with no neighbors within 6 arcsec. The blue line is the Poisson estimate of the SNR, assuming a sky background rate of 0.0011 photons/sec/sq arcsec (=0.0025 photons/sec/pixel), a source area of 110 sq arcsec (=49 sq pixels), and a median effective exposure time of 183522 sec. The horizontal green dashed line marks an SNR of 5; the vertical green dashed line is drawn at 26 mag.

Figure 6 Errors in recovered fluxes of artificial stars added to the NUV image as a function of true flux. The artificial stars (black points) have magnitudes of 22.5 mag, 23.0 mag, 23.5 mag, …, 26.5 mag. For clarity, the points are shown in the figure with their x-axis positions "jittered." The large red circles are the median difference between the recovered and input fluxes.

Figure 7 Distribution of count rate errors, scaled by the estimated errors, for the artificial stars. The red line is a Gaussian fit with μ = -0.07, σ = 1.24. The median difference between the recovered and input magnitudes for all artificial stars is -0.009 mag. The median difference is -0.004 mag for input magnitudes ≤ 25.5 mag; the median difference is -0.108 mag for input magnitudes > 25.5 mag.

Figure 8 NUV flux (counts per seconds) versus CFHT u magnitude for isolated CFHT sources (i.e., sources without a neighbor within 6 arcseconds). The dashed red lines are drawn at count rates corresponding to NUV = 25.0, 25.3, and 26.0 mag. The locus of points heading to the upper left in the plot are well-detected NUV sources.

Figure 9 Histogram of NUV count rates for sources with u ≥ 27.5, g ≥ 27.5, r ≥ 27.5, and no neighbors within 6 arcseconds. These positions are taken as a proxy for blank sky. The red line is a Gaussian fit to the histogram with μ = 0.00027 counts per second, σ = 0.0014 counts per second. 99% of the "sources" have NUV mag < 25.3 mag. We take 25.3 mag to be the 99% confidence detection limit for the deblended NUV EGS catalog.