Seeing-sorted Visible Multi-Object Spectrograph U-band Imaging of the GOODS-south Field^*

Nonino et al. (2009) presented the results of deep U-band and R-band imaging of the GOODS-south field. In the current paper, we present the seeing-sorted version of their U-bandimages, which also have a closer zero-point match to other GOODS multi-wavelength photometry. By utilizing a number of image processing techniques, Nonino et al. (2009) were able to create an image mosaic out of the 552 single chip images taken from the VIMOS instrument on the Very Large Telescope (VLT) in Chile. The final mosaic from this work had a full width half maximum of 08 and reached a depth of m_AB ≃ 29.8 mag (Nonino et al. 2009).

Ashcraft et al. (2018) used observations from the Large Binocular Telescope of the GOODS-North field and used a sub-stacking method to create two image mosaics: (1) the best resolution mosaic, generated with images using seeing FWHM ≤08, and (2) the best depth mosaic, cut off at 18 seeing (Ashcraft et al. 2018). This stacking method is useful to mitigate atmospheric effects associated with ground-based imaging. The optimal resolution mosaic is best for studying bright galaxies, and shows structure in galaxies more clearly than the best depth mosaic. The optimal depth stack, however, is more sensitive to lower surface brightness objects and structures. Ashcraft et al. (2018) showed using galaxy number counts that the optimal depth mosaic detects more of the faintest galaxies.

The work presented here focuses on the GOODS-south field, builds on the data of Nonino et al. (2009) and uses the seeing-sorted stacking method of Ashcraft et al. (2018). An additional step was incorporated to reduce uncertainty and variability in the zero-points of individual images following a process similar to one described in T. McCabe et al. (2021, in preparation).

The 552 individual U_V -filter images used in this work were taken between 2004 August and 2006 October by the VIsible Multi-Object Spectrograph (VIMOS) on ESO's VLT (Nonino et al. 2009). VIMOS is a four-chip charge coupled device (CCD) camera with a pixel scale of 0205/pixel and a field of view of 4 × 7' × 8'. In imaging mode, each chip uses a 2048 × 2440 pixels EEV 44–82 backside illuminated CCD. The chips are separated by 2' gaps. Special care was taken when dithering to get uniform coverage of the field (Nonino et al. 2009).

Ashcraft et al. (2018) found a ∼0.2 mag difference in the U-bandphotometric zero-points between their optimal resolution and optimal depth LBT/LBC mosaics. This difference was attributed to variations in transparency among different exposures and across different nights. Here, we therefore correct the flux scale of each image to match the U-bandphotometry in the 3D-HST photometric catalog of Skelton et al. (2014). Figures 1(e) and (f) show an example of the uncorrected and corrected flux ratios between the VIMOS images and the 3D-HST catalog for the first four nights of observations. Prior to these corrections, the average ratio across all nights was 97.6% ± 0.6%. After applying these U-bandzero-point corrections, this ratio increased to 99.8% ± 0.1%, better matching the 3D-HST catalog.

Zoom In Zoom Out Reset image size

All 552 single chip exposures were sorted based on their seeing (Figure 1(c)). swarp (Bertin et al. 2002; Bertin 2010) was used to combine the individual images. Our best resolution mosaic, cut off at FWHM ≲ 08, used 288 images, just over 50% of the VIMOS U_V data (Figure 1(a)). Our best depth mosaic used images with seeing FWHM ≲ 15, corresponding to 100% of the VIMOS U_V data (Figure 1(b)). SExtractor (Bertin & Arnouts 1996) was used to create source catalogs of the mosaics. The mosaics shared the same SExtractor configuration parameters, which are similar to those in Ashcraft et al. (2018). The galaxies were separated from stars as described in, e.g., Windhorst et al. (2011). Figure 1(d) shows the resulting galaxy number counts from our U-bandcatalogs. The best resolution mosaic U-bandcounts start to turn over around m_AB ≃ 26.8 mag, and fall off more quickly than the best depth U-bandmosaic, which turns over at m_AB ≃ 27 mag in U-band. For examples of the differences in resolution and depth between our best resolution and best depth mosaics we refer the reader to Figures 2–5 of Ashcraft et al. (2018), which are also representative for our data.

We present the results of deep U-bandimaging of the GOODS-south field from VIMOS on the VLT using data and image reductions from Nonino et al. (2009). Two image mosaics were made following the seeing-sorted stacking method of Ashcraft et al. (2018). All 552 single chip images were sorted based on FWHM. The optimal resolution mosaic was assembled using only images with FWHM ≤ 08. The optimal depth mosaic was made from images with FWHM ≤ 15. Before creating the mosaics, 139 four-chip exposures were corrected to better match the 3D-HST U-bandflux (Skelton et al. 2014). With the completion of these mosaics, three of the four UVCANDELS fields will have ground-based images complementary to HST data. New LBT U-banddata for the EGS and COSMOS fields are currently being reduced, thereby completing the seeing-sorted stacking analysis for the UVCANDELS program. In the coming years, these mosaics will complement JWST 1–5 μm observations and help us better understand galaxy assembly over the past 10 billion years. The image mosaics and source catalogs from this work will be released as part of the UVCANDELS program later this year, found at http://uvcandels.ipac.caltech.edu.

This work used data from the 3D-HST Treasury Program (HST-GO-12177 and 12328). We also acknowledge support from UVCANDELS grant HST-GO-15647 provided by NASA through the Space Telescope Science Institute, which is operated by the Association of Universities for Research in Astronomy, Inc., under NASA contract NAS 5-26555. We acknowledge support from NASA JWST Interdisciplinary Scientist grants NAG5-12460, NNX14AN10G and 80NSSC18K0200 from GSFC. M.N. acknowledges INAF 1.05.01.86.20.