Context. The elliptical galaxy NGC 3923 is known to be surrounded by a number of stellar shells, probable remnants of an accreted galaxy. Despite its uniqueness, the deepest images of its outskirts come from the 1980s. On the basis of the modified Newtonian dynamics (MOND), it has recently been predicted that a new shell lies in this region. Aims. We obtain the deepest image ever of the galaxy, map the tidal features in it, and search for the predicted shell. Methods. The image of the galaxy was taken by the MegaCam camera at the Canada-France-Hawaii Telescope in the g band. It reached the surface-brightness limit of 29 mag/arcsec2. In addition, we reanalyzed an archival HST image of the galaxy. Results. We detected up to 42 shells in NGC 3923. This is by far the highest number among all shell galaxies. We present the description of the shells and other tidal features in the galaxy. A probable progenitor of some of these features was discovered. The shell system likely originates from two or more progenitors. The predicted shell was not detected, but the new image revealed that the prediction was based on incorrect assumptions and poor data.
Our proximity and external vantage point make M31 an ideal testbed for understanding the structure of spiral galaxies. The Andromeda Optical and Infrared Disk Survey (ANDROIDS) has mapped M31s bulge and disk out to R=40 kpc in $ugriJK_s$ bands with CFHT using a careful sky calibration. We use Bayesian modelling of the optical-infrared spectral energy distribution (SED) to estimate profiles of M31s stellar populations and mass along the major axis. This analysis provides evidence for inside-out disk formation and a declining metallicity gradient. M31s $i$-band mass-to-light ratio ($M/L_i^*$) decreases from 0.5 dex in the bulge to $sim 0.2$ dex at 40 kpc. The best-constrained stellar population models use the full $ugriJK_s$ SED but are also consistent with optical-only fits. Therefore, while NIR data can be successfully modelled with modern stellar population synthesis, NIR data do not provide additional constraints in this application. Fits to the $gi$-SED alone yield $M/L_i^*$ that are systematically lower than the full SED fit by 0.1 dex. This is still smaller than the 0.3 dex scatter amongst different relations for $M/L_i$ via $g-i$ colour found in the literature. We advocate a stellar mass of $M_*(30mathrm{kpc})=10.3^{+2.3}_{-1.7}times 10^{10}mathrm{M}_odot$ for the M31 bulge and disk.
We present wide-field near-infrared J and Ks images of the Andromeda Galaxy taken with WIRCam on the Canada-France-Hawaii Telescope (CFHT) as part of the Andromeda Optical and Infrared Disk Survey (ANDROIDS). This data set allows simultaneous observations of resolved stars and NIR surface brightness across M31s entire bulge and disk (within R=22 kpc). The primary concern of this work is the development of NIR observation and reduction methods to recover a uniform surface brightness map across the 3x1 degree disk of M31. This necessitates sky-target nodding across 27 WIRCam fields. Two sky-target nodding strategies were tested, and we find that strictly minimizing sky sampling latency does not maximize sky subtraction accuracy, which is at best 2% of the sky level. The mean surface brightness difference between blocks in our mosaic can be reduced from 1% to 0.1% of the sky brightness by introducing scalar sky offsets to each image. The true surface brightness of M31 can be known to within a statistical zeropoint of 0.15% of the sky level (0.2 mag arcsec sq. uncertainty at R=15 kpc). Surface brightness stability across individual WIRCam frames is limited by both WIRCam flat field evolution and residual sky background shapes. To overcome flat field variability of order 1% over 30 minutes, we find that WIRCam data should be calibrated with real-time sky flats. Due either to atmospheric or instrumental variations, the individual WIRCam frames have typical residual shapes with amplitudes of 0.2% of the sky after real-time flat fielding and median sky subtraction. We present our WIRCam reduction pipeline and performance analysis here as a template for future near-infrared observers needing wide-area surface brightness maps with sky-target nodding, and give specific recommendations for improving photometry of all CFHT/WIRCam programs. (Abridged)
