Presented here is a second look at those S4G galaxies containing
Type III breaks, as published
in Laine
et al. (2016). This list includes all galaxies that displayed one
or more Type III breaks in any band (3.6, u, g, r, i, or z), including
those with hybrid profiles (e.g., Type III + II). Some (a total of
4/175) galaxies in this list hence were not denoted as containing a
Type III break in 3.6 micron imaging, but were included for the sake
of completeness. In all that follows, we used only the 3.6 micron
data and the surface brightness profiles originally measured by Laine
et al. (2016), which were made using fixed position angle and
ellipticity apertures.
Under this new analysis, breaks were automatically detected using
"change point analysis" (a description of which can be
found here),
with minimal human supervision. The algorithm detects changes in the
slope of the surface brightness profile as a function of radius, then
assesses the significance of these changes using bootstrap resampling.
There are three pitfalls of the method (as we reconstructed it from
the web page linked above) that require supervision:
1. If the disparity in slope between two regions
is large, the mean slope will be highly skewed and only the strongest
break will be detected. More subtle breaks are found by excluding the
region of the disk beyond the strongest break and running the analysis
again.
2. If the slope in a region is not constant, the
analysis finds spurious break locations (every point is a change
point). In such cases, only the first break location is used (as the
location where the disk changes to a profile not well-fit by an
exponential). If no part of the disk can be considered exponential,
all 'breaks' found are rejected.
3. Likewise, if a break occurs over a large radial
range, multiple change points will be identified throughout the
break. In such cases the average of all of these change points is
taken as the break radius.
A maximum of three breaks were allowed for each galaxy. The most
significant break may be defined as that which falls at the global
minimum of the cumulative sum profile. Then the profile is split into
two --- one on either side of the first break --- and again is
assessed for disk breaks in each section. This occasionally leads to
some breaks that are significant being ignored. We have marked such
breaks in these pages using parentheses.
After breaks are found, the surface brightness profile is
perturbed by +/- 1 standard deviation (including the sky RMS
measured in the S4G P4 analysis and Poisson noise in the flux within each
annulus) and the analysis is repeated. Each break must survive all
perturbations, or else it is rejected as spurious.
Additionally, initial break identification was done using the
original S4G masks, but all galaxies were checked again using masks
grown through use of a 3 pixel radius circular filter. Breaks in the
disk outskirts that lost significance with the harsher masking were
rejected, as these potentially were spuriously introduced by light
leakage. Some breaks previously remarked in parentheses as possible
but not significant breaks gained significance with the harsher
masking: these were thus included. In total, only 10 breaks were
rejected through this masking procedure.
The following links show a series of plots demonstrating the breaks
for each galaxy. There are two sets of plots per galaxy, which are as
follows (from top-left to bottom-right for each set):
- RADIAL PROFILES
- Radial surface brightness profile (3.6 micron, AB magnitudes;
black dots show the mean, while gray dashed line show the median,
which was not used in the break-finding analysis)
- Radial profile of position angle (in degrees) for the galaxy as
it appears in the sky
- Radial profile of position angle (in degrees) for a deprojected
image of the galaxy
- Radial profile of the m=1 Fourier mode amplitude, normalized to
the surface brightness in each radial bin
- The local slope of the surface brightness profile as a function
of radius, measured at each point using the adjacent 2 points on
either side (following Pohlen & Trujillo 2006)
- Radial axial ratio profile (b/a) of the galaxy as it appears in
the sky
- Radial axial ratio profile (b/a) of the deprojected image of the
galaxy
- Radial profile of the m=2 Fourier mode amplitude, normalized to
the surface brightness in each radial bin
- IMAGES
- The on-sky image of the galaxy, 3.6 micron. Star/noise masks
are shown in green
- Unsharp-mask on-sky image of the galaxy, 3.6 micron, masks same as
above
- On-sky image of the galaxy, 3.6 micron, masked of all pixels with fluxes
above the third quartile of the flux distribution in each radial
bin used for surface photometry (masks outline brightest
structures and so occasionally uncover, e.g., subtle spiral arms)
- The on-sky image of the galaxy, g-band (if available). Masks
are the same as those used in the 3.6 micron imaging.
- Unsharp-mask on-sky image of the galaxy, g-band (if available).
- Deprojected image of the galaxy, 3.6 micron (for aid in
identifying oval disks or Type IIIs breaks).
In each radial plot, the following are marked:
- Black dots show the mean surface brightness profile
- Gray dashed lines show the median surface brightness profile
- Red crosses in the slope profile plots denote the slopes
before median smoothing is applied, while black dashed lines
show the slope profile after smoothing (smoothing is done using
a kernel 0.1x the width of the radius array). Gray dashed lines
show the smoothed profile for the median surface brightnesses
(see above)
- Red shaded regions denote regions excluded from the break
analysis. R_outer was taken from Laine et al. (2016), R_inner
excludes all regions with clear influence from central
bulges/bars, including inner rings, inner lenses, and inner
spiral arms (if they emerge directly from the ends of the
bars).*
- Green dashed lines denote radii of rings, lenses, etc. from
Buta et al. (2015).
- Blue dotted lines denote the break radii from Laine et
al. (2016)
- Gold dashed lines denote the break radii from this
analysis
* NOTE: a handful of galaxies show evidence of being "oval
galaxies" (see Section 3.2
of Kormendy
and Kennicutt 2004). The central regions of these galaxies,
though appearing as normal disks, are offset in position angle and
ellipticity from the outermost isophotes, hence appear "oval" in
shape when the galaxy is deprojected. In unambiguous cases the
entirety of the "oval" was avoided in the selection of R_inner
(mostly to simplify the break selection, as such galaxies appear
almost as two disks, each with its own set of breaks).
In each image plot, we show only the inner and outer radius
boundaries (red dashed ellipses) and our break radii (gold dashed
ellipses), to avoid clutter.
Finally, for each galaxy are given Ron Buta's classification (a
detailed description of these classifications can be found in
Buta et
al. 2015, as well as
in table
form) the break type classifications for all breaks found in
this analysis, and brief notes justifying the choice of break
classification. I adopt here the following break classification
scheme:
- Type II breaks have the following subclasses:
- Type IId: break is clearly associated with symmetric
features found in the disk, e.g., spiral
structure, rings, or lenses
- Type IIa: break follows the ridgelines of asymmetric
features such as tidal streams and single spiral arms
- Type II: it is unclear with what feature the break
is associated, hence no added letter
- Type III breaks have the following subclasses:
- Type IIId: akin to IId, the break is associated with
symmetric features in the disk
- Type IIIa: the break appears related to asymmetric
isophotes, whether this be single spiral arm modes,
lopsided isophotes, or tidal features
- Type IIIs: the break occurs where the isophotes
begin becoming continually rounder with radius but
remain symmetric
- Type III: it is unclear with what feature the break is
associated, hence no letter added
- Type I breaks are now also given subclasses:
- Type I: a standard, no-break exponential
- Type 0: denotes that the profile is not well-fit by an
exponential anywhere