meeting 2025 11 03 gw - JacobPilawa/TriaxSchwarzschild_wiki_6 GitHub Wiki

• Following the confusing reply from Hu Zou, we decided to move on with the WISExSUPERCOSMOS data, so I've downloaded that and produced some diagnostics on the catalog.

• Cross-matched WISE and SUPERCOSMOS those sources which have W1 and W2 from wise, with B and R from SCOS. Each raw dataset contains ~1 billion sources each.

  • Note that both WISE and SCOS suffer at low galactic latitudes, so a full sky catalog is not possible from these data alone.

• WISE:

  • AllWISE data release is quoted here (Cutri+13)
  • Raw data contain 747 million sources with S/N >=5 in at least one bad.
  • 95% completeness (averaged over large areas) for W1< 17.1 and W2<15.7
    • Photometry from WISE are the w?mpro mags, based on PSF profile-fit measurements where ? is the channel number (1-4). The w?mpro measurements are available for all objects in WISE, whereas existing attempts to handle extended sources are somewhat heterogeneous (w?gmags which are elliptical apertures derived from 2MASS XSC sources only exist for ~483,000 largest WISE galaxies). There's also circular apertures of different radii, but the angular sizes of the sources have not been determined; the photometry also doesnt account for source ellipticies which is prone to contamination from nearby objects.
  • WISE magnitudes are Vega

• Some notes on the WISE cleaning:

  • All sources with W1<13.8 are eliminated, leaving behind galaxies typically smaller than the WISE resolution threshold and thus well-described by PSF magnitudes (though the fluxes might be slightly underestimated from WISE).
  • Sources must have S/N > 2 in W1 and W2 (practically equivalent to S/N W1 >=5)
  • Removal of "obvious artifacts" (cc flags[1,2]=‘DPHO’) and saturated sources (w?sat > 0.1)
  • "In order to optimize all-sky uniformity, we applied a global magnitude cut of W1 < 17." This removes ~20% of the AllWISE data, mostly around the ecliptic poles where the depth is the greatest.
  • The result is ~488 million AllWISE sources.
  • Galactic extinction is corrected for, following Indebetouw+05 and SF+11. Assumes AW1/E(B-V) = 0.169 and AW2/E(B-V)=0.130 as coefficients to be applied to the original SFD+98 maps.

• SCOS

  • Selected all sources for which gCorMagB and gCorMagR are not null in the database (i.e., are deteceted)
  • Quality warnings < 2048 (no strong warnings or severe defects)
  • SCOS reports a morphology class, and they select those which are classfied as non-stellar and non-noise (meanClass = 1)
  • They also apply a hemisphere correction as described in the text, but probably not that important for now/seems well documented elsewhere.
  • Galactic extinction is corrected for, following Indebetouw+05 and SF+11. Assumes AB/E(B-V) = 3.44 and AR/E(B-V)=2.23 as coefficients to be applied to the original SFD+98 maps.
  • They adopt optical limits or B <21 and R<19.5 for reliability
  • Final sample is of ~85 million galaxies
  • SCOS magntiudes are AB

• Crossmatching Process

  • All cross matching done within 2" radius
  • Astrometric accuracy of both WISE and SCOS is superb (<0.15'' and < 0.3'', respectively), so spurrious results not likely (given that they're detected in both catalogs in all bands)
  • The resulting cross-matched catalog still contains a very large number of contaminating stars and quasars, which are further removed with colors cuts.

• Further sample purification:

  • Quasars are removed with color cuts, motivated by cross-matching their sample with GAMA and SDSS (equation 3). They estimate maybe ~0.6% of the sources after this cut are contaminating quasars.
  • Extensive position dependent star removal outlined in Appendix A based on latitude/longitude and W1 and W2 colors
  • Bright end of the sample (W1<13.8) aslo removed for two reasons: (a) they already have precise redshifts determined frmo 2MPZ; (b) most of the bright sources which are NOT already in 2MASS XSC are stars and concentrated toward the galactic plane.

• The final mask:

  • The purification steps all help greatly, leaving behind ~21.5 million sources, but there are still some problems to be dealt with.
  • Using an iterative clipping approach (based on expected surface density), they produce a final mask. The remaining sky area is roughly 68% of the sky and contains 18.7 million sources. Mean surface density is 670 sources / deg^2, over 20x 2MASS.

• Completness and purity of the final sample:

  • Need to fill this in from the text!

• First, here's the final map of the ~18.7 million sources that pass all the galaxy quality cuts (and are outside the final Bayesian mask):

Source Map

• And the resulting distribution of the different magnitudes in redshift bins for the ~18.7 million galaxies:
  • Note that the R and B are in AB magntiudes; the W1/W2 are Vega magntiudes. Everything has been extinction corrected.

Magntiude Distributions	Color Distributions

• The catalog gives me basically everything I need to know to estimate the stellar masses using the W1 and W2 magntiudes. As a quick reminder, here's the steps of the calculation:

1. Calculate luminoisty in the W1 band. I do this by converting the redshifts given in the catalog to a luminosity distance, assuming H0 = 70 km/s/Mpc and Omega_m = 0.3 (then calculate and apply the distance modulus). This gives me absolute magnitudes in each band. The W1 luminosity (assuming MW1sun = 3.24) is given by:

2. Calculate the mass-to-light ratio from the W1-W2 rest frame colors (in Vega magnitudes, which is how they're quoted in the WISExSUPERCOSMOS catalog). This relation in particular is their full-sample relation (equation 2):

[ \log(M/L) = \begin{cases} -0.246 - 2.100,(W1 - W2), & \text{(jarrett13)} \ -0.030 - 1.960,(W1 - W2), & \text{(cluver14)} \ -0.376 - 1.053,(W1 - W2).clip(-0.2,0.4), & \text{(jarrett23)} \end{cases} ]

• Some notes for the above equations:
  • The W1 and W2 are the w1mpro and w2mpro reported in the WISExSUPERCOSMOS catalog. These are Vega magnitudes and are extinction corrected.
  • .clip(-0.2,0.4) means that any (W1-W2) outside of this range gets fixed to the nearest value.
  • In jarrett23 above, I have not applied the recommended k-corrections which could be non-negligible for our galaxies, since we're no longer focused on the MASSIVE galaxies alone. I'll add a jarrett23_full column which does the entire recommended calculation.

3. As noted just above, the fiducial "jarrett23" results do not perform the full k-correction as recommended, since it's a bit more in depth. I've added this however, and here's a brief description of that process:

     f_{\rm W1}[{\rm rest}] = \frac{f_{\rm W1}[{\rm obs}]}{f_0} \times f_{\rm W1}[{\rm obs}]

     [W1-W2]_{\rm rest} = [W1-W2]_{\rm obs} - \delta[W1-W2]

     M_\star = L_{\rm W1} \times \Upsilon_{\rm W1}

4. The final stellar mass for each case is then LW1 * M/L as described above.

• After going through the process above, the resulting stellar mass distributions look like this (first panel is the full dataset, the remaining panels are redshift slices):

Mass Distributions

• We can also look at, for each source, how the mass estimates compare. Here are a few ways of seeing this:
  • One-to-One: each dot here is a given source, with each panel being a different mass esimate vs. Cluver+14
  • Sample of points, Colored by W1-W2: hoping to convey here that the "spread" in the differences is driven by the color.
  • Binned differences: hoping to quantify, in general, how the different methods agree. This is done by binning the data, computing mean +/- standard deviation as a function of Cluver+14 mass estimate.

Case	Plot
One-to-One
Sample of Points, Colored by W1-W2
Binned Differences

• And here's a plot which helps to show why the masses can be so discrepant -- in the top panel I'm plotting the difference in log(M/L) one would obtain as a function of color. The bottom panel shows the actual color distribution of our sources. At the reddest end, the difference maps to >1 dex in M* when using the Jarrett+23 relation.

ML Differences with W1-W2 color distribution

• My plan going forward was to cross match the WISExSUPERCOSMOS results from above with some of the other catalogs we've been playing with to compare the mass estimates. Crossmatching with WISE2MBH was very straightforward, so I started there. I very quickly cross matched sources here (managed to find ~1.1 million sources in common across the two catalogs), and quickly compared their redshifts and stellar masses:

Redshift

• And a comparison of the various stellar mass estimates vs. the WISE2MBH masses:
  • In general, there seems to be a pretty nice level of agreement in particular for intermediate masses, across the different estimates, in particular with the full jarrett+23 calculation.

Plot

• This matching needs to be improved a bit (I think I have some mismatches currently), but I think I've found roughly ~150,000 galaxies that are in both the Siudek VAC and the WISExSUPERCOSMOS catalogs.

Redshift

• And the mass comparisons:

Plot