Biases of model magnitude for red galaxies

[chto]

@moustakas It seems that the colors based on the best-fit model are biased blue for red galaxies. This bias becomes more pronounced for low redshift galaxies.

The following plot, based on the fastphot-fuji-sv1-dark.fits catalog, demonstrates this finding. I first cut out galaxies with DES>0 to make sure DECAM filter is used for calculations. The y-axis is the fractional error of color based on FLUX_SYNTH_MODEL_band and color based on Flux_band. Triangles with the error bar show the median and error of the median. The blue band shows the histogram of the samples going into the median calculation. The top panel shows low redshift galaxies while the bottom panel shows the full samples.

Since the width of the redsequence is ~0.025, it would be great to have biases smaller than this value.

[moustakas]

Documenting a Slack thread with @chto and Risa from earlier this month. Work on this ticket continues (but no PR yet).

Here’s one of example fit. This object has an observed g-r color of 2.887 and a synthesized (model) color 
of 2.038. You can see that the model can’t get any redder even with A_V=0.85, so I’m optimistic that 
allowing super-solar metallicity will help.

Here’s a fit of the same object with (1) Z=0.030 (compared to Z=Z_sun=0.019 for the nominal fitting); 
(2) allowing models which are older than the age of the universe (note the light-weighted age is now 
~13.3 Gyr vs ~5.0 Gyr previously); and (3) a velocity dispersion of 300 km/s (vs 150 km/s previously). 
Allowing for larger reddening values didn’t help this particular case, but a different attenuation curve 
may.

Take-away: the data are still redder than the model. Next, here’s a fit without W1 (just to see if it’s a 
forced photometry issue in the imaging catalogs). The model really really doesn’t like the g-band!

 OK, here’s the last thing I have to try. Here I switched to using a Charlot & Fall - like attenuation curve 
(with a simple power-law dust slope of -0.7) rather than the nominal Fitzpatrick & Masa (Milky Way) 
extinction curve. The fit is actually a little better even thought Charlot & Fall is actually grayer than the 
MW curve, so it actually doesn’t go in the direction we want. (But the fit is better because the model 
ends up wanting more V-band extinction with this attenuation curve.)

Finally, here's this object in the viewer. There's a nearby companion but it's not clear to me that the photometry would be biased in the direction of being too red based on the nearby source-- https://www.legacysurvey.org/viewer-desi/?ra=221.0252&dec=-1.5677&layer=ls-dr9&zoom=16&targetid=39627752197849848&targets-dr9-sv3-dark

[moustakas]

Following up on the extensive new modeling capabilities incorporated in #95, it is clear that--for this particular test object--it's the data that are (unphysically) red. And in fact you can see from the image cutout that the photometry for this target may be compromised by the nearby companion.

[moustakas]

@chto can you please review the latest model fitting results in Iron and let me know if the model colors have improved. I do believe that the deviations you've documented are unrealistically red observed colors which are due to measurement noise.

[moustakas]

@chto hopefully this issue is resolved, but if not please feel free to re-open.