Small systematic redshift error for galaxies

[londumas]

When looking how redrock performs on ELG+LRG truth table from eBOSS we observe a small systematic redshift difference of around 10 km/s.
Obviously it is not that important but it would be nice to know why.
I am investigating how the redshift in the truth table has been calibrated,

[londumas]

There seems to be no shift between galaxy templates in idlspec2d and in redrock. It could be linked to the broadbands in idlspec2d.

[londumas]

I did a simple test: I ran the current master branch of redrock on some eBOSS galaxy plates. The current redrock templates are based on desisim galaxies.
I ran the same redrock replacing the galaxy templates by the SDSS galaxy templates.
I get back this 10 km/s.
Conclusion: This small shift is linked to differences in the galaxy templates, not to broadbands.
Different possibilities:

• different definition of wavelength?
• systematic shift of the (or a) lines in the data

[londumas]

Here is the distribution of velocity difference between SDSS (spEigenGal-55740.fits) and redrock (desisim templates).

[londumas]

@moustakas, a difference of 10 km/s corresponds to a difference of 0.1 A at z=0. This must come from the definition of the OII line in desisim. Could you tell me what are the values chosen for the two line of this double? I don't find it anywhere in the list given in #111. Let's see if this is that simple. Do we even care about a systematic error of 10 km/s ?

[moustakas]

The adopted rest wavelengths of the forbidden lines (in vacuum) are here:
https://github.com/desihub/desisim/blob/master/py/desisim/data/forbidden_lines.ecsv

[moustakas]

@londumas Can you make a plot of the data and the best-fitting templates (for idlspec2d and redrock) for a handful of objects with ~10 km/s velocity shifts? We should be able to easily tell if the problem is with the (rest-frame) position of the emission lines.

[londumas]

@moustakas, ok zooming on which line? The OII doublet ?

[moustakas]

Sure.

…

On Wed, Jun 27, 2018, 8:40 PM Hélion du Mas des Bourboux < ***@***.***> wrote: @moustakas <https://github.com/moustakas>, ok zooming on which line? The OII doublet ? — You are receiving this because you were mentioned. Reply to this email directly, view it on GitHub <#103 (comment)>, or mute the thread <https://github.com/notifications/unsubscribe-auth/ABXZDH9YbthSbJrq3tNohHEoQFw5Q8m2ks5uBCYVgaJpZM4SW-gj> .

[londumas]

Here is an example. Not very easy to see anything. Maybe our best way is using GAIA or something of the sort.

[londumas]

The science requirement document asks for less than 60km/s systematic error: https://desi.lbl.gov/DocDB/cgi-bin/private/RetrieveFile?docid=318;filename=DESI_L123_driver.pdf;version=6. This error of 10 km/s is thus under the requirement.

[londumas]

Reopening this issue because even though it is not very important for the science of DESI, it could be important for people using the catalogs from DESI to do other type of science. One could easily investigate this issue by taking one or two standard objects and see what redshift is missing.