Update to throughputs in v16 of DESI-347 about desimodel

[dkirkby]

I am working on this now...

[dkirkby]

We are still using v13 throughputs but there is now an approved v16 (18 Mar 2020) which fixes an error in the blue fiber transmission. Quoting from DESI-5647:

The disagreement in the throughput of the BLUE cameras reported in the
previous version of this note has been traced back to a mistake in the fiber
transmission data used in DESI-347. This is now fixed in version v15.2 of this
document.

@julienguy @sbailey Is anyone already working on this or shall I do it?

[dkirkby]

The names of both of the files that throughput.update reads from DESI-347 have changed, as well as the spreadsheet row with the total throughput. After fixing these, I get updated read noise and dark currents in desi.yaml and new throughputs in throughput/thru-[brz].fits.

However, I want to check that these throughputs are correct as calculated:

thru = (DESI-347 "sky thru") / (DESI-347 "spectro thru") * (DESI-334 "xxx-thru.txt")

In particular, does DESI-334 need to be updated to as-built throughputs in order to get self-consistent results? @julienguy

[dkirkby]

Here are the throughput changes from v13 (current DESIMODEL) to v16 (current DESI-347):

The changes are quite significant on the blue end but are still missing the narrow collimator absorption feature.

[julienguy]

In your formula above it looks like the normalization of the spectrograph throughput is defined by DESI-334 which is from 2014, whereas the as-built throughput per spectro is given in DESI-347 (tab 'spectrograph throughput'). So that doesn't look correct. More on this offline (with Pat Jelinksy in the loop).

[julienguy]

As for the dip in the blue, yes, this is due to a correction in the fiber transmission.

[julienguy]

The fine grid throughput per spectrograph is in https://desi.lbl.gov/DocDB/cgi-bin/private/ShowDocument?docid=5501

[dkirkby]

Unfortunately there is no average spectrograph throughput in 5501 so we either need to add that there or else update throughput.update to loop over each spectrograph and do its own average.