NRL SSI data is now v3

[mathomp4]

@sdrabenh asked about updating the NRLSSI files that @dr0cloud was making for us. And since we have the scripts used in here, I thought "Should be easy".

Well, nope. NRL moved to a new v3. For example, before the tsi-ssi files looked like:

time (yyyy-MM-dd), sunspot_darkening_function, facular_brightening_function, source
1882-01-01,165.59720,0.15277700,final
1882-01-02,251.05570,0.15338300,final
1882-01-03,370.62390,0.15312100,final
1882-01-04,155.95590,0.15305900,final
1882-01-05,368.64070,0.15228000,final
1882-01-06,513.66220,0.15237600,final
1882-01-07,471.04350,0.15339100,final
1882-01-08,204.96640,0.15354100,final
1882-01-09,67.623100,0.15281300,final

and with v3:

time (yyyy-MM-dd),Bolfac,Bolfac_unc,Bolfac_sm,Bolspot,Bolspot_unc,Ftcomp,Ftcomp_unc,source
1874-05-09,0.974380,0.125400,0.623336,0.658303,0.093900,137.641998,2.905400,final
1874-05-10,0.960666,0.125100,0.618316,0.681516,0.097100,137.938293,2.911100,final
1874-05-11,0.867380,0.120800,0.613791,0.494231,0.071400,130.532303,2.812800,final
1874-05-12,0.753316,0.116600,0.609493,0.312700,0.046500,124.715797,2.775100,final
1874-05-13,0.637969,0.111900,0.605439,0.276922,0.041600,116.454803,2.752600,final
1874-05-14,0.553863,0.108700,0.601768,0.220943,0.033900,110.703796,2.771300,final
1874-05-15,0.498240,0.106700,0.598877,0.155193,0.025000,107.155098,2.799700,final
1874-05-16,0.454371,0.105300,0.596568,0.106697,0.018400,104.368103,2.841000,final
1874-05-17,0.416254,0.104100,0.595575,0.047383,0.010400,101.467598,2.903900,final

I found some docs here:

https://www.ncei.noaa.gov/pub/data/sds/cdr/CDRs/Total%20Solar%20Irradiance/AlgorithmDescription_01B-32.pdf

I'm going to go to ChatGPT and see what I can see...

[mathomp4]

Note, GEOS ingests data like:

> head NRLSSI2/NRLSSI2.v2023.txt
# NRLSSI2 daily input
# treat daily values as valid at 12:00 GMT
# yyyy doy TSI:W/m2 MgIndex   SBindex
  1882 001 1360.808 0.152777  165.5972
  1882 002 1360.844 0.153383  251.0557
  1882 003 1360.740 0.153121  370.6239
  1882 004 1360.852 0.153059  155.9559
  1882 005 1360.624 0.152280  368.6407
  1882 006 1360.555 0.152376  513.6622
  1882 007 1360.721 0.153391  471.0435

[mathomp4]

Okay. I yelled at ChatGPT for a bit and it figured out...something:

[Mg_SB] Fitted v03->legacy:
  Mg:  Mg_v02 ≈ 0.007506 * Bolfac + 0.150331  (R^2=0.9049, n=51864)
  SB:  SB_v02 ≈ 1941.389784 * Bolspot + 42.906433  (R^2=0.9940, n=51864)

It apparently did a linear fit.

I would be more confident of that if it wasn't "magic numbers". That is "Oh yeah, 1941.389784 is the Johnson-Foobar Constant" or whatever