add gas weight conversion for N2O to N and fix weight conversion for SC-SF6

[corakingdon]

I think that computing the social cost of N2O was previously returning the social cost per ton of N not N2O, because the units in the emissions component are in tons of N. This PR normalizes the pulse size to be in tons of N2O instead

[corakingdon]

@davidanthoff I also just noticed in the documentation that the units for SF6 emissions in FUND are "million metric tonnes of carbon dioxide equivalent." Does that mean that in order to report the SC-SF6 in units of dollars per ton of SF6, we would actually need to multiple by some GWP? (i.e., is the compute_sf6 function currently returning the value in $/ton of CO2e?)

[lrennels]

@davidanthoff to revisit ASAP

---

For N2O we have this change already in version 3.8.6 used for IWG, so I'm confident we should change it here ... and I also checked with the below units validation.

We connect our marginal pulse into the globn2o variable of the climaten2ocycle component, which pulls from the same variable in emissions, which is then based on parameter n2oem in emissions. This is exogenously forced.

global check

• Looking at explore(m) globn2o is about 4.6 in 1990 and 10.1 in 2100.
• The IPCC scenario they are based on shows 4.6 million tons N in 1990 and 8.7 in 2100.
• Assuming the difference in 2100 is some calibration step, we focus on 1990 and see thet globn2o must indeed be in million tons of N not million tons of N2O

USA check

• the EPA shows about 460 million metric tons CO2eq of N2O in 2020 in the US = 1.5 million metric tons N2O = 0.95 million metric ton N which matches the values coming out of emissions:n2o around this time

---

For SF6 I don't think we're using CO2 eq given the conversion is 1:22,600, but the comments do say that SF6 is kt not Mt ... and given this which puts us 8,100 tons in 2020 this acutally seems about right ... so we are off by a factor of 1000.

This is corroborated by:

• https://www.ipcc-nggip.iges.or.jp/public/gp/bgp/3_9_Global_Sources_Industrial_Processes_SF6.pdf
• https://acp.copernicus.org/articles/20/7271/2020/
• PAGE2002 Social Cost of SF6 is around 200k
• FUND paper here http://www.economics-ejournal.org/dataset/PDFs/journalarticles_2014-31.pdf

we removed SF6 from version 3.8.6 for clarity

[codecov-commenter]

Codecov Report

Merging #68 (46dd1b5) into master (45cd6b8) will increase coverage by 0.18%.
The diff coverage is 90.00%.

@@            Coverage Diff             @@
##           master      #68      +/-   ##
==========================================
+ Coverage   79.10%   79.28%   +0.18%     
==========================================
  Files          39       39              
  Lines         938      956      +18     
==========================================
+ Hits          742      758      +16     
- Misses        196      198       +2     

Flag	Coverage Δ
unittests	79.28% <90.00%> (+0.18%)	⬆️

Flags with carried forward coverage won't be shown. Click here to find out more.

Impacted Files	Coverage Δ
src/new_marginaldamages.jl	91.15% <90.00%> (-0.43%)	⬇️

---

Continue to review full report at Codecov.

Legend - Click here to learn more
Δ = absolute <relative> (impact), ø = not affected, ? = missing data
Powered by Codecov. Last update 45cd6b8...46dd1b5. Read the comment docs.

[lrennels]

use these to gut-check from
source: http://www.economics-ejournal.org/dataset/PDFs/journalarticles_2014-31.pdf