8% difference between AGTP100 values using ar6 scripts and values reported in Table 7.SM.7 AR6 WGI

[TitouanGreffe]

I tried to recompute AGTP100 values for CH4 and N2O using ch4_analytical and n2o_analytical but I face this difference:

AGTP100 for CH4 I obtain is 1.9438492024794e-15 while it is reported as 2.121780e-15 in Table 7 SM7 i.e. a 8.49% difference (https://github.com/chrisroadmap/ar6/blob/main/notebooks/340_chapter7_table7.SM.7.ipynb)

AGTP100 for N2O I obtain is 8.683614565204947e-14 while it is reported as 9.192690e-14 in Table 7 SM7 i.e. a 5.5% difference (https://github.com/chrisroadmap/ar6/blob/main/notebooks/340_chapter7_table7.SM.7.ipynb)

`ts_per_year = 1
H_max = 100 # years

co2 = 409.9
ch4 = 1866.3
n2o = 332.1`

For CH4:
rf_ch4, agwp_ch4, agtp_ch4, iagtp_ch4 = ch4_analytical( np.linspace(0, H_max, H_max * ts_per_year + 1), d=np.array([3.424102092311, 285.003477841911]), #q=np.array([0.443767728883447, 0.319591049742508]), # this is what is in Zeb's file sent 25 Feb q = np.array([0.443767728883447, 0.313998206372015]), # this is what Bill used co2=co2, ch4=ch4, n2o=n2o, ch4_ra = -0.14, alpha_ch4 = 11.8, ch4_o3=1.4e-4, ch4_h2o=0.00004 )

For N2O:

rf_n2o, agwp_n2o, agtp_n2o, iagtp_n2o = n2o_analytical( np.linspace(0, H_max, H_max * ts_per_year + 1), d=np.array([3.424102092311, 285.003477841911]), #q=np.array([0.443767728883447, 0.319591049742508]), # this is what is in Zeb's file sent 25 Feb q = np.array([0.443767728883447, 0.313998206372015]), # this is what Bill used co2=co2, ch4=ch4, n2o=n2o )

Moreover, 335_chapter7_generate_metrics,ipynb to calculate AGTP100 values does not work for at least two reasons:

1. meinshausen function is missing in fair.forcing.ghg. I do not know if the function meinshausen2020 is supposed to replace it or not.
2. from fair.defaults.thermal import q, d raise the following error: ModuleNotFoundError: No module named 'fair.defaults.thermal' both using latest fair version or version 2.1.0 which may have been used when ar6 package was released.

I did not find a way to solve this issue as well.

[chrisroadmap]

Hi, yes this is legacy code that has become a bit outdated (it was right at the time, and worked at the time, but things move on and our software development skills were not as good back then).

in your environment, if you roll the fair version to 1.6.4, this should fix the problems 1 and 2 above you identify. You could reproduce the results with 2.x but the interface is different.

metrics like GWP and GTP are the bane of my life. But anyway, I wonder if the difference arises because you don't include the carbon cycle adjustment (cells 10 and 11, notebook 335).

[TitouanGreffe]

Hi, thanks for your very quick answer.
I used fair version 1.6.4 and then packages were uploaded! I run all cells of "335_chapter7_generate_metrics.ipynb", including carbon_cycle_adjustment (i.e. cell 10, i.e. needed for calculting AGTP100 of CH4) and n2o_analytical (i.e. cell 12, i.e. needed for calculting AGTP100 of N2O).
Then for CH4 i.e. table[1]['AGTP100'] is equal to 1.3682387486072224e-15, still far from 2.121780e-15 in Table 7 SM7
For N2O i.e. table[2]['AGTP100'] is equal to 8.45665472950499e-14, still far from 9.192690e-14 in Table 7 SM7.

Obviously, the fair package have been updated and improved since v1.6.4, but I wonder if values presented in Table 7 SM7 were computed with an older version of ar6.metrics functions because for the moment I cannot explain such difference.
Could you share the code used to obtain values displayed in Table 7 SM7?

I thank you very much for your time and your future answer.