Extend historical volcanic forcing to 2300

[penguian]

Closes issue #92

[atteggiani]

Thank you @penguian.

This looks good.
I left two comments, but I don't think they are extremely important, so feel free to merge without addressing them.

[MartinDix]

Is there an example output file somewhere?

[penguian]

@MartinDix See /g/data/tm70/pcl851/CMIP7/esm16_ancil/modern/historical/atmosphere/forcing/resolution_independent/2025.11.25/volcts_cmip7.dat

[MartinDix]

Data in this file seems to have a one month time offset. E.g. Pinatubo erupted in June 1991 and the input4MIPS and ESM1.5 CMIP6 file show AOD increasing in July but it's not until August in your file

/g/data/vk83/experiments/inputs/access-esm1p5/modern/historical/atmosphere/forcing/resolution_independent/2021.06.22/volcts_cmip6.dat
1991     4    97    68    75  103
1991     5    93    70    76  107
1991     6   165   109    78  116
1991     7   304   465   400  161
1991     8   422  1415   964  379

/g/data/tm70/pcl851/CMIP7/esm16_ancil/modern/historical/atmosphere/forcing/resolution_independent/2025.11.25/volcts_cmip7.dat
1991    4   79.1   49.1   44.5   65.0
1991    5   77.3   48.7   44.0   65.7
1991    6   81.6   49.2   44.7   60.8
1991    7   86.4   50.7   45.0   75.0
1991    8  108.2  268.2  463.8  258.2
1991    9  294.4  722.4  843.4  322.3

[penguian]

Good catch. See /home/851/pcl851/gdata/CMIP7/esm1p6_ancil/2025.11.27/modern/historical/atmosphere/forcing/resolution_independent/2025.11.27/volcts_cmip7.dat for the corrected file output.

[MartinDix]

Pinatubo now looks spot on, but there are some other eruptions where the peak in your file is different to what I get from the data.

Zooming in on Fig 5 of https://docs.google.com/document/d/1blX5kv0We1BteqWzMKs0OuhazAcAonay I estimate 0.081 for the 1912 peak, so agreeing with my value. This eruption was at high N latitude so perhaps something off with the latitude weights?

The band order is also reversed in your file. The code is awkward but https://github.com/ACCESS-NRI/UM7/blob/f69fbab14ab57dcfa26117a9ffd9ccab899d293f/umbase_hg3/src/control/top_level/glue_rad-rad_ctl2.F90#L2325-L2342 shows that the first band is 30N-90N.

/g/data/tm70/pcl851/CMIP7/esm1p6_ancil/2025.11.27/modern/historical/atmosphere/forcing/resolution_independent/2025.11.27/volcts_cmip7.dat
1912    5   44.1   28.9   26.3   46.6
1912    6   46.7   30.8   71.1  345.7
1912    7   53.9   62.8  207.4 1214.9

/g/data/vk83/experiments/inputs/access-esm1p5/modern/historical/atmosphere/forcing/resolution_independent/2021.06.22/volcts_cmip6.dat
1912     5   142    50    51  64
1912     6   592    68    52  68
1912     7  1387   127    54  72

See https://github.com/ACCESS-NRI/CMIP7-Input/blob/92-historical-volcanic-check/examples/esm1p6/historical/atmosphere/volcanic/volcanic_aod.ipynb

[penguian]

As for the latitude bands, I did check the Fortran code, but didn't notice that the first index used in VOLCTS was decreasing from 4 down to 1. I will rectify this.

[penguian]

See the last commit and see gadi:/g/data/tm70/pcl851/CMIP7/esm1p6_ancil/2025.12.01/modern/historical/atmosphere/forcing/resolution_independent/2025.12.01/volcts_cmip7.dat for the updated output. I used cell.point in the comparison as well as reversing the order of the latitude bands.

The weighting at CMIP7/esm1p6/atmosphere/volcanic/cmip7_volcanic.py#L31 uses iris.analysis.cartography.cosine_latitude_weights(cube) after constraining the cube to a latitude band.