Emissions ScenarioMIP (Testing): CMIP6Plus 0-3-0

[jkikstra]

The first set of emissions scenarios is available for testing.
It includes two different future scenarios: scendraft1 (lower emissions) and scendraft2 (higher emissions).

This data is for testing (both of the forcing data and of modelling workflows) only.
It is therefore placed under "CMIP6Plus"

• IIASA-IAMC-scendraft1-0-3-0
• IIASA-IAMC-scendraft2-0-3-0

input4MIPs web page:
https://input4mips-cvs.readthedocs.io/en/latest/dataset-overviews/anthropogenic-slcf-co2-emissions/#scenariomip

Changelog from earlier versions (which were shared in smaller groups), with additional description (this document will be populated further in the coming weeks):
https://docs.google.com/document/d/1E7Wv2APCRY-LRfI6II9pkwtfySGvRm2zPE2oyYP6mak/edit?usp=sharing

At this point we welcome as much feedback as possible!

Some possible things we would love feedback on:

• Is all data that you need there?
• Does the metadata look alright?
• Does the data not create any weird jumps with historical emissions data?
• Are there any other data quality issues that would create problems for CMIP7 runs?

[eleanororourke]

@cecilehannay, @MartinDix, @jmulcahy11, @seferian - the IAM teams are really keen to have feedback on these test emissions scenarios datasets as soon as possible (ideally end of week 31/10/25). This would allow them to finalise the data of the first two scenarios in the running order (H and VL). Delay in this will delay the dependent datasets.

[GregoryFaluvegi]

I have started to take a look at these samples to alter/test GISS ModelE workflow for incorporating CMIP emissions. While I don't have much to say about your bullet points so far (sorry) I did come up with a few initial questions:

1. we will not get CO2 for openburning, right? I do not see it in the current batches, and I seem to recall this is consistent with CMIP6...

2. Time appears to be yearly 2022-2025 and then in 5-year steps to 2100. But, to my eye, 2024 is missing from scendraft2-0-3-0 but exists in scendraft1-0-3-0?

3. If I am not hallucinating, back in CMIP6 we had speciated VOCs for openburning (e.g.: NMVOC-C3H8-em-speciated-VOC-openburning*nc). In these sample sets I only find bulk NMVOC for openburning. Does this sound right? Is the plan for the final emissions to have speciated or no?

Many thanks for all your work,
Greg

[jkikstra]

Thanks a lot @GregoryFaluvegi for starting to look at this batch of test data!

1. correct

2. also correct - I'm afraid in creating this 0-3-0 set, one script did not run through correctly for scendraft2-0-3-0.

3. correct again. Speciated VOC were not ready yet in 0-3-0, but will be included with the next set, 0-4-0, soon.

order of dimensions

that's a good one for us to double-check before finalising 0-4-0.
I assume (would be great if you can confirm) that the best way to go for a scenario timeseries is virtually always to conform to the historical timeseries?

[GregoryFaluvegi]

Thanks for confirming 1, 2, 3. On the order of dimensions I would guess (and this is only a guess as far as other people are concerned) that the same order as the CMIP6 SSP files would be most 'expected'. Whether that's the same as the CMIP7 era historical, I am having a hard time checking at the moment because of some download problems.
(And when there is a sector dimension, then the same order as the historical, yes.)

[GregoryFaluvegi]

Just wanted to add: the practice of putting the global totals as meta-data in the files (often they put first year and last year totals) is very useful for people like me who always worry they did something wrong. It is a great way to check 'final' ESM input files...

[GregoryFaluvegi]

Just in case it leads to something not already known: some emissions go negative at times. I happened to noticed this at:
NH3 Agri. Waste Burn., 2100 Nov
CH4 Agri. Waste Burn., 2080 Dec
...scendraft1

[GregoryFaluvegi]

Since you asked specifically about jumps vs. historical: apologies in advance if I made any mistakes on my end. If so, please let me know! Most things seem to match up well, but here are a few issues I noticed (for the species I had to deal with at least):

• It seems to me most species' anthro sector 1 (ENE) are 0 in the drafts 1 and 2? Exceptions: speciated VOCs and CO2 seem to meet up with historical OK (at least for the VOC species I used)
• Some speciated VOCs which had non-zero sector 5 (SLV) in the historical are now 0 "immediately".
• Certain speciated VOCs for the sector 4 (RCO) meet up nicely for draft1 but are 0 for draft 2 "immediately" e.g. VOC23-ketones, VOC22-other-alka, VOC21-methanal.

That's all from me! Thanks.

[cecilehannay]

Hi @eleanororourke:
We downloaded the "CMIP6Plus" emissions for scenarioMIP

IIASA-IAMC-scendraft1-0-3-0
IIASA-IAMC-scendraft2-0-3-0

@bgaubertNCAR is going to look at them and run his scripts to process them to see if he runs into any issues.

[matrss]

Hey!

I am looking at this dataset right now in preparation for us at FZJ. So far the only issue I found is that I am missing C2H2 and H2 biomass burning emissions. If I understand correctly then C2H2 should be part of the speciated openburning VOCs coming with the next version, but what about H2? I can't find a mention of it anywhere.

[jkikstra]

OK, so I had a quick look at CMIP6 scenario files first. Hope I did this correctly.
Multiplied total emissions by the share files to obtain sectoral emissions.
Then divided the sectoral emissions to get the CO/H2 ratio per gridcell.

Contrary to my expectation, I do not obtain a single value for each gridcell.
Perhaps (speculating here) one of the reasons could be that there was some regridding (/spatial aggregation) done, as I am looking here at 0.5deg gridcells?

Savanna (sector=2)
While most datapoints are somewhat in the range of the value above (37.06), the actual values per gridpoint can differ quite a lot.

Agricultural Waste Burning (sector=0)
Here, the same thing as above, with 1 specific gridcell being on a different order of magnitude (so that gridcell could be a good candidate to figure out why the factors can vary).

[jkikstra]

CMIP7 historical files
Here we can look directly at the (native) 0.25 degree resolution.

Savanna (SAVA)

OK, here all values are basically exactly 37.06, so indeed using the suspected ratio that I mentioned above, indeed following what @gidden explained.

Details (code underlying the analysis)

# ---
# jupyter:
#   jupytext:
#     text_representation:
#       extension: .py
#       format_name: percent
#       format_version: '1.3'
#       jupytext_version: 1.16.4
#   kernelspec:
#     display_name: Python 3 (ipykernel)
#     language: python
#     name: python3
# ---

# %%
import xarray as xr
from pathlib import Path
import matplotlib.pyplot as plt
import numpy as np
from pathlib import Path

# %% [markdown]
# # Check that CMIP6 scenario files and their EFs are consistent with the auxiliary files

# %%
# load data
bb_data_cmip6_location = Path("C:/Users/kikstra/IIASA/ECE.prog - Documents/Projects/CMIP7/IAM Data Processing/concordia_cmip7_v0-4-0/input/gridding/esgf/cmip6")
h2_total = xr.open_dataset(bb_data_cmip6_location / "H2-em-openburning_input4MIPs_emissions_ScenarioMIP_IAMC-MESSAGE-GLOBIOM-ssp245-1-1_gn_201501-210012.nc")
h2_perc_allsectors = xr.open_dataset(bb_data_cmip6_location / "H2-openburning-share_input4MIPs_emissions_ScenarioMIP_IAMC-MESSAGE-GLOBIOM-ssp245-1-1_gn_201501-210012.nc")
co_total = xr.open_dataset(bb_data_cmip6_location / "CO-em-openburning_input4MIPs_emissions_ScenarioMIP_IAMC-MESSAGE-GLOBIOM-ssp245-1-1_gn_201501-210012.nc")
co_perc_allsectors = xr.open_dataset(bb_data_cmip6_location / "CO-openburning-share_input4MIPs_emissions_ScenarioMIP_IAMC-MESSAGE-GLOBIOM-ssp245-1-1_gn_201501-210012.nc")

# set variables to look at
SECTOR = 2 # Grassland Burning (SAVA)
TIME = "2015-01-16" # first year


# %%
# calculate savannah emissions
h2_sava_cmip6 = (
    h2_total.sel(time=TIME).H2_em_openburning *
    h2_perc_allsectors.sel(time=TIME,sector=SECTOR).H2_openburning_share
)
co_sava_cmip6 = (
    co_total.sel(time=TIME).CO_em_openburning *
    co_perc_allsectors.sel(time=TIME,sector=SECTOR).CO_openburning_share
)

# calculate the gridpoint ratios
co_div_h2_sava_cmip6 = co_sava_cmip6 / h2_sava_cmip6

# plot map
co_div_h2_sava_cmip6.plot()

# min/max
co_div_h2_sava_cmip6.min()
co_div_h2_sava_cmip6.max()
co_div_h2_sava_cmip6.mean()

# plot histogram of values
plt.figure(figsize=(10, 5))
plt.hist(co_div_h2_sava_cmip6.values.flatten(), bins=50, edgecolor='black', alpha=0.7)
plt.xlabel('CO / H2 Ratio')
plt.ylabel('Frequency')
plt.title('Distribution of CO/H2 Ratios (Grassland Burning)')
plt.grid(True, alpha=0.3)
plt.tight_layout()
plt.show()

# %%
# calculate agricultural waste burning emissions
h2_awb_cmip6 = (
    h2_total.sel(time=TIME).H2_em_openburning *
    h2_perc_allsectors.sel(time=TIME,sector=0).H2_openburning_share
)
co_awb_cmip6 = (
    co_total.sel(time=TIME).CO_em_openburning *
    co_perc_allsectors.sel(time=TIME,sector=0).CO_openburning_share
)

# calculate the gridpoint ratios
co_div_h2_awb_cmip6 = co_awb_cmip6 / h2_awb_cmip6

# plot map
co_div_h2_awb_cmip6.plot()

# min/max
co_div_h2_awb_cmip6.min()
co_div_h2_awb_cmip6.max()
co_div_h2_awb_cmip6.mean()

# plot histogram of values
plt.figure(figsize=(10, 5))
plt.hist(co_div_h2_awb_cmip6.values.flatten(), bins=50, edgecolor='black', alpha=0.7)
plt.xlabel('CO / H2 Ratio')
plt.ylabel('Frequency')
plt.title('Distribution of CO/H2 Ratios (AWB)')
plt.grid(True, alpha=0.3)
plt.tight_layout()
plt.show()

# list of values - remove np.nan and order
co_div_h2_awb_cmip6_values = np.sort(co_div_h2_awb_cmip6.values.flatten()[~np.isnan(co_div_h2_awb_cmip6.values.flatten())])
print(co_div_h2_awb_cmip6_values)

# %% [markdown]
# # Check CMIP7 historical files and their EFs (consistent with GFED4.1s?)
bb_data_cmip7_location = Path("C:/Users/kikstra/IIASA/ECE.prog - Documents/Projects/CMIP7/IAM Data Processing/concordia_cmip7_v0-4-0/input/gridding/esgf/bb4cmip7")

ds_h2_total = xr.open_dataset(bb_data_cmip7_location / "H2_input4MIPs_emissions_CMIP_DRES-CMIP-BB4CMIP7-2-1_gn_190001-202312.nc")
ds_h2_perc_sava = xr.open_dataset(bb_data_cmip7_location / "H2percentageSAVA_input4MIPs_emissions_CMIP_DRES-CMIP-BB4CMIP7-2-1_gn_175001-202312.nc")
ds_co_total = xr.open_dataset(bb_data_cmip7_location / "CO" / "gn" / "v20250612" / "CO_input4MIPs_emissions_CMIP_DRES-CMIP-BB4CMIP7-2-1_gn_190001-202312.nc")
ds_co_perc_sava = xr.open_dataset(bb_data_cmip7_location / "COpercentageSAVA" / "gn" / "v20250612" / "COpercentageSAVA_input4MIPs_emissions_CMIP_DRES-CMIP-BB4CMIP7-2-1_gn_175001-202312.nc")

# %%
# calculate savannah emissions
h2_sava_cmip7 = (
    ds_h2_total.sel(time=TIME).H2.squeeze() *
    ds_h2_perc_sava.sel(time=TIME).H2percentageSAVA.squeeze()
)
co_sava_cmip7 = (
    ds_co_total.sel(time=TIME).CO.squeeze() *
    ds_co_perc_sava.sel(time=TIME).COpercentageSAVA.squeeze()
)


# %%
# calculate the gridpoint ratios
co_div_h2_sava_cmip7 = co_sava_cmip7 / h2_sava_cmip7.plot()

# plot map
co_div_h2_sava_cmip7.plot()

# %%
# plot histogram of values
plt.figure(figsize=(10, 5))
plt.hist(co_div_h2_sava_cmip7.values.flatten(), bins=50, edgecolor='black', alpha=0.7)
plt.xlabel('CO / H2 Ratio')
plt.ylabel('Frequency')
plt.title('Distribution of CO/H2 Ratios (SAVA) CMIP7 DRES 2-1')
plt.grid(True, alpha=0.3)
plt.tight_layout()
plt.show()
# %%

[jkikstra]

OK, have implemented the translation based on the GFED4.1s emissions factors now, as that is what was used for CMIP7 historical openburning emissions.
So H2_em_openburning should be there in the next version.

[matrss]

Great, thanks!

[durack1]

@mjevanmarle ping! You might be able to enlighten @jkikstra with methodology insights