some negative values in openburning of IIASA-IAMC-vl-1-0-0

[GregoryFaluvegi]

Dataset source ID(s)

Dataset source ID(s): IIASA-IAMC-vl-1-0-0

Describe the issue

I noticed there are some negative emissions in the openburning. The ones I noticed were all in the AWB (sub-)sector. (In the single case I printed below, the negative min is small compared to the positive max. I haven't done a full search through the data, but I think you'll find negatives in multiple species at least in years 2030, 2045, 2065, and maybe 2100.)

Expected data

I was expecting that the only negative values allowed would be in the CO2 anthro sectors "8: BECCS; 9: Other Capture and Removal."

Screenshots

Looking at CO in the "h" scenario, the min is 0:

ncap2 -O -s 'print(CO_em_openburning.min()); print(CO_em_openburning.max());' CO-em-openburning_input4MIPs_emissions_ScenarioMIP_IIASA-IAMC-h-1-0-0_gn_202201-210012.nc nowhere.nc
CO_em_openburning, size = 1 NC_DOUBLE, value = 0
CO_em_openburning, size = 1 NC_DOUBLE, value = 2.11025866961e-07

But same on the "vl" scenario has negative(s):

ncap2 -O -s 'print(CO_em_openburning.min()); print(CO_em_openburning.max());' CO-em-openburning_input4MIPs_emissions_ScenarioMIP_IIASA-IAMC-vl-1-0-0_gn_202201-210012.nc nowhere.nc
CO_em_openburning, size = 1 NC_DOUBLE, value = -2.81027715356e-09
CO_em_openburning, size = 1 NC_DOUBLE, value = 3.50798498511e-07

[durack1]

@jkikstra any chance you have some insights here, or can redirect to the data provider

[jkikstra]

Thanks for the ping @durack1 and thanks for bringing it up @GregoryFaluvegi.

Having looked at this now for a very brief time, I'm afraid it's likely a bug.
My first observations and suspicions: (a) it relatively small [for CO: -0.2 MtCO/yr maximum global-sum negative, relative to 22.6 in 2023], (b) it affects all species [BC, CH4, CO, N2O, NH3, NMVOC, NOx, OC, SO2], (c) the original scenario goes to zero, but not below, (d) my suspicion is this is a problem caused by a downscaling choice that relates to base year patterns which does not account well for a phase out of agricultural waste burning that is this rapid.
Will investigate more and advise how to go forward ASAP.

[GregoryFaluvegi]

Thanks you for looking into it. If it turns out to be a large enough bug that you have to fix it :-(, then I would also note in case it is related, the very small positive values sometimes seen. Example:
CO-em-anthro_input4MIPs_emissions_ScenarioMIP_IIASA-IAMC-vl-1-0-0_gn_202201-210012_annual_totals_by_sector.csv (the global totals) near the end of the timeseries, last sector, a global total of: 4.4115077546550165e-21 instead of 0.

[jkikstra]

Thanks you for looking into it. If it turns out to be a large enough bug that you have to fix it :-(, then I would also note in case it is related, the very small positive values sometimes seen. Example: CO-em-anthro_input4MIPs_emissions_ScenarioMIP_IIASA-IAMC-vl-1-0-0_gn_202201-210012_annual_totals_by_sector.csv (the global totals) near the end of the timeseries, last sector, a global total of: 4.4115077546550165e-21 instead of 0.

OK thanks, will check that too!

[vnaik60]

Hi @jkikstra - circling around to see if you have an ETA on updated files? Thanks!

[jkikstra]

Hi @jkikstra - circling around to see if you have an ETA on updated files? Thanks!

Aiming for upload by Tuesday.

[jkikstra]

I believe that this has been resolved now.
The upload is running on my side, will inform the forcings task team leads tomorrow.

As far as I can see for now, it does not have any material implications, but it's good to provide an update nonetheless to remove these tiny negative values (and also capture a different small update in the CO2 sector Other Capture and Removal; which did not affect concentrations).

[jkikstra]

Forget to write it here earlier, but just coming in to say that this should have been resolved with the v1-1-0 upload to ESGF some time ago.

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

As far as I'm concerned, this can be closed.

How does it look for you @GregoryFaluvegi ?

[GregoryFaluvegi]

Hello. I admit that I did not check as closely this time! (both because of other projects and because it looked to me that the latest available csv 'checksum' files are still 1-0-0.) When I used those 1-0-0 csv files vs 1-1-0 nc, I did notice some changes > 1% in the OCR, if I remember correctly -- but you did flag this in your comment in this thread already.

In any case, I think the original problem is fixed, as I just checked year 2100 CO ag waste burning minimum and it's 0.
So, thank you for your attention to this issue.