search

IAMconsortium / concordia

Apache License 2.0
0 stars 3 forks source link

Revisions to harmonization draft #9

Closed gidden closed 7 months ago

gidden commented 1 year ago

Following our discussion on 2023-08-16:

Mattermost discussion regarding CO2:

@gidden Let me start a separate thread on harmonization issues - and please let me know if you would like me to put this on github as well for posterity. I am looking at the results for the cp044-OAE_off run. Note that there are now significant discrepancies in the near term for CO2. This was not the case for the previous set of scenario results . It looks like the primary reason for the change between versions is due to revisions in industry (further away) and transportation (previously underestimated, now closer).

@nicobauer OK, what about the change from 2015 to 2020. REMIND increases emissions from energy and industry by slightly more than a giga ton of CO2 (1050 Mt more precisely). A look into CEDS shows for the period 2015 to 2019 an increase of 650 and for EDGAR from 2015 to 2018 an increase of 550. Industrial process increased about 100Mt in both cases, but there some tendency towards stagnation. Now a look into BP statistics indicates an increase of energy related emissions by 960MtCO2. The increases are consistent with the energy part (net of process emissions) compared with CEDS and EDGAR. In REMIND the emissions are mostly driven upwards by the energy supply sector (about 700Mt). In CEDS the increase from 2015 to 2019 is nearly 490Mt and in EDGAR from 2015 to 2018 it is also about 480Mt. Conclusion: there are differences in the accounting of emissions that imply differences in CO2 emissions 2015. Assuming a change in emissions from 2015 to 2020 of about +1GtCO2 seems reasonable. The increase that I see in the figure for the harmonized data seems too small.

coroa commented 1 year ago

Updated harmonization with ratio overrides for chn CO2 is available as version 2023-08-18 from http://rescue-task1.3.s3-website.eu-north-1.amazonaws.com/harmonization/harmonization-2023-08-18-facet-PkBudg_cp0400-OAE_on-plotly.html#co2-energy-sector . This closes the gap somewhat, but not substantially.

merfort commented 1 year ago

Hi Matt and Jonas, there is a new version of of the REMIND-MAgPIE output in CEDS format, in which energy sector emissions are split further into Energy Sector|Supply and Energy Sector|Extraction. You find it in the google drive project folder under "Task 1.3 > Harmonization > REMIND-MAgPIE Output > REMIND-MAgPIE-CEDS-RESCUE-Tier1-2023-09-14.csv". Could you try to rerun the harmonization script to check, if the new split aligns with the historic data? Please note that also the scenarios themselves got updated again, so also the other emissions may differ slightly compared to the previous version. Additionally to the scenarios that were designed to be fed into the emulator for extension after 2100 ("Extension") there are now also the scenarios that are used directly in the ESMs ("Direct").

merfort commented 1 year ago

As for the other open issue with the OAE emissions, this will take a bit longer and won't be finished before the GA next week.

coroa commented 1 year ago

Thanks, leon (@merfort). Can you please confirm the attached sector mapping. Updates in yellow, ie. 1B1 (fugitive solid fuels) and 1B2 (fugitive petrol fuels) emissions are extraction (even though they also contain some amount of emissions happening during refining f.ex.), while 1A1 emissions are then the supply emissions.

sector_mapping_extraction.xlsx

merfort commented 1 year ago

There is just one variable, that we would map differently: 1B2b_Fugitive-NG-distr is part of Energy Sector|Supply (instead of Energy Sector|Extraction), since it is rather a question of how the energy system distributes the energy. Apart from that, we can confirm the mapping, thanks!

coroa commented 1 year ago

@merfort Hmm ... my initial reaction after looking at the new harmonization plots (facet) is, that the split seems to increase the differences between model and history.

I was also wrong: with the new sector mapping there are quite some co2 extraction emissions.

merfort commented 1 year ago

@coroa Thanks for the new plots! I just go through all the gases with a few comments:

gidden commented 1 year ago

Hi all, I did a quick look as well - thanks @merfort for already going through.

In short, separating out extraction emissions without additional overrides (e.g., 2100 offset convergence) has made the overall harmonization worse (e.g., total CO2 before and after). I would suggest either applying the overrides for this sector or reverting back to our previous approach.

I would lean towards reverting back (ie., not including extraction explicitly), because in most (all?) cases, we use a convergence method anyway. That means that this offset goes to 0 'automatically'. At least this is true for GHGs, but perhaps aerosols like VOC not.

VOCs I agree is odd, but I think this may be a sector mapping or aggregation issue @coroa - in the 8-18 version, total VOCs in energy were about 50 Mt. In the 9-16 version supply is about 20 and extraction is about 50. So we are in excess somehow.

So in short, I agree we need to double check the process and compare against the previous version.

gidden commented 1 year ago

@merfort - we figured out that CEDS in their newest release doubled up a variable. @coroa will update this and post a new file

coroa commented 1 year ago

@merfort So, i finally made the new results comparable at http://rescue-task1.3.s3-website.eu-north-1.amazonaws.com/harmonization/harmonization-2023-10-03-facet-PkBudg_cp0400-OAE_off.html

You see old results from 2 months ago w/o the split next to the new ones harmonized separately per subsector and then aggregated to the full Energy Sector, for example at:

It looks like the split overall does not have a big effect on improving the harmonized values. And actually for VOC (see here) the huge amount of extraction emissions in CEDS is blowing up the small REMIND assumption well out of proportion and leads to a very unrealistic composite trajectory:

image

We would therefore suggest to go back to the aggregated Energy Sector variables and only continue breaking out the non-modelled 7A_Fossil-fuel-fires emissions

merfort commented 1 year ago

Ok, I see that in particular for VOC the split makes it worse, so I guess it really makes sense to revert it back so that only the aggregated "...|Energy Sector" variable is reported. Just one final thing: did you take a closer look at the VOC energy emissions in SSA as they appear in the CEDS dataset? Just because it is so much higher than what we report. I tracked our values back and found out that they originally come from ECLIPSE and there we really only have about 2 kt VOC/yr (0.002 Mt VOC/yr) for the regions "8 Western Africa" and "10 South Africa" (what we map to SSA).

merfort commented 1 year ago

@gidden On the other issue regarding the reallocation of OAE calcination emissions.

If I got that right, our idea was to add these emissions to the different end use sectors, because this will define where the synfuels produced from OAE-calcination-based CO2 are burned and thus the CO2 is released to the atmosphere. These sectors are the following 5:

CEDS+|9+ Sectors|Emissions|CO2|Industrial Sector
CEDS+|9+ Sectors|Emissions|CO2|Residential Commercial Other
CEDS+|9+ Sectors|Emissions|CO2|Aircraft
CEDS+|9+ Sectors|Emissions|CO2|International Shipping
CEDS+|9+ Sectors|Emissions|CO2|Transportation Sector

I now realized that this problem, that the ESMs actually need all types of emissions based on a tailpipe accounting, does not end with OAE calcination emissions. For example, the industry process cement calcination emissions that are captured and then used for synfuels have the same issue - so far we account them within the industry sector. Here the magnitude is much smaller, though, so keeping them as part of the industry sector is not that big of a problem. For all biofuel emissions we also have the same issue - we touched upon that issue a few times already, I think. In REMIND we account biofuels to be completely carbon neutral, so these emissions do not appear anywhere in our emissions balances, but for the ESMs it is important that the CO2 is taken up by the plants in one place and released to the atmosphere (by burning the fuel) in another. Same is true for DAC. To have the full information maximally transparently available, we would need to report everything on a tailpipe-basis with a "CO2-source" suffix, i.e. in the following way (exemplary for the Aircraft sector):

CEDS+|9+ Sectors|Emissions|CO2|Aircraft
= 
  CEDS+|9+ Sectors|Emissions|CO2|Aircraft|Fossil
+ CEDS+|9+ Sectors|Emissions|CO2|Aircraft|Cement Process
+ CEDS+|9+ Sectors|Emissions|CO2|Aircraft|OAE Calcination
+ CEDS+|9+ Sectors|Emissions|CO2|Aircraft|Biomass
+ CEDS+|9+ Sectors|Emissions|CO2|Aircraft|DAC

At the same time we'd have to add new variables tracking negative emissions from Biomass and DAC separately (i.e. tracking, where and how much CO2 is taken up by the plants).

The problem now is, that we - so far - do not track these emissions at all, as I said (carbon neutrality assumption). We are currently considering to rewrite our reporting such that we report all emissions also in a full tailpipe accounting. However, this will take time and will definitely not be available soon.

Thus, the only thing we could do is to limit the sector sub-split to

CEDS+|9+ Sectors|Emissions|CO2|Aircraft Sector
= 
  CEDS+|9+ Sectors|Emissions|CO2|Aircraft|Fossil
+ CEDS+|9+ Sectors|Emissions|CO2|Aircraft|Cement Process
+ CEDS+|9+ Sectors|Emissions|CO2|Aircraft|OAE Calcination

(admitting that I am not sure, how easy it is to reallocate Cement Process emissions, maybe we'd have to drop these as well).

Thus, we would anyways be inconsistent with the reporting and I would rather suggest to not do the split explicitly but only implicitly reallocate the OAE calcination emissions (and maybe cement process emissions) to the end-use sectors without having the explicit split between Fossil, Cement Process and OAE calcination. This way we would not unnecessarily bloat the reporting with incomplete information.

In order for the ESMs to still have the information on how much net CO2 limestone calcination emissions there are, we would keep the variable Emi|CO2|CDR|OAE|+|Calcination Emissions, noting that this must not be added to the total emissions, since the emissions would now part of the 5 end-use sectors.

What do you think about that?

gidden commented 1 year ago

Hey @merfort - I agree on being pragmatic with time and availabilities. We can also flag these as issues we want to tackle in the Tier 2 emissions datasets.

One question of clarification on my end:

Thus, we would anyways be inconsistent with the reporting and I would rather suggest to not do the split explicitly but only implicitly reallocate the OAE calcination emissions (and maybe cement process emissions) to the end-use sectors without having the explicit split between Fossil, Cement Process and OAE calcination. This way we would not unnecessarily bloat the reporting with incomplete information.

So in the end the OAE emissions would end up in *|Transport etc? I think that is fine, and is ultimately what we would use to do the downscaling and gridding.

If we're all on the same page, then I think that's decided!

merfort commented 1 year ago

@gidden Yes, exactly. Great. Then I will do it like that.

merfort commented 1 year ago

Hi all, I uploaded a new version of the emissions (REMIND-MAgPIE-CEDS-RESCUE-Tier1-2023-10-17.csv) in the google drive folder. In this version the OAE calcination emissions are reallocated to the 5 end-use subsectors. As explained above, just to have that information, these emissions are still reported in a separate variable (CEDS+|9+ Sectors|Emissions|CO2|OAE Calcination Emissions, I renamed it from named CEDS+|9+ Sectors|Emissions|CO2|CDR OAE Calcination Emissions since the term "CDR" is actually misleading). Please make sure to exclude this variable from the harmonization and downscaling process, since that would lead to a double accounting.

Furthermore, I reverted the split between Energy Sector|Extraction and Energy Sector|Supply emissions back so that there is now again just one Energy sector.

That should solve the last two remaing issues. Please let me know, if there is anythings else.

coroa commented 1 year ago

Hi all,

The new version of the emissions have finally made it through our harmonization and reporting pipeline. Find the results on AWS: http://rescue-task1.3.s3-website.eu-north-1.amazonaws.com/harmonization/2023-11-03/ . I kept the split into Modelled and Non-Modelled Energy Sector emissions.

I think that should conclude this exercise.