Feature #635

CMIP6 emissions

Added by Twan van Noije about 5 years ago. Updated over 2 years ago.

Target version:
Start date:
Due date:
% Done:



Pre-industrial emissions for CMIP6, covering the period 1750 to 1850, have recently become available from

I have started writing code for reading the CMIP6 emissions into TM5 (so far only the CB05 version). Unfortunately, separate data sets with different conventions are provided for anthropogenic sources and biomass burning. In particular, the NMVOC speciation for the biomass burning emissions is different from other emission datasets used in TM5. For aviation, only the total NMVOC emissions are provided. Recommendations on how to distribute the NMVOC emissions from aviation over individual species are given by the data provider. Other anthropogenic emissions follow the ACCMIP/CMIP5/AR5 speciation.

The anthropogenic and biomass burning data sets are now treated as separate inventories in the code, which can be activated with LCMIP6 and LCMIP6BMB. This is still work in progress. To finish the code, I still need to include the VOC splits for the emissions from both biomass burning and aviation.

The code is in a local directory on $PERM, in a new project called cmip6_emissions. I haven't committed the code to the repository yet.

BC_emis_TM5_CMIP6.png View (45.1 KB) Twan van Noije, 10/28/2016 03:09 PM

BC_emis_TM5_CMIP5.png View (42.8 KB) Twan van Noije, 10/28/2016 03:09 PM

Emissions_CMIP6-CMIP5.pdf (39.1 KB) Twan van Noije, 10/28/2016 03:09 PM


#1 Updated by Twan van Noije almost 5 years ago

The implementation of CMIP6 emissions for the pre-industrial and historical periods is now complete. Tests have been performed for the year 2010 with TM5 driven by ERA-Interim. I have compared the results with a corresponding simulation using the existing emission routines and emissions from the CMIP5 RCP4.5 scenario.

The spatial patterns of the emissions applied in the model look correct. As an example the attached png files show the emission maps for Black Carbon (BC) for both simulations.

I have also checked and compared the annual totals applied in both model simulations. The numbers are in good agreement with the totals calculated from the emission files directly (see attached pdf).

Anthropogenic CH4 emissions have not yet been released, and are still taken from CMIP5. For the rest, the code is ready to be merged. It has been committed to the aero branch.

#2 Updated by Twan van Noije over 4 years ago

The code has been adapted to work with the latest version (1.1) of the CMIP6 biomass burning emissions. In the new version, the carbon emissions are on average about 5% lower.

#3 Updated by Twan van Noije over 4 years ago

A new version (1.2) of the biomass burning emissions was released on 13 December 2016. This is supposed to be the final version to be used for CMIP6. I have replaced the biomass burning emissions in ec:EMISSIONS/CMIP6 with the new version, and made a minor change in the TM5 code to make it work with the new version.

#4 Updated by Twan van Noije over 4 years ago

The treatment of the carbonaceous aerosol emissions has been changed in several respects in revision r564:

  • Different median particle sizes can be specified for the emissions due to fossil-fuel combustion from different CMIP6 anthropogenic sectors (currenly energy, industry, transportation, shipping , aircraft), for both the soluble and insoluble fractions.
  • The contributions from solid biofuel combustion to the CMIP6 sectors is now treated separately. This doesn't change the total emitted mass, but may change the distribution over soluble and insoluble modes and the number of emitted particles for the relevant CMIP6 sectors (in particular RCO, IND, ENE and TRA).
  • To account for non-resolved ageing of black carbon on subgrid scales, part of the emitted black carbon can be assumed to be soluble. Different mass fractions can be set for fossil fuel combustion, biofuel combustion and open biomass burning. As for organic matter, the soluble fraction of the black carbon from biofuel combustion and biomass burning is emitted in the accumulation mode.
  • Most of the parameter settings for particulate emissions have been moved from chem_param.F90 to emission_data.F90.

#5 Updated by Twan van Noije about 4 years ago

TM5 now reads the latest version of the CEDS anthropogenic emissions of ozone and aerosol precursors (release date 18 May 2017). Differences in global emission amounts compared to the previous release are very small, except for VOC species. I have verified that the differences in emission totals applied in TM5 are within numerical precision the same as the differences calculated directly from the CEDS data sets. The model now also uses methane emissions from CEDS, which so far have only been provided for the period 1970-2014.

#6 Updated by Twan van Noije almost 4 years ago

CEDS anthropogenic methane emissions for the period before 1970 have recently become available at 10-year intervals, starting in 1850. So the years provided are 1850, 1860, ..., 1960, 1970, and yearly from then on.

The new data have been included in the model (Revision r640). At intermediate years, a linear interpolation between the monthly values at the start end end of the decade is applied. For the period prior to 1850, emissions of 1850 are applied. This is specific to the methane emissions from CEDS; CMIP6 emissions of other species and CMIP6 methane emissions from biomass burning are provided on a yearly basis back to 1750.

An update of the CEDS data set of aircraft emissions has just been announced, and will be included as soon as the new files are available.

The scenario data sets should also become available soon.

#7 Updated by Twan van Noije over 2 years ago

  • Status changed from In Progress to Resolved

I have included the future scenario emissions for anthropogenic sectors and open biomass burning for SSP3-7.0 and SSP3-7.0-LowNTCF, which will be run for CMIP6 as part of ScenarioMIP and AerChemMIP.

For black carbon and organic aerosol emissions, we apply a different treatment for the emissions from solid biofuel burning. As these were not provided for the scenarios, following the recommendation by the data providers, I have assumed that the fraction of BC/OC solid biofuel emissions compared to the total BC/OC anthropogenic emissions for each sector and month is the same as during the last 10 years of the historical period (2005-2014). In cases where there are emissions from a particular sector in the future, but not during 2005-2015, the fraction cannot be calculated and the emissions from solid biofuel buring are set to zero. In this way, I have created the missing input files (using cdo).

The input files containing the future emissions data sets are provided in ec:/nks/EMISSIONS/CMIP6/.

#8 Updated by Twan van Noije over 2 years ago

Simulations for the year 2015 and 2016 run without problems, and produce similar optical depth, methane and ozone mixing ratio fields as for 2014.

The code changes for reading the scenario emissions and boundary conditions are committed in revisions 944-957.

There is one very minor issue that I am not sure about yet, and that is whether or not the biomass burning emissions in February need to be scaled by 28/29 in leap years. I have asked this to the data providers, and will adapt the code once I have their answer.

#9 Updated by Twan van Noije over 2 years ago

Following a mail conversation with Steve Smith (copied below), I have removed the leap-year correction for the anthropogenic emissions. The biomass burning emissions also shouldn't be corrected.

Hi Twan,

I guess I wasn’t thinking of leap years.

What I had been thinking was, if a model, for example, uses 12 30-day months per year (evidently there might be some that do this), then I was thinking one would might want to >adjust fluxes so that totals match. (But then again, if you’re normal year isn’t 365 days, then perhaps there’s a different way to think about that?).


On Dec 21, 2018, at 3:55 PM, Noije van, Twan (KNMI) <> wrote:

Hi Steve,

I see. I presume this also applies for the anthropogenic emissions, right?

However, in that case I must have misunderstood your comment in the CEDS readme file (

"Note that total monthly emissions were converted to fluxes using a 365 day calendar. If your model uses a different calendar, you should ideally adjust fluxes such that the integrated annual fluxes are maintained."

Should I ignore this comment?

Thanks for explaining,


From: Smith, Steven J (PNNL-JGCRI) <>
Sent: Friday, December 21, 2018 5:53 PM
To: Noije van, Twan (KNMI); Werf, G.R. van der
Subject: RE: CMIP6 biomass burning emissions; leap-year correction

Hi Twan,

The calendar used in the data files is indicated in the metadata under the time dimention.

So, for example, in the files we processed the calendar is:
:calendar = "365_day";

(so 365 days with no leap year)

Since these are fluxes, which are per unit time and spatial dimensions, we have already corrected for this in the data. So you don't need to change fluxes in leap years. You can use the same flux value for the leap day. There would just be one more day with the same flux as in the rest of that month.



From: Noije van, Twan (KNMI) []
Sent: Friday, December 21, 2018 5:08 AM
To: Werf, G.R. van der; Smith, Steven J (PNNL-JGCRI)
Subject: CMIP6 biomass burning emissions; leap-year correction

Dear Guido, Steve,

Just a minor question about the open biomass burning emissions you provided to CMIP6. It is about the treatment of the February emissions in leap years.

For the anthropogenic emissions, I understand that the provided emission fluxes (kg/m2/s) were calculated from the monthly totals using a non-leap year calendar. Therefore it is recommended to reduce the February fluxes by 28/29 in leap years, when running a model with a standard calendar, in order to produce the exact same total emissions.

As the future open burning emissions are provided in a similar format as the anthropogenic emissions, I assume the same holds for those emissions. Is that correct?

The question then also arises whether I should also apply that correction factor to biomass burning emissions in the historical period.

I know it's a minor issue, but still I prefer to do it correctly.

Thank you very much for letting me know.

Best wishes,


#10 Updated by Philippe Le Sager over 2 years ago

  • Status changed from Resolved to Closed
  • % Done changed from 0 to 100

Twan van Noije wrote:

The code changes for reading the scenario emissions and boundary conditions are committed in revisions 944-957.

Merged into the trunk in r964.

#11 Updated by Twan van Noije over 2 years ago

I have made some runs to test the future emissions applied in the model. Tests were made with the offline TM5 model for 2014 (using historical emissions) and for 2015 and 2016 using SSP3-7.0 (v1.1). Global emission totals as analyzed by the TM5 budget routine in three runs are shown in the table below.

Species (unit) 2014 2015 2016
NOx (Tg N) 60.4997 60.3604 61.1454
CO (Tg) 1073.0868 1044.1069 1053.4314
SO2 (Tg S) 60.3432 53.8773 53.8167
NH3 (Tg) 75.5520 75.7979 76.6568
ISOP (Tg) 572.3237 572.1790 573.8042
TERP (Tg) 95.8234 95.6277 95.8527
NMVOC (Tg) 132.2576 131.2626 132.6771
SO4 (TgS) 1.5473 1.3815 1.3799
POM (Tg) 57.8877 55.6870 56.1629
BC (Tg) 9.7554 9.7423 9.8686

For all species except SO2 and SO4, the changes in the total emission amounts are small. For SOx=SO2+SO4, the emissions (Tg S) drop by 6.63 (11.8%), from 61.89 in 2014 to 55.26 in 2015.

This is consistent with the emission totals calculated from the provided SO2 emission files, where the anthropogenic emissions (Tg S, excl. aircraft) drop by 6.60, from 55.84 in 2014 to 49.24 in 2015. Of course, biomass burning emissions and aircraft emissions also change, but these are much smaller.

The tests indicate that the code for reading the future emission data sets, and for interpolating them to intermediate years works as expected.

#12 Updated by Twan van Noije over 2 years ago

The drop in SO2 emissions is consistent with the data presented in Figure 7 of Gidden et al. (2018; The fact that the discontinuity is largest for SO2 is likely related to the fact that in the harmonization calculations data up to 2017 were included for sulfur emissions in China, "because of the drastic reduction in these emissions in the most recently available datasets" (see footnote on page 7).

Also available in: Atom PDF