Information can be found on the project pages: http://www.globalchange.umd.edu/ceds/ http://www.globalchange.umd.edu/ceds/ceds-cmip6-data/ https://github.com/JGCRI/CEDS *************************************************************************************************************************************** DATE: May 18, 2017 ---------------------------------------------------------------------------------------------------- SUMMARY This emissions dataset was produced by the Community Emissions Data System (CEDS) Project (http://www.globalchange.umd.edu/ceds/). *** IMPORTANT **** The May 18, 2017 data version corrects issues identified in the gridded versions of the previous data. This release supersedes previous data releases for all years. Emissions by country and sector are unchanged from the July 26, 2016 CEDS data release. An analysis of the differences in the gridded datasets can be found at google drive here: https://goo.gl/tGrc27 Note that the filename format for the May 2017 gridded data release was updated at the request of PCMDI for consistency with CMIP6 data naming conventions. As noted below, as a result of this naming update, some netCDF variables names have changed from the previous versions. Included in this release are SO2, NOx, CO, NMVOC, NH3, BC, OC, CO2, and CH4 emissions. (CH4 emissions provided from 1970 onward.) Please see the CMIP6 MIP protocol documents for use of these files. While we have supplied emissions over 1750-1850, the CMIP6 protocol calls for pre-industrial controls runs to use 1850 emissions. While we recommend using this data for CMIP6 simulations, note that the absolute magnitude of these changes are relatively small, particularly in 1850. We anticipate that these changes would not have a substantial impact on pre-industrial control simulations using the previous data versions. We do not, therefore, recommend re-running simulations if they are already underway. Gridded emissions are available at: https://esgf-node.llnl.gov/search/input4mips/ and should also be accessible through other ESGF nodes. All supplemental data is available at the project web site: http://www.globalchange.umd.edu/ceds/ (look for the "CEDS CMIP6 page" link) Known issues with the released CEDS data are listed at the end of this document. This data is documented in the following paper: Hoesly R M,Smith S J,Feng L ,Klimont Z ,Janssens-Maenhout G ,Pitkanen T T,Seibert J J,Vu L D,Andres R J,Bolt R M,Bond T C,Dawidowski L ,Kholod N ,Kurokawa JI ,Li M ,Liu L ,Lu Z ,Pinto De Moura M C,O'Rourke P ,Zhang Q 2017. "Historical (1750 - 2014) anthropogenic emissions of reactive gases and aerosols from the Community Emission Data System (CEDS)" Geoscientific Model Development Discussions http://www.geosci-model-dev-discuss.net/gmd-2017-43/ At the end of this document is a history of changes to this README document. ********************************************************* GRIDDED DATA The emissions from 1750-2014 are provided at monthly resolution, on a 0.5 degree grid, with 50-years per data file as per ESGF formatting conventions (with 1850 CMIP6 pre-industrial emissions in their own file). (Finer grids can be produced if needed.) Emissions are relatively smooth over the early portion of this time period, but annual data is supplied for consistency across the dataset. The files are in netcdf v4 (HDFv5) format with CF-compliant metadata. NOTE: the original June (1850) and July (1851-2014) data releases were formatted with 8 sectors as separate variables in each netCDF file. It was later discovered that this format is not compatible with the software that replicates data across to other (non-PCMDI) ESGF nodes. Subsequent data releases (August 2016 and May 2017) are in a new format with data in one variable, with sectors included as a dimension. The sectors in the netCDF files are: sector Description 0: AGR Non-combustion agricultural sector 1: ENE Energy transformation and extraction 2: IND Industrial combustion and processes 3: TRA Surface Transportation (Road, Rail, Other) 4: RCO Residential, commercial, and other 5: SLV Solvents 6: WST Waste disposal and handling 7: SHP International shipping (including VOCs from oil tanker loading/leakage) Within the netCDF files the sector ids are: sector:ids = "0: Agriculture; 1: Energy; 2: Industrial; 3: Transportation; 4: Residential, Commercial, Other; 5: Solvents production and application; 6: Waste; 7: International Shipping" ; Gridded aircraft emissions are also supplied in a separate file with 25 vertical layers (using the CMIP5 historical emissions Lamarque et al. 2010, as drawn from Lee et al., as a template). Note that aircraft emissions are zero in early years, but files are provided for all years consistency. The gridded emissions incorporate a monthly seasonal cycle by sector drawing largely from the ECLIPSE project. VOC speciation is provided at the 23 species resolution used in HTAP and RETRO. For use in setting aerosol size distribution and additional speciation (if desired), an auxiliary dataset emissions providing emissions from solid biomass combustion is also provided. Note that these are a subset of emissions in the main files. These data, therefore, should NOT be added to the emissions in the main files. (Note that no data files for CO2 emissions from solid biomass are released, as CO2 emissions from CEDS are from fossil fuels only.) Supplementary "checksum" .csv text files that provide total global mass for each sector, month, and year are also available at the project web site. Filenames are in the following format: EMISSION DATA FILE NAME FORMAT Bulk Gridded emissions: [em_species]-em-anthro_input4MIPs_emissions_CMIP_CEDS-2017-05-18_gn_YYYY01-ZZZZ12.nc (where YYYY is the starting year contained in this file, and ZZZZ is the ending year) netCDF variable name: [em_species]_em_anthro Gridded aircraft emissions: [em_species]-em-AIR-anthro_input4MIPs_emissions_CMIP_CEDS-2017-05-18_gn_YYYY01-ZZZZ12.nc netCDF variable name: [em_species]_em_AIR_anthro SUPPLEMENTAL EMISSION DATA FILE NAME FORMAT Gridded biomass emissions: [em_species]-em-SOLID-BIOFUEL-anthro_input4MIPs_emissions_CMIP_CEDS-2017-05-18-supplemental-data_gn_YYYY01-ZZZZ12.nc netCDF variable name: [em_species]_em_SOLID_BIOFUEL_anthro VOC speciation grids: VOC01-alcohols-em-speciated-VOC-anthro_input4MIPs_emissions_CMIP_CEDS-2017-05-18-supplemental-data_gn_YYYY01-ZZZZ12.nc netCDF variable name: VOC01_alcohols_em_speciated_VOC_anthro (There is a separate README file with further information on VOC speciation -- see project web site. Note there is a variable naming inconsistency for these file - described in the VOC README) **** NOTE some netCDF internal variable names have changed from previous versions **** At the request of PCMDI, netCDF internal variable names are now consistent with the variable_id in the metadata and variable_id in the filename. As a result, variable names changed compared to the previous CEDS releases for aircraft emissions, and for the speciated VOC and and solid biofuel supplemental emission files. (As in previous versions, dashes ("-") are replaced by underscores ("_") for the variable name to avoid errors with some software packages.) Emissions are provided in units of mass flux, as a monthly average flux, as noted in the netCDF files. Units are total mass of the indicated species as follows: SO2 : Mass flux of SOx, reported as SO2 NOx : Mass flux of NOx, reported as NO2 CO : Mass flux of CO NMVOC : Mass flux of NMVOC (total mass emitted) NH3 : Mass flux of NH3 BC : Mass flux of BC, reported as carbon mass (e.g., molecular weight 12) OC : Mass flux of OC, reported as carbon mass (e.g., molecular weight 12) CO2 : Mass flux of all oxidized carbon, reported as CO2 CH4 : Mass flux of CH4 For reference, the file: CEDS_gridcell_area_05.nc provides the grid cell areas used. (see project web site.) 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. In addition to gridded data, aggregate data by country and sector is also available in units of kilo-tonne (kt) per year (which have been published in the .zip journal paper supplement). Note that country totals in these summary files do not include international shipping or aircraft emissions, which are reported under the "global" iso. GRIDDING METHODOLOGY Emissions were first estimated at the level of country, sector, and fuel. Emissions by sector were then mapped to spatial grids by country and sector (several intermediate gridded sectors were combined to form the final release sectors for CMIP6). Grid cells that contain more than one country have portions of emissions from each country. For recent decades, emissions were mapped to the grid level largely using the distribution of emissions from EDGAR (usually using year-specific EDGAR grids from 1970 through 2008). For some sectors, the emissions distribution varies over time, while for other sectors it was held constant. For most emission species, residential combustion is the dominant source by 1850, so emissions from the RCO sector were distributed using HYDE population distributions by 1900 (with blended spatial distributions between 1900 and 1970). For other sectors the emissions distribution within each country was held fixed before 1970. We aim to improve these distributions in the future, and would welcome contributions. Because of the simplifying assumptions, emissions distributions, particularly in earlier years, should not be taken literally. Overall, however, anthropogenic emissions become small relative to either natural sources or mid-to-late 20th century emissions so we anticipate that these assumptions are not likely to have major impacts on global or regional modeling results. ********************************************************* CEDS COMPARED TO RCP EMISSIONS USED IN CMIP5 The following supplemental files: OC_Global_RCP_Comparison.pdf BC_Global_RCP_Comparison.pdf CO_Global_RCP_Comparison.pdf NH3_Global_RCP_Comparison.pdf NMVOC_Global_RCP_Comparison.pdf NOx_Global_RCP_Comparison.pdf SO2_Global_RCP_Comparison.pdf provide global CEDS emissions graphed against the historical emission data points used in the RCP/CMIP5 process (Lamarque et al. 2010) and are available at the project web site. Note that the CEDS emissions contain seasonality in all sectors, particularly the residential sector. This is an additional change from the RCP/CMIP5 emissions that is not apparent in the graphs showing annual emissions. ********************************************************* BRIEF NOTE ON 1850 EMISSIONS DATA VALUES For the emissions in this period, the assumptions for residential biomass are the dominant emissions for most species. (Exceptions are NH3 where manure emissions are dominant, and NOx, where industrial combustion (higher temperatures), residential, and manure are all important). There are some notable differences compared to the 1850 data used in CMIP5 from Lamarque et al. (2010): BC and OC: these are slightly different (BC slightly lower) due largely to updates in the SPEW (Bond et al.) emission factor database. CO: Is higher due to the use of a biomass emission factor appropriate for lower temperature, less efficient residential combustion. NOx: Is lower for the same reason, the use of a biomass emission factor that is lower than that used in the previous version which did not distinguish between coal and biomass combustion in this period. (These NOx and CO differences persist in the early years, but are not indicative of emissions differences for more recent decades.) Emissions prior to 1850 are estimated with an assumption of constant per-capita residential biomass consumption, fossil fuel emissions extrapolated using CDIAC CO2 emissions by fuel type, and emissions associated with metal smelting, coal coke and iron production extrapolated using metal production data by country. ********************************************************* GENERAL PROJECT AND DATA INFORMATION An overview of the CEDS project is available from the project web site, and a paper describing the CMIP6 dataset has been submitted to the CMIP6 GMD special issue (see top of this document). Briefly, our aim for the CMIP6 data release is emissions data that reflect a reasonable estimate of trends over time by country and sector. This emissions data is calibrated to country-level emission estimates wherever these are available, with default data used for sectors and countries where defaults are not available. For most species EDGAR 4.3 (from 1970 - 2014) is used as the default data, while for SO2, region and sector-specific assumptions updated from Smith et al. are combined with EDGAR 4.3 process emissions. In order to produce a dataset for release in time to be used for CMIP6, the initial focus of the project was to develop consistent trends over 1970 - 2014. Pre-industrial emissions were then finalized. In order to estimate 1850 emissions, residential and industrial coal and biomass consumption were estimated for the entire time period (drawing from CDIAC, Bond et al, and other sources). With these endpoints in place, emissions trends for the intervening years have been estimated. The result will be a continuous, consistent historical emission time series for CIMP6. Where relevant, emission factors will be converged to the selected early-industrial values. Emissions from fuel combustion are estimated using fossil and solid biofuel combustion statistics over all years. For years before 1971 (non-OECD) or 1960 (OECD), fuel combustion is estimated using either direct estimates or by trending fossil fuel combustion using CDIAC CO2 emissions. Emissions from non-combustion sectors are taken from inventory data in recent decades, and trended back before (generally) 1970 using an appropriate proxy data series. Some of these proxies are only available globally. While global and regional trends using this procedure appear to be reasonable, some anomalies are present at the country level. In the future we will refine these estimates by collecting additional country-level proxy data. As noted above, anthropogenic emissions start to become smaller relative to other sources by the beginning of the 20th century, so we anticipate that are not likely to have major impacts on global or regional modeling results. This dataset will continue to be refined after the CMIP6 data release and we welcome collaboration and engagement to improve historical emissions data. Note that there are some spurious emission results for early years at the sectoral level in the current data base due to the the sectoral resolution of the data available to extend emissions back in time. For example, aircraft emissions are present back to 1851, even though actual aircraft emissions did not begin until the early 20th century. There are also non-zero process emissions in early years in countries where this might not be realistic (and also the reverse). The magnitude of these emissions are small and will not materially impact model results. Please let us know if you identify any emissions results that should be examined. Country level emission data that are incorporated in this data release include: European Countries: Country data as reported to EMEP New Zealand, Belarus: UNFCCC reported data USA: US EPA Trends Canada: Environment Canada Argentina: Country submission to UNFCCC China: MEIC Inventory (2008, 2010, 2012) South Korea: http://airemiss.nier.go.kr/ Japan: REAS historical inventory (preliminary updated) Australia: National Emissions Inventory Taiwan: National Emissions Inventory Other Asia: REAS 2.1 historical inventory The methods and results for this data are documented in the following paper: http://www.geosci-model-dev-discuss.net/gmd-2017-43/ Please contact: ssmith@pnnl.gov for further details or questions about this dataset. To receive updates you can sign up for an information distribution list. (This list will be used only for CEDS-related information and e-mail addresses will not be shared or distributed.) To subscribe to the listserv, please send an email to listserv@listserv.umd.edu with the email body: “subscribe cedsinfo”. You can also go directly to the website for the listserv at: https://listserv.umd.edu/archives/cedsinfo.html. You do not have to be registered to view the archive on the website. ********************************************************* KNOWN ISSUES These issues will be corrected in a future data version (beyond CMIP6). **South Korea BC Emissions** From comparisons with literature and observations, it appears that BC emissions for South Korea are overestimated (see discussion in Hoesley et al. 2017). **Data break for several small countries** Combustion emissions becomes zero in earlier years for several countries that have inconsistent temporal coverage in the IEA energy data. These include: Sint Maarten, Suriname, Cambodia, Mongolia, Palau, Botswana, Namibia, and Niger. ********************************************************* DOCUMENT HISTORY 8-08-2017 Added link to analysis of differences in gridded data between release version 2017-05-18 and previous gridded data releases. (e.g. https://goo.gl/tGrc27 ) 5-18-2017 Updated data version number to reflect an additional correction to the data, release version 2017-05-18 5-17-2017 Updated to describe 2017-05-01 interim release of gridded data that fixes previously identified issues Note that filename format was updated at the request of PCMDI for consistency with CMIP6 data naming conventions. For reference, the naming conventions for PREVIOUS gridded data releases (all releases before May 2017) was: [em_species]-em-anthro_input4MIPs_emissions_CMIP_CEDS-v2016-06-18-sectorDim_gr_YYYY01-ZZZZ12.nc [em_species]-em-AIR-anthro_input4MIPs_emissions_CMIP_CEDS-v2016-06-18-sectorDim_gr_YYYY01-ZZZZ12.nc [em_species]-em-SOLID-BIOFUEL-anthro_input4MIPs_emissions_CMIP_CEDS-v2016-06-18-sectorDim-supplemental-data-18_gr_YYYY01-ZZZZ12.nc VOC01-alcohols-em-speciated-VOC_input4MIPs_emissions_CMIP_CEDS-v2016-06-18-sectorDim-supplemental-data-18_gr_YYYY01-ZZZZ12.nc **Seasonality** (Issue corrected as of 2017-05-01) The v2016-06-18 (pre-industrial) and v2016-07-26 (1851-2014) CMIP6 data releases contained a small error in monthly emissions distribution. This mostly shows up as larger emission fluxes in February as compared to other months (5-8% higher fluxes globally in February, with correspondingly lower emissions in other months). Annual totals by country and sector were not impacted, only the seasonal distribution. A correction for this error has been released on the project web site that can be applied to correct previous CEDS data releases. This is corrected in new data released as of May 2017. **Inconsistent Year-to-Year Gridding** (Issue corrected as of 2017-05-01) The v2016-06-18 (pre-industrial) and v2016-07-26 (1851-2014) CMIP6 data releases contain a number of year-to-year inconstancies in how emissions were mapped to a grid within some specific countries and sectors. This was caused by a logic error that caused population to be used to distribute emissions within a country rather than the sector-specific spatial distribution. (While there were some cases where this is desired, this occurred more often than it should have). This does not impact emissions by sector and country - e.g. the sum of emissions within each country and gridding sector are equal to the published CEDS emission values, however the spatial distribution can be in-consistent within a country from year-to-year. This is corrected in new data released as of May 2017. 4-13-2017 Updated to note two issues with the current emissions dataset. Added link to submitted discussion paper. 11-11-2016 Updated to clarify molecular weight of BC/OC emissions, and to note speciated VOC issue. 9-26-2016 Updated comparison between CEDS and RCP OC emissions. (RCP OC emissions were incorrectly displayed in previous graphs.) 8-24-2016 Update to note reformatting of gridded data to be compatible with ESGF software