M1 Collection of pollutant-dependent emission temporal profiles and associated uncertainties

CO2MVS RESEARCH ON SUPPLEMENTARY OBSERVATIONS M1 Collection of pollutantdependent emission temporal profiles and associated uncertainties Due date of deliverable 31/12/2023 Submission date 11/12/2023 File Name CORSO-M1-V2 Work Package /Task WP1 / T1.2 Organisation Responsible of Deliverable Barcelona Supercomputing Center Author name(s) Marc Guevara Revision number V2.0 Status Issued to consortium Dissemination Level / location www.corso-project.eu The CORSO project (grant agreement No 101082194) is funded by the European Union. Views and opinions expressed are however those of the author(s) only and do not necessarily reflect those of the European Union or the Commission. Neither the European Union nor the granting authority can be held responsible for them. CORSO M1 2 Table of Contents 1 Introduction .................................................................................................................... 3 2 Methods ......................................................................................................................... 3 3 Results ........................................................................................................................... 5 4 Clarification on the pollutant-dependency ...................................................................... 7 5 References .................................................................................................................... 9 CORSO M1 3 1 Introduction Using emission inventories in modelling activities requires the original aggregated annual emissions to be broken down into fine temporal resolutions using emission temporal profiles. These profiles, which are typically constructed using long term averaged statistical data or meteorological parametrisations, contribute, together with the emission factors, activity factors and spatial proxies, to the overall uncertainty of high resolution spatiotemporal resolved emission inventories used for inverse modelling and data assimilation efforts. This document shortly describes the methods and results of quantifying the uncertainty associated to anthropogenic day-of-the-year (daily) emission temporal profiles. We focus on the daily scale as it is the targeted temporal scale of the future IFS-based CO2 Monitoring and Verification Support (CO2MVS) capacity to provide observation-based estimates of CO2 emissions. The work estimates the uncertainty associated to the temporal profiles of the main pollutant sources contributing to total CO2 anthropogenic emissions, namely: energy industry (power plants), road transport, residential/commercial combustion, aviation, shipping and manufacturing industry. The document is mainly for CORSO WP1 users, to provide guidance on the use of the temporal profiles and associated uncertainty to produce a 4D CO2 emission ensemble using the FFDAS system under T1.4. 2 Methods Considering that the input datasets and approaches used to estimate emission temporal profiles vary from sector to sector, uncertainties in the final profiles can arises from many factors, including: 1) the representativeness of the proxy selected to model the temporal distribution of the emissions for a specific sector (i.e., is the proxy selected the best one?), 2) uncertainties associated to the potential variability of the constructed profiles between years and countries 3) uncertainties associated to the input parameters considered in the meteorological parametrizations and 4) the quality of the proxy data considered to derive the profiles. Uncertainties from the chosen proxy to construct the temporal profiles, e.g. congestion statistic for the road transport sector, are assumed to result in systematic errors, which are more difficult to consider. Similarly, uncertainties derived from the quality of the proxy data are ignored, because we lack information on this. To quantify the uncertainty in the daily temporal profiles, an ensemble of profiles was created for each pollutant source based on proxies linked to their temporal emission variability, including electricity production statistics, traffic congestion statistics, AIS-based shipping emissions and air traffic statistics, among others. These profiles are from different years / countries / oceans and seas, so that the full range of possibilities is included. In the case of the residential/commercial combustion sector, the ensemble of profiles was created using the heating degree day (HDD) approach, which is an indicator used as a proxy variable to reflect the daily energy demand for heating a building (Quayle and Diaz, 1980). Uncertainty ranges were defined for the two key input parameters of the HDD approach, namely the critical temperature or temperature threshold (Tb), above which a building needs no heating (i.e., heating appliances will be switched off), and the non-heating fraction (f), which defines the share of residential/commercial combustion emissions that are not related to space heating but to other activities that remain constant throughout the year such as water heating or cooking . For these two parameters, uncertainty ranges were defined using the information provided by Ciais et al. (2022) and Eurostat et al. (2023), respectively: 𝑇𝑏=14.8 ± 1.8 ºC 𝑓 = 0.17 (+134%/−65%) The following table summarises the proxies considered for the construction of the ensembles for each pollutant sector. For the energy industry, road transport and residential/commercial CORSO M1 4 combustion sector the ensembles of profiles constructed are country-dependent, while for aviation, shipping and the manufacturing industry only one global ensemble is produced due to the lack of more detailed information. Table 1 Summary of the proxies considered for the constructions of daily temporal profile ensembles per sector Pollutant sector Proxy Spatial resolution Energy industry National electricity generation statistics (1) Country level Road transport Tomtom daily congestion statistics (2) Country level Residential/commercial combustion HDD approach + uncertainties associated to Tb and f Country level (3) Aviation EUROCONTROL daily flight statistics (4) Global Shipping CAMS-GLOB-SHIPv3.2 AISbased daily CO2 emissions (Jalkanen et al., 2016) (5) Global Manufacturing industry Industrial natural gas consumption statistics (Zhou et al., 2023) Global (1) ENTOS-E (https://transparency.entsoe.eu/content/static_content/Static%20content/knowledge%20base/SFTPTransparency_Docs.html#how-to-connect), NPP (https://npp.gov.in/dgrReports), CAMMESA (https://cammesaweb.cammesa.com/informe-sintesis-mensual/), COES (https://www.coes.org.pe/Portal/portalinformacion/generacion), ESKOM (https://www.eskom.co.za/dataportal/emissions/ael/), EPA - Air Markets Program Data (https://campd.epa.gov/data/bulk-data-files), AEMO (http://nemweb.com.au/Data_Archive/Wholesale_Electricity/MMSDM/2021/MMSDM_2021_02/MMSDM_Histor ical_Data_SQLLoader/), ONS (https://dados.ons.org.br/dataset/geracao_termica_despacho), OCCTO (https://occtonet3.occto.or.jp/public/dfw/RP11/OCCTO/SD/LOGIN_login#), PGCB (https://pgcb.gov.bd/site/page/0dd38e19-7c70-4582-95ba-078fccb609a8/-), AESO (https://www.aeso.ca/market/market-and-system-reporting/data-requests/historical-generation-data/), DEWA (https://www.dubaipulse.gov.ae/data/dewa-consumption/dewa_gross_power_generation_mwh-open#), NCSI (https://data.gov.om/OMELCT2016/electricity), TEIAS (https://www.teias.gov.tr/en-US) (2) https://www.tomtom.com/en_gb/traffic-index/ (last accessed: December 2023) (3) Original temporal weight factors were estimated at the grid cell level and then averaged at the country-level by considering the gridded EDGARv8 CO2 emissions reported for 2021. (4) https://ansperformance.eu/data/ (last accessed: December 2023) (5) Available on ECCAD (https://eccad.sedoo.fr/#/catalogue, last accessed: December 2023) CORSO M1 5 3 Results For each pollutant sector, the following CSV-files are provided: - CORSO_day_of_the_year_profiles_ensemble_<pollutant_sector>_v1.0.csv: Contains the temporal weight factors per day-of-the-year (represented as 2021 yearly Julian calendar) per ensemble member (ens_mem_xxx) and country (ISO3). The number of ensemble members varies per country and sector, as a function of the availability of proxy data. The ISO3 codes ROW and EU27 refer to Rest of the World and the European Union, respectively. - CORSO_day_of_the_year_profiles_uncertainty_<pollutant_sector>_v1.0.csv: Contains the mean value (FD_mean), standard deviation (FD_sd) and lower and upper limit of the 95% confidence interval (FD_q2.5 and FD_q97.5) of the temporal weight factors per day-of-the-year (represented as 2021 yearly Julian calendar) and country (ISO3). The ISO3 codes ROW and EU27 refer to Rest of the World and the European Union, respectively. Note that the average of the weight factors of each temporal profile in each ensemble is 1 for a full year, so that the temporally distributed emissions always add up to the annual total. The datasets describe the uncertainties in terms of a 95% confidence interval, with a lower and upper limit and the expected value. The distribution function of the estimated temporal weight factors is assumed to be normal following EMEP/EEA (2023), which indicates that this assumption can be done is the standard deviations is less than 30 % of the mean value. For the residential sector, several countries with substantial land area located in the tropics report a flat profile in all the members of the ensemble. These is because there is not a single day in which outdoor temperatures are below the temperature threshold values defined when estimating the HDD. Therefore, the associated uncertainty to these profiles is assumed to be 0 (emissions are constant across the year). For the energy, road transport and residential combustion sector, an additional CSV is provided (i.e. CORSO_day_of_the_year_profiles_proxies_missing_countries_<pollutant_sector> _v1.0.csv) indicating the information of which country should be considered for those countries with no data (e.g., for the road transport sector in China, it is recommended to use the temporal profiles and uncertainties estimated for Taiwan). The mapping between countries was done based on geographical proximity. Figure 1 shows an example of the daily temporal profiles and associated uncertainties computed for the energy, road transport and residential/commercial combustion sectors for Spain and Japan. The solid line represents the mean value of the temporal weight factors for each day of the year, while the pink shaded areas in each plot represent the confidence limits defined by the 2.5 percentile and 97.5 percentile points. Figure 2 shows the global daily temporal profiles and associated uncertainty computes for the shipping and aviation sectors. CORSO M1 6 Figure 1 Daily temporal profiles and associated uncertainty (95% CI) computed for the energy, road transport and residential/commercial combustion sectors for Spain and Japan. Figure 2 Global daily temporal profiles and associated uncertainty (95% CI) computed for the shipping and aviation sectors. CORSO M1 7 4 Clarification on the pollutant-dependency The constructed profiles and associated uncertainties presented in this Milestone are for CO2 emissions as well as for co-emitted species (i.e., CO and NOx). Therefore, the constructed profiles are countryand sector-dependent, but not pollutant-dependent, as this feature does not apply to the sectors/species we are covering. The reasons for that are as follows: 1. The temporal distribution of CO2, CO and NOx emissions from the combustion sectors we are including in the dataset (energy, aviation, residential, road transport, shipping and manufacturing industry) can be assumed to be very similar. An exception could be done for road transport. However, we are limited by how the EDGAR and CAMS-GLOB-ANT inventories report road transport emissions (see point 3 for more details). 2. To construct the profiles and associated uncertainties we are using proxies that give us information on the temporal variability of the activity of the sectors (e.g., number of flights, electricity generated, levels of congestion, natural gas consumed, changes in the outdoor temperature), but not of the individual emitted species. Exceptionally, for the shipping sector the profiles we constructed use as a basis the CAMS-GLOBSHIPv3.2 AIS-based CO2 emissions. For this specific sector, we could indeed construct pollutant-dependent profiles (i.e., using the CAMS-GLOB-SHIPv3.2 CO2, CO and NOx emissions). However, the profiles derived for CO2 and NOx were found to be almost identical (see Figure 3). Based on these results, we decided it was enough to stick to the assumption made in point 1. 3. If road transport emissions in current global emission inventories were reported per fuel type (i.e., diesel, gasoline, others), we could then consider the pollutantdependencies derived from: a) cold-start emissions for CO gasoline road transport emissions and b) the temperature penalty for NOx diesel road transport emissions, respectively. However, both EDGAR and CAMS-GLOB-ANT report road transport emissions under a unique category, and therefore these effects cannot be considered in the constructed profiles. Our plan is to further investigate this aspect under the framework of Milestone 4: "Fluctuations of emission ratios in urban plumes". In this task we will put the focus in different European urban areas, and there we are planning to use the CAMS-REG inventory, which actually splits road transport emissions by fuel type. Therefore, when analysing the fluctuations of emission ratios in EU urban plumes we will develop specific CO/NOx temporal profiles for gasoline/diesel road transport emissions, which will be different from the ones developed for CO2 road transport under the present Milestone 1. Pollutant-dependency can be significantly relevant for specific sectors (mostly process-based) and species, but not for the ones we are currently studying. An example of this is agricultural soil, where NH3 and CH4 emissions present very different temporal patterns, as the first one is mainly related to the use of fertilizers in crops and the second one is mainly driven by rice fields, which have a very specific calendar. CORSO M1 8 Figure 3 Example of daily CO2 and NOx temporal profiles derived from CAMS-GLOB-SHIPv3.2 2022 emissions 0.6 0.7 0.8 0.9 1 1.1 1.2 1 9 17 25 33 41 49 57 65 73 81 89 97 105 113 121 129 137 145 153 161 169 177 185 193 201 209 217 225 233 241 249 257 265 273 281 289 297 305 313 321 329 337 345 353 361 Shipping daily temporal profiles (2022) CO2_FD NOx_FD CORSO M1 9 5 References Ciais, P., Bréon, F. M., Dellaert, S., Wang, Y., Tanaka, K., Gurriaran, L., Francoise, Y., Davis, S. J., Hong, C., Penuelas, J., Janssens, I., Obersteiner, M., Deng, Z., and Liu, Z: Impact of lockdowns and winter temperatures on natural gas consumption in Europe, Earth's Future, 10, e2021EF002250, https://doi.org/10.1029/2021EF002250, 2022. EMEP/EEA, 2023. EMEP/EEA air pollutant emission inventory guidebook 2023. Uncertainties. Available at: https://www.eea.europa.eu/publications/emep-eea-guidebook-2023/part-ageneral-guidance-chapters/a-5-uncertainties-2023/view (last accessed December 2023) Eurostat, 2023. Energy consumption in households. Available at: https://ec.europa.eu/eurostat/statisticsexplained/index.php?title=Energy_consumption_in_households (last accessed December 2023) Jalkanen, J.-P., Johansson, L., and Kukkonen, J.: A comprehensive inventory of ship traffic exhaust emissions in the European sea areas in 2011, Atmos. Chem. Phys., 16, 71–84, https://doi.org/10.5194/acp-16-71-2016, 2016. Quayle, R.G., Diaz, H.F., 1980. Heating degree day data applied to residential heating energy consumption. J. Appl. Meteorol. 19(3): 241–246. Zhou, C., Zhu, B., Davis, S. J., Liu, Z., Halff, A., Arous, S. B., de Almeida Rodrigues, H., and Ciais, P.: Natural gas supply from Russia derived from daily pipeline flow data and potential solutions for filling a shortage of Russian supply in the European Union (EU), Earth Syst. Sci. Data, 15, 949–961, https://doi.org/10.5194/essd-15-949-2023, 2023.