Background & Summary

The European Union and the United Kingdom is the world’s third energy consumer and CO2 emitter, accounting for 12% of global emissions1,2,3. The EU27 announced the Green Deal in December 2019, and set up a road map for cutting greenhouse gas emissions by at least 55% by 2030 and reaching carbon neutrality by 20504. With increasing focus and effort on reducing CO2 emissions, there is a growing need for more reliable data as well as for data with a lower latency. Most CO2 emission inventories including national inventories reported to the United Nations Framework Convention on Climate Change (UNFCCC) lag reality by one years or more2,5,6,7,8,9,10.

Lower latency estimates of CO2 emissions are generally delayed by several months as data must be gathered from numerous sources and then verified. For example, Eurostat has been producing early estimates of annual and country-level CO2 emissions in the EU since at least 2012, with a delay for about 5 months11. The Global Carbon Project publishes projections of the current year’s global fossil CO2 emissions since 2012 by the end of the current year12. Few EU countries and the UK publish quarterly or monthly estimates of emissions from preliminary energy data13,14,15,16,17. As the COVID-19 pandemic profoundly disrupted human activities in the year 2020, existing inventories were not able to monitor changes in activity and assess the COVID-19 impacts on CO2 emissions during that period and the following recovery. Le Quéré et al.18 and Forster et al.19 developed methods for estimating daily, global CO2 emissions, based on confinement index and Google mobility indexes respectively18,19. These two approaches captured first order pandemic related reductions but were not tested for subsequent variations when lockdowns were finished, and did not continue to provide near-real-time, daily and country-level estimates emissions. Further, the private company Kayrros released a daily estimate of emissions from regulated sectors (European Trading Scheme) in November 2021 using site-level satellite monitoring of industrial activity, called Carbon Watch20. The Carbon Monitor international research initiative developed a new near-real-time daily dataset of CO2 emissions with global coverage and country-level estimates for 12 countries which lags reality by only one month1,3,21,22,23,24,25,26. Building on this foundation, they have also developed the Carbon Monitor China27, which provides provincial-level near-real-time daily emissions data for China, and the Carbon Monitor United States28, which provides state-level daily emissions data for the US. Additionally, they have launched the Carbon Monitor Cities29,30, covering near-real-time daily emissions data for over 1500 cities worldwide, the Carbon Monitor Power31, which provides near-real-time hourly-to-daily data on power generation, and the Carbon Monitor GRACED32,33, a dataset of global near-real-time grid-level CO2 emissions.

Here, we present a near-real-time, sector-specific, country-level estimates of daily fossil fuel and cement CO2 emissions for 27 European Union countries and the United Kingdom based on an extension of Carbon Monitor called Carbon Monitor Europe (CM-EU). We present the methodology and the results for changes in emissions between January 1, 2019 and December 31, 2021. Details of our data sources and approach are in the Methods section. An evaluation of CM-EU against national quarterly emissions estimates and Carbon Watch is provided in the technical validation section. The data Carbon Monitor Europe are publicly available at Figshare34 and our website https://eu.carbonmonitor.org/.

Methods

Carbon Monitor Europe (CM-EU) is a new regional improved version of the global Carbon Monitor system, providing near-real-time daily estimates of CO2 emissions for six sectors over the globe with separate estimates for 12 countries or groups of countries1,21. CM-EU presents country-level estimates of daily CO2 emissions from January 2019 through December 2021 for 27 countries of European Union (EU27) and the United Kingdom (UK). Following the method presented in the global Carbon Monitor publications1,21, we used annual CO2 emissions of EU27 & UK from the Emissions Database for Global Atmospheric Research (EDGAR)8 as the baseline data for emissions in the year 2019, then disaggregated and extrapolated in time this estimate into daily scale for 2020 and 2021 based on time variations of activity data. Below we describe the calculation process in detail.

Annual country-level and sectoral CO2 emissions in the baseline year 2019

Annual fossil fuel combustion and cement production CO2 emissions by sector in 2019 for all European Union countries and United Kingdom are directly obtained from the Fossil CO2 emissions of all world countries - 2020 Report8 released by the Emissions Database for Global Atmospheric Research (EDGAR). Emissions of EDGAR are derived using the IPCC Tier 1 approach according to the 2006 IPCC Guidelines35, with 35 sectors based on IPCC categories. We aggregated 11 energy-related and cement production sectors of EDGAR into six main sectors, including power, industry, ground transportation, residential (public, commercial buildings and households), domestic and international aviation. In this dataset, emissions from international flights are assigned to the country of departure, which is different from the accounting used by UNFCCC from IPCC guidelines where international flights are assigned to a specific bunker fuel category. Table 1 shows the correspondence table between aggregated sectors of CM-EU and EDGAR sectors based on IPCC (first column).

Table 1 The mapping relationships between sectors in this study and EDGAR sectors based on IPCC categories.
Full size table

Data acquisition and processing of daily CO2 emissions in 2019, 2020 and 2021

Carbon Monitor follows the IPCC guidelines for emissions reporting35 in computing CO2 emissions from a country/region by multiplying activity data (AD) by corresponding emissions factors (EF):

$$Emis=\sum \sum \sum A{D}_{i,j,k}\times E{F}_{i,j,k}$$

where i, j, k denote regions, sectors, and fuel types respectively. We assume that emission factors and structure of each sector remain unchanged for each country in 2020 and 2021 compared with 2019. Thus, the rate of change of the emission is calculated based solely on the change of the activity data in 2020 and 2021 compared to the same period of 2019. The emissions were calculated following this equation separately for the power sector, the industrial sector, the ground transportation sector, the aviation sector (including the domestic and international aviation sector) and the residential sector.

Power sector

For the power sector, daily emissions were calculated from hourly electricity generation data by production types at resolution of 1 h to 15 min. Data of 22 EU countries (Austria, Belgium, Bulgaria, Croatia, Cyprus, Czech Republic, Denmark, Estonia, Finland, France, Germany, Greece, Hungary, Ireland, Italy, Latvia, Netherlands, Poland, Portugal, Slovakia, Slovenia and Spain) and United Kingdom are available from the ENTSO-E Transparency platform (https://transparency.entsoe.eu/dashboard/show) and Balancing Mechanism Reporting Service (BMRS) (https://www.bmreports.com/) respectively (Table 2). We removed outliers and filled the N/A values by using the linear interpolation function in Python Pandas packages. The electricity generation data used in this study from ENTSO-E and BMRS are include several fuel types with coal, gas, oil and peat. The average yearly emission factor of the whole power sector is assumed to remain constant at its 2019 value, so that daily emissions are estimated by:

$$Emi{s}_{power,daily,2019}=Emi{s}_{power,yearly,2019}\times \frac{A{D}_{power,daily,2019}}{A{D}_{power,yearly,2019}}$$
$$Emi{s}_{power,daily,2020or2021}=Emi{s}_{power,daily,2019}\times \frac{A{D}_{power,daily,2020or2021}}{A{D}_{power,daily,2019}}$$
Table 2 Sources of activity data for different sectors.
Full size table

For Lithuania, Luxembourg, Malta, Romania and Sweden (other EU) not covered by ENSTO-E, we assume a linear relationship between their daily emissions and the total daily emissions of the 22 EU countries and United Kingdom (EU22 + UK) and their daily emissions are estimated as:

$$Emi{s}_{power,daily,2019,EU5}=Emi{s}_{power,yearly,2019,otherEU}\times \frac{Emi{s}_{power,daily,2019,EU22+UK}}{Emi{s}_{power,yearly,2019,EU22+UK}}$$
$$Emi{s}_{power,daily,2020or2021,otherEU}=Emi{s}_{power,daily,2019,EU5}\times \frac{Emi{s}_{power,daily,2020or2021,EU22+UK}}{Emi{s}_{power,daily,2019,EU22+UK}}$$

Industry sector

Daily emissions from industry are estimated using the monthly industrial production index (IPI) from several datasets and daily power generation data (Table 2). As daily production data are not available for industrial and cement production, the monthly CO2 emissions are estimated by using monthly statistics of industrial production and daily data of electricity generation to disaggregate the monthly CO2 emissions into a daily scale. This approach is based on two assumptions: 1. A linear relationship is assumed between industrial production index and emissions from industrial and cement production. 2. A linear relationship is assumed between daily industry activity and daily electricity production, from ENSTO-E and our approach for the five Other-EU countries. Therefore, the monthly and daily industry emissions are estimated following:

$$Emi{s}_{industry,monthly,2019}=Emi{s}_{industry,yearly,2019}\times \frac{A{D}_{industry,monthly,2019}}{A{D}_{industry,yearly,2019}}$$
$$Emi{s}_{industry,monthly,2020or2021}=Emi{s}_{industry,monthly,2019}\times \frac{A{D}_{industry,monthly,2020or2021}}{A{D}_{industry,monthly,2019}}$$
$$Emi{s}_{industry,daily,2019,2020or2021}=Emi{s}_{industry,monthly,2019,2020or2021}\times \frac{A{D}_{power,daily,2019,2020or2021}}{A{D}_{power,monthly,2019,2020or2021}}$$

Ground transportation sector

Carbon Monitor uses TomTom congestion global level data (Table 2) from the TomTom website (https://www.tomtom.com/en_gb/traffic-index/) to capture the daily variations in the ground transportation activity. The TomTom traffic congestion level (called G hereafter) represents the extra time spent on a trip in congested conditions, as a percentage, compared to uncongested conditions. TomTom congestion level data cover 203 cities across 24 EU countries (Austria, Belgium, Bulgaria, Czech Republic, Denmark, Estonia, Finland, France, Germany, Greece, Hungary, Ireland, Italy, Latvia, Lithuania, Luxembourg, Netherlands, Poland, Portugal, Romania, Slovakia, Slovenia, Spain, Sweden) and the United Kingdom at a temporal resolution of one hour (Table 3). Note that a zero-congestion level means the traffic is fluid or ‘normal’ but does not mean there are no vehicles and zero emissions. The lower threshold of emissions when the congestion level is zero was estimated using real-time data from an average of 70 main roads in the city of Paris. The daily mean car counts (called Q hereafter) were calculated by a sigmoid function based on regression:

$$Q=a+\frac{b{G}^{c}}{{d}^{c}+{G}^{c}}$$

Where a, b, c, d are regression parameters. Then we allocated the country-level emissions from EDGAR into a daily scale using:

$$Emi{s}_{groundtransport,d}=Emi{s}_{groundtransport,yearly}\times \frac{{Q}_{d}}{{\sum }_{d=1}^{n}{Q}_{d}}$$

where Qd is the mean vehicle number per hour in day d, Emisground transport,d is the ground transport emission in day d, Emisground transport,yearly is the annual road transportation emissions from EDGAR, n is the number of days in a year. For countries not covered by TomTom (Croatia, Cyprus, Malta), we assume that their emission changes follow the patterns of total daily emissions from the average other 24 EU countries and the United Kingdom.

Table 3 Cities (203 across 24 EU countries and UK) where TomTom congestion level data are available.
Full size table

Aviation sector

Emissions of the aviation sector separated into domestic and international aviation, are estimated by individual commercial flights data (Table 2) from the Flightradar24 database (https://www.flightradar24.com). We compute CO2 emissions by assuming a constant emissions factor of aviation in CO2 emissions per km flown, called EFaviation across the whole fleet of aircraft (regional, narrowbody passenger, widebody passenger and freight operations) as the share of flight types has not significantly changed since 2019. The aviation sector separates domestic flights and international flights departing from countries considered in this study. The flights within EU but different countries are considered as international. The daily emissions of the aviation sector are estimated as:

$$Emi{s}_{aviation,daily}=DKF\times E{F}_{aviation}$$

where DKF is the distance flown that is computed using great circle distance between the take-off, cruising, descent and landing points for each flight and are cumulated over all flights. For countries with overseas territories (mainly France, UK, Denmark) aviation emissions of flights to/from overseas territories are counted in this study as international emissions, whereas they would be reported as domestic emissions by national estimates.

Residential sector

We use the fluctuation of air temperature (Table 2) to simulate daily variations in the energy consumption and direct emissions of residential and commercial buildings. The calculation of residential emissions was performed in three steps: (1) Calculation of population-weighted heating degree days each day for each country of EU27 & UK and for each day based on the ERA5 reanalysis of 2-m air temperature36,37, (2) Using annual residential emissions in 2019 from EDGAR as the baseline. For each country, the residential emissions were split into two parts, i.e., cooking emissions and heating emissions, according to the EDGAR guidelines. The emissions from cooking were assumed to remain stable, while the emissions from heating were assumed to depend on and vary by the heating demand. (3) Based on the change of population-weighted heating degree days in each country, we scaled the EDGAR 2019 residential emissions to 2020 and 2021. Since the index of heating degree days are daily values, we get daily emission updates for the residential sources. The EDGAR residential emissions are downscaled to daily values based on daily variations in population-weighted heating degree days as follows:

$$Emi{s}_{residential,monthly}=Emi{s}_{residential,yearly,2019}\times \frac{{\sum }_{m}HD{D}_{daily}}{{\sum }_{m,2019}HD{D}_{daily,2019}}$$
$$\begin{array}{l}Emi{s}_{residential,daily}=Emi{s}_{residential,monthly}\times Rati{o}_{heating,monthly}\times \frac{HD{D}_{daily}}{{\sum }_{m}HD{D}_{daily}}\\ +Emi{s}_{residential,monthly}\times \left(1-Rati{o}_{heating,monthly}\right)\times \frac{1}{{N}_{m}}\end{array}$$
$$HD{D}_{daily}=\frac{\sum \left(Po{p}_{grid}\times \left({T}_{grid,daily}-18\right)\right)}{\sum Po{p}_{grid}}$$

Where m is the month, HDDdaily the population-weighted heating degree day, Ratioheating,monthly the percentage of residential emissions from heating demand monthly, Nm the number of days in month m, Popgrid the gridded population data derived from Gridded Population of the World, Version 436, Tgrid,daily is the daily average air temperature at 2 meters derived from ERA537, and 18 is a HDD reference temperature of 18 °C following ref. 38.

Code description

Python code for data generation is provided (link in the Code Availability section). The codes in CM_EU_v2.py produce daily emissions for 27 EU countries and the United Kingdom in the dataset.

Data Records

The CM-EU dataset is an CSV file containing country-level emission of 27 European Union countries and the United Kingdom from 01/01/2019 to 31/12/2021 for six sectors: power, industry, ground transportation, domestic and international aviation. The dataset comprises five columns: country, date, sector, value, and timestamp. The value column denotes the daily CO2 emissions in Mt CO2. The timestamp is included to help servers recognize time. At the time of writing this article, this dataset has been updated to December 31, 2021, and the full dataset can be downloaded at Figshare34. Latest updates and related information are available for view and download on our website https://eu.carbonmonitor.org/.

The power sector emissions for Lithuania, Luxembourg, Malta, Romania, and Sweden, in addition to ground transport emissions for Croatia, Cyprus, and Malta, have been estimated by assuming a linear relationship with the daily emissions totals from the remaining countries in their respective sectors. This estimation method was employed due to the lack of coverage or incorrect data provided by the activity data.

Figure 1 shows daily CO2 emissions for the years 2019, 2020, and 2021 in the EU27 & UK, as well as for specific countries including Germany, the United Kingdom, Italy, Poland, France, Spain, the Netherlands, and other European countries. Figure 2 represents the daily CO2 emissions for six sectors within the EU27 & UK from 2019 to 2021. Figure 3 shows the changes in emissions within the EU27 & UK, from 2019 to 2020 and 2021, across regions and sectors. It highlights the contributions made by major countries and sectors during the study period. Figure 4 shows changes in annual CO2 emissions at country level, making comparisons between 2020 and 2019, as well as between 2021 and 2020, and 2021 and 2019 with other published datasets.

Fig. 1
figure 1

7-days smoothed (running mean) daily CO2 emissions for EU27 & UK, Germany, the United Kingdom, Italy, Poland, France, Spain, Netherlands and the rest of European countries in 2019, 2020 and 2021. Grey shaded areas indicate the changes between 2019 and 2020. The percentage and numbers in the top of each panel reflect the relative and absolute change in 2021 compared with 2019. Bars at the bottom show the sectoral shares of annual emissions in 2021. (Aviation includes the domestic and international aviation).

Full size image
Fig. 2
figure 2

7-days smoothed (running mean) daily CO2 emissions for EU27 & UK for six sectors in 2019, 2020 and 2021. Grey shaded areas indicate the changes between 2019 and 2020. The percentage and number in the top of each panel reflect the relative and absolute change in 2021 compared with 2019.

Full size image
Fig. 3
figure 3

Changes in EU27 & UK emissions from 2019 to 2020 and 2021 across regions and sectors (a). Contribution by major countries/emitters and sectors in 2019, 2020 and 2021 (b).

Full size image
Fig. 4
figure 4

Country-level changes in annual CO2 emissions in 2020 compared with 2019 (left), in 2021 compared with 2020 (middle), in 2021 compared with 2019 (right).

Full size image

Technical Validation

Comparison with national emissions estimates, yearly and sub-yearly in selected countries, and with other near-real-time estimates

To evaluate the CO2 emissions data from CM-EU, we compared our results with three different existing emissions estimates of various timescale (Table 4). The first evaluation data are annual country-level CO2 emissions of EU27 & UK in 2019 and 2020 from EDGAR8, BP7, IEA39, GCP2 and Eurostat40. The second evaluation data are preliminary emissions publicly disclosed on a monthly or quarterly scale with a latency of a few months to one year by national statistics offices/inventory agencies of Netherlands, Sweden, UK, France and Germany, which are from Centraal Bureau van Statistiek16, Statistics Sweden17, gov.uk15, Citepa14 and Umwelt Bundesamt13 respectively. The third evaluation data include four data sources: annual CO2 emissions reported to the UNFCCC6, near-real-time daily and monthly CO2 estimates of EU from CICERO (updated from ref. 41) and Centre of Research on Energy and Clean Air (CREA)42, and daily country-level CO2 emissions of EU countries and UK from 2019 to 2021 from the Carbon Watch data of Kayrros20 focusing on regulated sectors (ETS).

Table 4 List of CO2 emissions datasets used for technical validation of CM-EU products.
Full size table

Figure 5 shows the comparison of annual CO2 emissions in 2019 and 2020 and changes between the two years for EU27 & UK, Germany, UK, Italy, Poland, France, Spain, Netherlands and the rest of EU between CM-EU, EDGAR, BP, IEA, GCP and Eurostat. In terms of annual total amount, the results of the six databases are relatively similar, and the main difference comes from the scope difference of each dataset. As for changes between 2019 and 2020, our estimates are lying in the middle range, close to most other datasets. Figure 6 shows the comparison of annual or quarterly CO2 emissions in 2019 and changes between 2020 and 2019 or 2021 and 2020 for Netherlands, UK, Sweden and Germany between this study and estimates from Citepa (France), Centraal Bureau van Statistiek (Netherlands), gov.uk (UK), Statistics Sweden (Sweden) and Umwelt Bundesamt (Germany). It illustrates that our estimates are close to the official estimates in terms of annual or quarterly totals for 2019 and for changes between 2019 and 2020 (at the beginning of COVID-19 pandemic), with relative differences from 0.07% to 3.17% at annual scale and 0.2% to 5.63% at quarterly scale.

Fig. 5
figure 5

Comparison of annual CO2 emissions in 2019 and 2020 and the relative changes between the two years among six datasets: CM-EU, EDGAR, BP, IEA, GCP and Eurostat for EU27 & UK, Germany, UK, Italy, Poland, France, Spain, Netherlands and the rest of EU.

Full size image
Fig. 6
figure 6

Comparative analysis of quarterly (a) and annual (b) CO2 emissions between CM-EU (darker color) and preliminary national CO2 emissions estimates (light color) for France, Netherlands, UK, Sweden and Germany (2019–2021).

Full size image

Our estimates differ more (in terms of relative differences) from these official estimates for quarterly and annual changes between 2020 and 2021 than between 2019 and 2020, with relative differences from 0.81% to 3.34% at annual scale and 0.19% to 9.94% at quarterly scale.

In Fig. 7, we compare CO2 emissions across 27 EU countries from 2019 to 2021, utilizing yearly, monthly, and daily time-steps with other estimates with CREA and CICERO. We omitted the CM-EU emissions from the international aviation sector and UK emissions due to the absence of extra-territorial and UK emissions data in CREA and CICERO. Although CM-EU and CREA offer daily estimates, we averaged them into monthly steps for an easier visual comparison. For monthly data, the emission estimates from all three sources are relatively similar, with CREA reporting a slightly higher emission level. We conducted a moving window correlation analysis of these two sets of data with CM-EU and averaged the results. As can be seen from the shaded portion of the line graph, there is substantial agreement between CM-EU and the other two data sets for the majority of the time, with the exception of July 2021, where a discrepancy was observed. For daily data, we employed a seven-day moving average to CM-EU and CREA emissions. Judging by the trend and shaded portion of the moving window correlation, the two sets of data exhibit a significant level of consistency, with R2 values greater than 0.6 occurring approximately 46% of the time. However, the CREA data for 2019 and 2020 are not truly daily, resulting seesaw variability pattern, explaining the difference with CM-EU during these years. Beginning in March 2021, CREA started to provide real daily data, which led to a very high consistency with CM-EU throughout the remainder of the year.

Fig. 7
figure 7

Annual, monthly and daily CO2 emissions in 27 EU countries from 2019 to 2021: comparing official reporting to the UNFCCC (dark blue), CM-EU (purple), CREA (green), and CICERO (orange) for annual and monthly data, and CM-EU and CREA for daily data using a seven-day running mean. The shaded area in the line chart represents the correlation coefficient using moving window correlation analysis. Lower daily correlations with CREA before March 2021 are because CREA data are not real daily data in this period (see text).

Full size image

We also compared the monthly CO2 emissions of EU27 & UK and France for four or five sectors (power, industry, ground transport, domestic aviation and residential) from 2019 to 2021, with the Kayrros Carbon Watch data and the official Citepa in Fig. 8.

Fig. 8
figure 8

Technical validation of short-term emissions changes with other available ‘high frequency’ data. (a) Comparison of monthly CO2 emissions for four sectors (power, industry, ground transport and domestic aviation) for EU27 & UK from 2019 to 2021 between CM-EU and Kayrros Carbon Watch data20. (b) Comparison of monthly CO2 emissions in France for power, industry, ground transport, domestic aviation and residential from 2019 to 2021 among CM-EU, Kayrros Carbon Watch data20 and CITEPA monthly emission bulletin14.

Full size image

For domestic aviation, we found that CM-EU emissions are lower than Kayrros both in EU27 & UK and France in 2019 (Fig. 8). A possible reason for this difference is that our definition of domestic flights in France only includes trips within metropolitan France, while Kayrros includes flights from France to EU countries and from metropolitan France to French overseas territories. This broader inclusion in Kayrros definition explains why their emissions are larger. The same difference in definitions applies to EU27 & UK. However, it is worth noting that Kayrros estimates for domestic aviation became lower than CM-EU after the outbreak of the COVID-19 pandemic. One possible reason for the lower estimates of domestic aviation emissions by Kayrros compared to CM-EU after the COVID-19 pandemic outbreak is that Kayrros took into account a passenger load factor in their estimates, while CM-EU did not. It is important to note that the passenger load factors of flights have significantly dropped since the COVID-19 pandemic43, leading to lower emissions estimates by Kayrros. The reason why Citepa emissions are higher than CM-EU and Kayrros for domestic aviation (Fig. 8) is because the definition of domestic flights for Citepa is all flights in and out of metropolitan-France (mainland), and half of flights between Mainland and French overseas territories (the other half is considered of the responsibility/accounting of the overseas territories). Non-commercial flights are also considered for the estimates of Citepa, which are not in CM-EU and Kayrros.

For industry, the CM-EU emissions are higher than Kayrros, but lower than Citepa. A possible reason for this is the scope of Kayrros covering the regulated industrial installations is lower than the total industry sector of CM-EU, and there are the balances of attributions between power and industry for power production on industrial sites in Kayrros. While all the manufacturing industries, construction and cement production are considered in CM-EU and Citepa. Citepa also considers other industrial productions.

For power, the CM-EU emissions in EU27 & UK are almost the same as Kayrros but small differences in winter peak and summer trough (Fig. 8). A possible reason for this is both datasets use the electricity production by fuel type as activity data from the same data sources. Kayrros can catch the differences from fuel burning efficiency in facility level while CM-EU can’t. But the differences canceled each other when aggregating data for all countries, though they used different annual power emissions as baseline and different emission factors. In France, CM-EU only consider the emissions from power generation in the ‘power’ sector. While urban heating and other electricity self-producers industries are considered in Citepa. Thus, Citepa emissions for ‘power’ are larger than CM-EU. While Kayrros used the annual power emissions from individual French ETS plants annual reporting as the baseline, which are nearly twice larger than those of CM-EU, with largest differences during the cold season. Kayrros also considered the fuel burning efficiency in facility level. This indicates that the aggregate ‘emission factor’ based on 2019 data from EDGAR in CM-EU is lower than the emission factors deduced by Kayrros from ETS declared plant level emissions, a difference that deserves further investigation in the future. Thus, the data of Kayrros have almost the same temporal pattern but are larger with the CM-EU. Due to the different splits of power and industry sector, we add the sum of power + industry for France in Fig. 8b to show that the sum of emissions is in better agreement.

For ground transport, the CM-EU emissions are almost the same than those of the Citepa, but the Kayrros emissions are larger in the summer and lower in the winter. The mobility from trains, expected to increase in the vacation period, was not filtered from vehicles in Kayrros, so the data in the period from July to September were removed.

For residential emissions, the CM-EU emissions are higher in the summer than Citepa. This may be because we assume that the residential emissions in the summer are equal to the emissions from cooking, and cooking emissions were assumed to remain stable all the year, while the emissions from heating were assumed to depend on and vary by the heating demand.

Technical validation for ground transportation sector and residential sector

In the ground transport sector, we assumed a linear relationship between traffic counts data and emissions or fossil fuel consumption, which assumes that large scale congestion (emissions per vehicle depend on their speed) and vehicle fleets types and emission factors have not changed significantly. To validate these assumptions, we compared annual ground transportation traffic counts with CO2 emissions and fossil fuel use in EU countries and the United Kingdom from 2010 to 2019 in Fig. 9. The annual traffic counts data we used come from Eurostat44, defined as motor vehicle movements on national territory (irrespective of registration country), covering the United Kingdom and EU countries except Denmark, Germany, Greece, Luxembourg, Malta, Portugal and Slovakia. The annual fossil fuel use of ground transport sector data come from IEA45, including four types of fossil fuel (coal, crude oil, oil products and natural gas). The CO2 emissions data of ground transport are from EDGAR8, including road transportation, rail transportation, inland navigation and other transportation. The comparison statistics shown in Fig. 9 indicate that the coefficient of determination (R2) values are 0.7891 between traffic counts and CM-EU ground transport emissions and 0.7834 between traffic counts and fossil fuel use of ground transport, respectively. These findings affirm the assumption’s validity for the EU region and interregional shifts.

Fig. 9
figure 9

Correlation of annual ground transportation traffic counts with CO2 emissions and fossil fuel use in EU countries and the United Kingdom.

Full size image

For the residential sector, only temperature is assumed to cause variation of emissions in our approach. Ciais et al.46 looked at changes of direct natural gas use in the residential sector for few European countries from ENSTO-G pipeline data during the first half of the year 2020 and found that temperature variations indeed dominated the observed changes despite the confinements in the cold season, excepted for France and Italy where an additional reduced consumption independent of temperature was found during the weeks of strictest confinements. Here we compared daily emissions from CM-EU with estimates derived from natural gas use from ENSTO-G47 in Austria, Belgium_Luxembourg (Belgium and Luxembourg combined), Bulgaria, Croatia, Denmark_Sweden (Denmark and Sweden combined), Finland, France, Germany, Greece, Hungary, Ireland, Italy, Latvia_Estonia (Latvia and Estonia combined), Netherlands, Poland, Portugal, Romania, Slovakia, Slovenia and United Kingdom. Here we use natural gas use data from ENTSO-G as activity data instead, and annual residential emissions in 2019 from EDGAR as baseline. The average yearly emission factor of the whole residential sector is assumed to remain constant at its 2019 value, so that daily emissions are estimated as:

$$Emi{s}_{residential,daily,2019}=Emi{s}_{residential,yearly,2019}\times \frac{A{D}_{residential,daily,2019}}{A{D}_{residential,yearly,2019}}$$
$$Emi{s}_{residential,daily,2020or2021}=Emi{s}_{residential,daily,2019}\times \frac{A{D}_{residential,daily,2020or2021}}{A{D}_{residential,daily,2019}}$$

The ENTSO-G consumption data was completed with Trading Hub Europe (THE)48 for the German and e-control49 for Austria. Note that the consumption sectors are not provided by e-control dataset. We further split the consumption into more detailed sectors, including household and public buildings heating, industry, and others based on energy balance datasets from Eurostat50 following the method from Zhou et al.51 The activity data used here are only from gas fuel of household and public heating, and we assume that there is a linear relationship between gas fuel use and total fossil fuel use for residential sector. The results in Figs. 10, 11 show that the variation of residential emissions in our study are similar to those of ENTSO-G in most countries (R2 ranges from 0.67 to 0.95). Latvia_Estonia (R2 = 0.52) and Portugal (R2 = 0.17) stand out with significant differences in their residential emissions data compared to the ENTSO-G results. A possible reason is that the variations of coal and petroleum uses for residential heating are not considered in the activity data from ENTSO-G. Therefore, in certain countries where coal and oil are extensively used in residential sector, gas consumption data may not be indicative of the overall energy use patterns in the residential sector.

Fig. 10
figure 10

Comparison of daily CO2 emissions of residential sector from 2019 to 2021 between CM-EU and ENTSO-G.

Full size image
Fig. 11
figure 11

Correlation between CM-EU residential emissions and estimates derived from natural gas use from ENSTO-G.

Full size image

Uncertainty analysis

There are two main sources of uncertainties in the CM-EU data. 1. The uncertainty inherited from the EDGAR annual national emissions used for the reference year 2019. 2. Uncertainty from daily activity data and models used to downscale them into daily emissions. The uncertainty analysis was conducted based on the 2006 IPCC Guidelines for National Greenhouse Gas Inventories35 and the methodology outlined in the global Carbon Monitor1,21. First, uncertainties were calculated for each sector of CM-EU.

For the power sector, CM-EU uses daily statistics of actual thermal production as activity data. When no uncertainty information is available, the 2-sigma uncertainty of the power activity data is assumed to be ±5% according to the IPCC recommended default uncertainty range of energy statistics35. In addition, for emission factors, the uncertainties mainly come from the variability of coal emission factors (as coal has a wide range of emission factors of different coal types) and changes in the mix of fuels in thermal power production. CM-EU calculates emission factors based on annual thermal production7 and annual power emissions8, and the uncertainty range is ±13%. We used error propagation equations to combine the aforementioned uncertainties of each part and estimated the uncertainties of annual power emissions as ±10%.

For the industry sector, a 2-sigma uncertainty (±36%) of CO2 emissions from industry and cement production is estimated from monthly production data and sectoral emission factors. The uncertainty of industrial output data is assumed to be ±20% in the industry sector52. For the sectoral emission factor uncertainty, according to Carbon Monitor methodology, we calculate national emission factors in 2010–2012 in USA, France, Japan, Brazil, Germany, and Italy according to data availability of monthly emission data and IPI data, and their 2-sigma uncertainties vary from ±14% to ±28. Thus, we adopt a conservative uncertainty of ±30% for CM-EU emissions in this sector.

For the ground transport sector, the global Carbon Monitor methodology assessed a 2-sigma uncertainty of ±9.3% from the prediction interval of the regression model built in Paris to estimate the emissions from this sector. Note that the regression model in Paris between car counts and the TomTom congestion index was based on assuming a relative magnitude in car counts; thus, emissions follow a similar relationship with the TomTom congestion index in Paris.

For the residential sector, global Carbon Monitor compares the estimates by using our methodology with estimates from publicly available natural gas daily consumption data by residential and commercial buildings for France (https://www.smart.grtgaz.com/fr/consommation). The 2-sigma uncertainty of the daily emission estimations is further quantified as ±40%.

For the aviation sector, global Carbon Monitor compares estimates by using two different activity data, i.e., the flight route distance (what we used in this study) and the number of flights and calculate the average difference to quantify the uncertainty of ±10.2% in the aviation sector.

Overall, the uncertainty ranges of the power, ground transportation, industry, residential, and aviation are ±10.0%, ±9.3%, ±30.0%, ±40.0%, and ±10.2%, respectively and the uncertainty in the emission of EDGAR for 2019 is estimated as ±7.1%53.

Then, we combine all the uncertainties by following the error propagation equation.

$${U}_{total}=\frac{\sqrt{\sum {U}_{s}\times {\mu }_{s}}}{\left|\sum {\mu }_{s}\right|}$$

where Us and μs are the percentage and quantity (daily mean emissions) of the uncertainty of sector s, s respectively. The overall uncertainty is quantified as ±13.6%.

We also make the technical validation for CM-EU. Therefore, the uncertainty is equal to the maximum value between the uncertainty ranges and the mean relative uncertainty from technical validation data. Finally, the overall uncertainty range of CM-EU is estimated as ±13.6%.

Usage Notes

The generated datasets are available from https://doi.org/10.6084/m9.figshare.20219024.v2. We recommend loading the data with a script that can handle large datasets. Users should also note that the unit of emissions in this dataset is Mt CO2. Latest updates and related information are available for view and download on our website https://eu.carbonmonitor.org/.