Background & Summary

Municipal wastewater treatment facilities are the main technical solution to mitigating water pollution. But wastewater purification in WWTPs and other treatment facilities always comes at the cost of energy consumption, use of chemicals and environmental impacts1,2, among which, GHG emissions are of most concern3,4. Even though GHG emissions from wastewater make only a small contribution to global anthropogenic GHG emissions, it is still important to map GHG emissions from wastewater treatment systems, and to set reasonable targets for mitigation of GHG emissions5,6. To achieve these purposes, a comprehensive GHG inventory is a prerequisite. There have been numerous studies establishing GHG accounts of WWTPs7,8,9,10,11,12,13, but there still exist challenges and problems.

Current GHG accounts often do not consider differences of treatment processes/technologies. The accounting of GHG emissions of WWTPs at the regional level mainly uses IPCC emission factors, where the centralized biological treatment processes are only categorized into aerobic and anaerobic processes but neglect the differentiation of sub-categories of aerobic or anaerobic technologies7,10,11,12,13,14, leading to large uncertainties of GHG emission factors. To accurately account GHG emissions in WWTPs, detailed processes/technologies should be considered and analysed.

Frequently, only CH4 and/or N2O are accounted for, excluding CO2 emissions of biological treatment processes as ‘these are generally derived from modern (biogenic) organic matter in human excreta or food waste and should not be included in national total emissions (IPCC 2019, Volume 5, Chapter 6, Page 7)’15. But intensive research has shown that a significant amount of fossil CO2 are directly emitted from WWTPs, and assuming that all direct CO2 emissions are biogenic may underestimate GHG emissions16,17,18,19,20.

Dissolved GHG in the treated effluent themselves have the potential to be released. In addition, many waterways are in eutrophic or nutrient-rich conditions, which can further induce discharged wastewater to increase GHG emissions15. However, GHG emissions from receiving waters are rarely accounted for, due to a lack of data of the water quality of the recipient body of water and downstream discharge pathways. Even though some studies considered off-site emissions from the treated effluent, only one discharge pathway of entering rivers, lakes or oceans was assumed7,8,9. To account emissions from different discharge pathways (such as direct discharge into rivers, lakes, reservoirs, seas, soil, and sewage irrigated farmland) is essential for identifying key emission sources, GHG composition and their contribution to the whole wastewater treatment system.

Existing reginal- or national-level studies on GHG emissions accounting of wastewater treatment systems are not comparable. This is mainly due to different emission factors and data sources in different studies. For example, Zhao, et al.10 used firm-level activity data and IPCC 2006 emission factors to calculate CH4 emissions, while emission factors of Yan, et al.11 were obtained from the average of four references excluding IPCC emission factors, and provincial-level activity data from China Environment Yearbook and China Statistical Yearbook. Differences in applied methodology and data sources contribute to a factor 38 difference in calculated CH4 emissions for the same year.

To solve the above gaps, we constructed a high-resolution (firm-level) and time series (from 2006 to 2019) GHG emission inventory of WWTPs in China. Emission sources include on-site emissions from biological treatment processes and off-site emissions from energy consumption and discharge pathways of the WWTP. We distinguished between 10 potential pathways: direct and indirect (after sewers) discharge into seas; direct and indirect discharge into rivers, lakes, reservoirs etc.; municipal WWPTs; direct discharge onto sewage irrigated farmlands; discharge onto land; other facilities (decentralized wastewater treatment facilities); centralized industrial WWTPs and other discharge pathways. To account for the different emission potentials of different treatment technologies, we calculated emissions based on 48 separate biological, physical, chemical and physicochemical technologies and their combinations. GHG emission factors of different biological treatment technologies in line with China’s conditions were obtained from the literature. Three GHG were estimated in this research, i.e., CO2, N2O and CH4. We did not distinguish between fossil CO2 and biogenic CO2 emissions from biological treatment but regarded CO2 emission as the sum of fossil CO2 and biogenic CO2 emissions.

Methods

We include GHG emissions from domestic wastewater treated by municipal WWTPs and other facilities in this paper. The other facilities mainly collect and treat wastewater discharged from residential areas, tourist facilities, resorts, nursing homes, airports, railway stations, and other public places. Domestic wastewater collected by both municipal WWTPs, and other facilities maybe mixed with industrial wastewater under certain conditions. In this case, IPCC 2019 suggests that the mixed domestic and industrial wastewater can be considered as domestic wastewater15.

GHG emissions of a WWTP result from on-site and off-site emissions. On-site emissions are usually defined as emissions induced by wastewater and sludge treatment processes of WWTPs21,22. In our study, the system boundary excludes GHG emissions from sludge treatment and disposal processes in a WWTP due to lack of data, even though it is reported that sewage sludge treatment and disposal processes account for about 40% GHG emissions in wastewater systems23. On the other hand, generated CH4 emissions from a WWTP are rarely recovered or flared in China, we regard recovered or flared CH4 emissions as being zero. Therefore, on-site emissions only refer to emissions from wastewater treatment procedures in this research. For various wastewater treatment technologies, biological treatment technologies generate on-site GHG emissions during wastewater treatment processes, but physical, chemical, and physicochemical treatment technologies do not. Off-site emissions refer to emissions from effluent, electricity consumption, production and transportation of chemicals. But we exclude off-site emissions generated by chemicals’ production and transportation due to lack of data for each WWTP, and they being negligible compared with electricity consumption13. CO2 emissions from electricity consumption are fossil CO2, because they come from coal-fired power generation, but CO2 emissions generated by on-site wastewater treatment and off-site effluent are mixed with fossil CO2 and biogenic CO2, as influent and effluent COD may contain both fossil and biogenic carbon.

Figure 1 shows a flowchart of the construction of the firm-level GHG emission inventory of wastewater treatment facilities from 2006 to 2019 in China. The first step to quantify GHG emissions of a WWTP is to judge the applied treatment technology. If the WWTP adopts a biological process, on-site emissions from the biological treatment process are calculated. Otherwise, off-site emissions from electricity consumption and the discharge pathway for each WWTP are quantified. Calculation of GHG emissions from each emission source was based on the multiplication of emission factors and activity data. The activity data for each WWTP was collected from China Environmental Statistics Database (CESD)24,25,26,27,28,29,30,31,32,33,34,35,36,37.

Fig. 1
figure 1

The flowchart of the construction of firm-level GHG emissions inventory of wastewater treatment facilities from 2006 to 2019 in China. Only biological treatment processes emit on-site GHG, but physical, chemical, and physicochemical treatment technologies do not generate on-site GHG emissions.

Full size image

Classification of wastewater treatment technologies and its priority

To examine GHG emissions of different wastewater treatment processes, we need to decide the category of technology applied in each WWTP. In most cases, a WWTP has a primary, secondary or tertiary treatment process, and for each process, especially in secondary treatment, more than one technology may be applied. It is impossible to quantify on-site GHG emissions for each technology, since we only collected data on concentration of influent and effluent pollutants for the whole WWTP, rather than for each technology or process. Therefore, to simplify the calculations of on-site GHG emissions, we first need to judge the main category of treatment technology of a WWTP, and then choose the corresponding emission factors of CH4, N2O and CO2 to calculate GHG emissions generated by biological treatment processes. The technology classification is presented in Table 1. A decision tree for determining the category of treatment technology of a WWTP is shown in Fig. 2.

Table 1 Classification of treatment processes of WWTPs in China .
Full size table
Fig. 2
figure 2

A decision tree for determining the category of treatment technology of a WWTP.

Full size image

On-site emissions from biological treatment

CH4, N2O and CO2 emissions estimated by this study

WWTPs or other treatment facilities, which have biological treatment processes, emit CH4, N2O and CO2 directly, which were calculated by Eq. 1.1, 1.2 and 1.3, respectively. The CH4, N2O and CO2 emission factors of different biological treatment processes adopted in this study were obtained from the literature, and most were studies on GHG emission factors of existing Chinese WWTPs. On the other hand, some emission factors were adopted from the IPCC 2019 report, laboratory-based studies or other models, because of a lack of studies on emission factors of full-scale wastewater treatment processes. Detailed CH4, N2O and CO2 emission factors from the literature were summarised in Table S1, Table S2 and Table S3, respectively. We obtained the minimum, maximum and average values of emission factors for each biological treatment process. The average values of emission factors were defined as the default emission factors in this study, and they are shown in Table 2. We also list IPCC 2019 emission factors of biological treatment processes in Table 2 for comparison. For those WWTPs or other treatment facilities adopted by the combined enhanced activated sludge treatment technology, their emissions factors are mean of emission factors of specific enhanced activated sludge treatment technologies (i.e., AO, A2O, OD or SBR). CH4 and N2O are converted to the CO2 equivalent by Global Warming Potential (GWP) values for 100 years. The GWP of N2O, CH4 and CO2 are 265, 28 and 1, respectively38.

$$\begin{array}{c}CH{4}_{bio,i}=E{F}_{bio,CH4,j}\ast A{D}_{bio,CH4,i}\ast 28\end{array}$$
(11)
$$\begin{array}{c}N2{O}_{bio,i}=E{F}_{bio,N2O,j}\ast T{N}_{in,i}\ast 265\end{array}$$
(12)
$$\begin{array}{c}CO{2}_{bio,i}=E{F}_{bio,CO2,j}\ast CO{D}_{removed,i}\end{array}$$
(13)
Table 2 Default GHG emission factors of biological treatment technologies in this study and in IPCC 2019.
Full size table

Where, \(CH{4}_{bio,i}\), \(N2{O}_{bio,i}\) and \(CO{2}_{bio,i}\) refer to CH4, N2O, and CO2 emissions (g CO2eq/year) from biological treatment processes in the i th WWTP. \(E{F}_{bio,CH4,j}\) (g CH4/kg COD removed or g CH4/kg COD), \(E{F}_{bio,N2O,j}\) (g N2O/kg TN influent) and \(E{F}_{bio,CO2,j}\) (g CO2/kg COD removed) are three GHG emission factors of the process j in the i th WWTP. \(A{D}_{bio,CH4,i}\) is activity data of biological CH4 emissions. There are two types of \(A{D}_{bio,CH4,i}\). When the unit of \(E{F}_{bio,CH4,j}\) for the process j is g CH4/kg COD removed, \(A{D}_{bio,CH4,i}\) is the removed COD per year (kg COD removed/year) in the ith WWTP. But \(A{D}_{bio,CH4,i}\) refers to the difference between influent COD mass and COD transferred to sludge if the unit of \(E{F}_{bio,CH4,j}\) is g CH4/kg COD. In the section of ‘Calculation of COD removed in the form of sludge’, we described how to estimate the COD transferred in the form of sludge for each process. \(T{N}_{in,i}\) is the annually influent TN mass (kg TN influent/year) in the ith WWTP, and \(CO{D}_{removed,i}\) is the annually removed COD (kg COD removed/year) in the ith WWTP.

CH4 and N2O emissions estimated by IPCC 2019

To make a comparison with our study, we also used the method of IPCC 2019 to calculate CH4 and N2O emissions from biological treatment processes. CH4 and N2O emission factors for each wastewater treatment process are from IPCC 2019 (Table 2).

$$\begin{array}{c}CH{4}_{IPCC\_bio,i}=[E{F}_{IPCC\_bio,CH4,j}\ast (CO{D}_{in,i}-{S}_{COD,i})-{R}_{CH4,i}]\ast 28\end{array}$$
(14)
$$\begin{array}{c}N2{O}_{IPCC\_bio,i}=E{F}_{IPCC\_bio,N2O,j}\ast T{N}_{in,i}\ast 265\end{array}$$
(15)

where, \(CH{4}_{IPCC\_bio,i}\) and \(N2{O}_{IPCC\_bio,i}\) refer to CH4 and CO2 emissions (g CO2eq/year) from biological treatment processes in the ith WWTP. \(E{F}_{IPCC\_bio,CH4,j}\) (g CH4/kg COD) and \(E{F}_{IPCC\_bio,N2O,j}\) (g N2O/kg TN influent) are IPCC 2019 CH4 and N2O emission factors of the process j in the i th WWTP. \(CO{D}_{in,i}\) is the annually influent COD mass (kg COD influent/year) in the ith WWTP. \({S}_{COD,i}\) (kg COD removed as sludge/year) is the COD removed in the form of sludge in the ith WWTP. \({R}_{CH4,i}\) is amount of CH4 recovered or flared from the ith WWTP. This value was regarded as being zero because there are very few CH4 recovered or flared in China. \(T{N}_{in,i}\) is the annually influent TN mass (kg TN influent/year) in the i th WWTP.

Calculation of COD removed in the form of sludge

$$\begin{array}{c}{S}_{COD,i}=CO{D}_{removed,i}\ast {Y}_{obs,j}\ast 1.42\end{array}$$
(16)
$$\begin{array}{c}CO{D}_{removed,i}=(CO{D}_{in}-CO{D}_{out})\ast {V}_{wastewater,i}\end{array}$$
(17)

where, \({S}_{COD,i}\) (g COD removed as sludge/year) is the COD removed in the form of sludge in the ith WWTP, \(CO{D}_{removed,i}\) (g COD/year) is the COD removed of the ith WWTP. \({Y}_{obs,j}\) (g VSS/ g COD) is the observed sludge yield of process j in the ith WWTP. 1.42 (g COD/ g VSS) is the conversion factor that determine biomass concentration in terms of COD39. \(CO{D}_{in}\) and \(CO{D}_{out}\) are influent and effluent COD concentration of the ith WWTP. \({V}_{wastewater}\) is the volume of treated wastewater in the ith WWTP. The coefficient of \({Y}_{obs,j}\) (g VSS/ g COD) for each process is from Chen et al.40. Since a membrane bioreactor (MBR) is the combination of an enhanced activated sludge process and a membrane process, its \({Y}_{obs,j}\) was estimated by the average value of observed sludge yield of an enhanced activated sludge process and a biofilm process. Coefficients \({Y}_{obs,j}\) of different treatment processes are shown in Table 3.

Table 3 Coefficients Yobs of biological treatment processes.
Full size table

Off-site emissions from discharge pathways

Treated wastewater was discharged in one of 10 different pathways. Table 5 shows emission factors of CO2, N2O and CH4 of each discharge pathway. The effluent emission factors of CH4 and N2O were adopted from IPCC 2019, while the CO2 emission factors of the treated effluent were derived from the appendix of IPCC 2019 (IPCC 2019, Volume 5, Chapter 6, Page 59-Page 60)15. The detailed derivation process of CO2 emission factor of effluent discharge refers to Supplementary Information ‘CO2 emission factor of effluent discharge’. Emissions from discharge pathways were calculated by Eq. 2.1–2.3:

$$\begin{array}{c}CH{4}_{eff,i}=E{F}_{eff,CH4,j}\ast CO{D}_{out,i}\ast 28\end{array}$$
(21)
$$\begin{array}{c}N2{O}_{eff,i}=E{F}_{eff,N2O,j}\ast T{N}_{out,i}\ast 265\end{array}$$
(22)
$$\begin{array}{c}CO{2}_{eff,i}=E{F}_{eff,CO2,j}\ast CO{D}_{out,i}\end{array}$$
(23)

where, \(CH{4}_{eff,i}\), \(N2{O}_{eff,i}\) and \(CO{2}_{eff,i}\) are CH4, N2O and CO2 emissions (g CO2eq/year) from the discharge pathway j in the ith WWTP. \(E{F}_{eff,CH4,j}\) (g CH4/kg COD effluent), \(E{F}_{eff,N2O,j}\) (g N2O/kg TN effluent) and \(E{F}_{eff,CO2,j}\) (g CO2/kg COD effluent) are effluent emission factors of the discharge pathway j of the ith WWTP. \(CO{D}_{out,i}\) (kg COD effluent/year) and \(T{N}_{out,i}\) (kg TN effluent/year) are annually effluent COD and TN mass of the ith WWTP.

Off-site emissions from electricity consumption

The calculation of GHG emissions from electricity consumption is shown in Eq. 3.1. Baseline emission factors for regional power grids in China41,42,43,44 were used in this study. Only CO2 is considered for emission factors for regional power grids without considering N2O and CH4 due to their small contributions. China’s baseline emission factors for regional power grids are presented in Table 4.

$$\begin{array}{c}CO{2}_{ele,i}=E{F}_{ele,CO2,j}\ast El{e}_{con,i}\end{array}$$
(31)

where, \(CO{2}_{ele,i}\) is the CO2 emission from electricity consumption (kg CO2/year). \(E{F}_{ele,CO2,j}\) (kg CO2/kWh) denotes the CO2 emission factor of province j of the studied WWTP. \(El{e}_{con,i}\) (kWh/year) refers to the electricity consumption of the ith WWTP.

Table 4 Baseline emission factors for regional power grid in China from 2006 to 2019 (Unit: kg CO2/kWh).
Full size table
Table 5 Emission factors of different GHG emissions from discharge pathways.
Full size table

Uncertainty analysis

The uncertainty of GHG emissions was mainly caused by emission factors. Since calculation of activity data of each WWTP was based on annual on-site monitored data of the volume of treated wastewater, influent and effluent concentration of pollutants and electricity consumption, there is no uncertainty for activity data. We analysed GHG emissions uncertainty induced by biological treatment processes and discharge pathways. The uncertainty caused by electricity consumption was not considered, because China’s regional power grid baseline emission factors are based on specific values rather than ranges.

For the emission factors of biological treatment processes, we acquired the minimum, maximum and average emission factors of each technology from the literature. Then, we used the following Eq. 4.1 and 4.2 to calculate the uncertainty of emission factors.

$$\begin{array}{c}Uncertainty\;lower\;bound\;Ulb=\frac{\left(E{F}_{min}-E{F}_{ave}\right)}{E{F}_{ave}}\ast 100 \% \end{array}$$
(41)
$$\begin{array}{c}Uncertainty\;upper\;bound\;Uub=\frac{\left(E{F}_{max}-E{F}_{ave}\right)}{E{F}_{ave}}\ast 100 \% \end{array}$$
(42)

Since the CH4 emission factor was determined by the multiplication of the maximum producing potential (B0) and the methane correction factor (MCF), its uncertainty was measured by Eq. 4.3. The uncertainty of B0 (UB0) is ± 30% in IPCC 2019, and the uncertainty of MCF (UMCF) was determined by Eq. 4.1 and 4.2. The uncertainties of N2O and CO2 emission factors of discharge pathways were calculated by Eq. 4.1 and 4.2.

$$\begin{array}{c}{U}_{CH4}=\pm \,\sqrt{{U}_{{B}_{0}}^{2}+{U}_{MCF}^{2}}\end{array}$$
(43)

We applied Monte Carlo simulations to analyse the combined uncertainty of emission factors and activity data. Emission factors of CH4, N2O and CO2 of biological treatment processes and discharge pathways all follow triangular distributions, because ‘upper and lower and a preferred value are provided (IPCC 2006, Volume 1, Chapter 3, Page 22)’15 in this study. Random sampling on emission factors was performed 100,000 times, then multiplied by activity data of each GHG in each WWTP, generating 100,000 values for GHG emissions. Finally, uncertainty ranges of 95% confidence intervals of GHG emissions were adopted.

Other causes that may induce uncertainties include ‘Measurement error’, ‘Lack of completeness’ and ‘Misreporting or misclassification’. With regard to the measurement error in a real WWTP, the measured influent and effluent concentration of pollutants and electricity consumption may be incorrect. But this uncertainty is difficult to quantify and control in this study. In terms of lack of completeness, the original data was incomplete for all WWTPs. For instance, data of some indicators was lacking, e.g., volume of treated wastewater, influent, or effluent concentrations of COD. When a WWTP does not have sufficient indicators, the WWTP was removed, and its emissions were not calculated. For the misreporting or misclassification, accurate classification of treatment technologies is the basis for calculating GHG emissions of secondary biological treatment processes, but uncertainties caused by misreporting and/or misclassification of treatment technologies are possible and cannot be easily rectified.

Data Records

The dataset of “Greenhouse gas emissions of wastewater treatment plants in China from 2006 to 2019” is made public under Figshare45. There are 400,512 data records in the dataset. These include:

  • 399,420 firm-level GHG emission inventories (57,060 firms, i.e., 57,060 WWTPs and other wastewater treatment facilities; for each firm, there are CH4, N2O and CO2 emissions from biological treatment processes, CO2 emissions from electricity consumption, and CH4, N2O and CO2 emissions from the discharge pathways);

  • 70 annual biological treatment GHG emission inventories (from 2006–2019, CH4 and N2O emissions calculated by IPCC 2019 methodology, and CH4, N2O and CO2 emissions calculated as described in the section of Methods);

  • 42 annual effluent GHG emission inventories (from 2006–2019, CH4 and N2O emissions calculated by the IPCC 2019 methodology, and CO2 emissions calculated by the method of this paper);

  • 14 annual electricity CO2 emission inventories (from 2006–2019);

  • 322 annual CO2eq emissions of different technologies from biological treatment processes (from 2006–2019, 23 technology categories);

  • 322 are annual CO2 emissions of different technologies from electricity consumption (from 2006–2019, 23 technology categories);

  • 322 annual CO2eq emissions of different technologies from discharge pathways (from 2006–2019, 23 technology categories).

In this study, the firm-level GHG emission inventory provides a foundation for the remaining emission inventories. Based on the firm-level GHG emission inventory, annual CH4, N2O and CO2 emission inventories of biological treatment processes, effluent and electricity consumption are presented, and annual total CO2eq emissions of different technologies from biological treatment processes, electricity consumption and discharge pathways are also quantified.

Figure 3 presents annual CH4, N2O and CO2 emissions from different emission sources and annual treated wastewater from 2006 to 2019. The pie charts in Fig. 4 show the structure of treatment technology in total CO2eq emissions in 2006, 2010, 2015 and 2019, respectively. Treatment technologies are classified by main categories of processes based on the classification in Table 1. Since the enhanced activated sludge process is the main wastewater treatment technology in China and it includes many sub-categories, the emission structure of sub-categories (i.e., AO, A2O, OD and SBR) of the enhanced activated sludge process is also shown in pie charts.

Fig. 3
figure 3

China’s GHG emissions from wastewater treatment (in million tons CO2eq) and treated wastewater (in billion cubic meters) 2006 to 2019. Ele, Bio and Eff indicate GHG emissions from electricity consumption, biological treatment processes and effluent discharge.

Full size image
Fig. 4
figure 4

Structure of treatment technology in total CO2eq emissions in 2006, 2010, 2015 and 2019 (in million tons CO2eq). GHG emissions from enhanced activated sludge processes and conventional activated sludge accounted for a large proportion (>80%) in 2010, 2015 and 2019. While the percentage from biological treatment process was very high (58%) in 2006, because for some WWTPs, their subcategories of biological treatment processes were not reported in the original dataset. In this case, their treatment technologies were named as biological treatment, and their GHG emissions were estimated by emission factors of the process of activated sludge treatment in this study.

Full size image

Technical Validation

Uncertainty analysis

Uncertainty of emission factors

The uncertainty of CH4, N2O and CO2 emission factors of biological treatment technologies is presented in Table 6. For comparison, we also list the uncertainty of CH4 and N2O emission factors based on IPCC 2019. The analysis by IPCC 2019 shows higher uncertainty in terms of CH4 and N2O emission factors of a majority of biological treatment technologies, due to its less detailed classification of technologies. For instance, different activated sludge technologies in IPCC 2019 possess the same emission factors and uncertainties, because IPCC 2019 classifies all activated sludge processes into one aerobic process category. However, processes of AO, A2O, SBR and OD are quite different, although they are all activated sludge technologies. Since we collected GHG emission factors based on different categories of traditional or enhanced activated sludge processes, emission factors and their uncertainties of processes of AO, A2O, SBR and OD are different and have different ranges (Table 6). However, on-site emission factors of certain processes are rarely reported in the literature, and we cannot obtain their emission factors based on detailed process classification. For example, we applied a CH4 emission factor (200 g CH4/kg COD) of the anaerobic process from IPCC 2019 to four different anaerobic processes (i.e., anaerobic hydrolysis, typical anaerobic reactors, anaerobic biofilter, and other anaerobic biological treatment), due to a lack of their on-site emission factors from references. Therefore, reported uncertainties (−30%,39%) for CH4 emission factors of the four anaerobic processes are the same. Overall, the uncertainties of GHG emission factors of different biological treatment technologies were relatively high. One of the main reasons is that GHG emission factors are strongly affected by different operational parameters46,47,48,49 (temperature, pH, dissolved oxygen (DO), sludge retention time (SRT), hydraulic retention time (HRT), influent chemical oxygen demand (COD) to total nitrogen ratio (C/N), influent chemical oxygen demand (COD) to total phosphorus ratio (C/P), etc.) of these WWTPs.

Table 6 Uncertainty of CH4, N2O and CO2 emission factors of biological treatment technologies.
Full size table

The uncertainty of CH4, N2O and CO2 emission factors of 10 discharge pathways is shown in Table 7. Since CH4 and N2O emission factors for the discharge pathway of ‘flowing sewer (open or closed)’ are zero in IPCC 2019, we assumed that there was no CO2 generation under this flowing condition. We regarded discharge pathways via municipal WWPTs, centralized industrial WWTPs and other facilities (decentralized wastewater treatment facilities) as discharge pathway of ‘flowing sewer’. Therefore, we do not report any uncertainty of CH4, N2O and CO2 emission factors of entering municipal WWTPs, industrial WWTPs and other facilities. We considered the discharge pathway of ‘other discharge pathways’ in this study as ‘discharge to aquatic environments (Tier 1)’ in IPCC 2019, and its uncertainties of CH4 (−100%, 148%) and N2O emission factors (−90%, 1394%) are the largest compared with other discharge pathways. Because there are very few studies on the CO2 emission factor of the treated effluent, we derived CO2 emission factors of lakes, rivers and reservoirs from the appendix of IPCC 2019 (IPCC 2019, Volume 5, Chapter 6, Page 59-Page 60)15, and we assumed that pathways of discharging into sea and ‘others’ also have the same CO2 emission factors. Thus, their CO2 emission factor uncertainties were all the same, with the uncertainty of (−12%, 20%).

Table 7 Uncertainty of emission factors of different discharge pathways.
Full size table

Combined uncertainty of GHG emissions

The combined uncertainty of GHG emissions of biological treatment processes is presented in Table 8 and Fig. 5(a–c). The shadow areas shown in Fig. 5 indicate the 95% confidence interval of GHG emissions. For comparison, CH4 and N2O emissions calculated by emission factors of IPCC 2019 are also shown in Fig. 5(a,b). From 2006 to 2019, the uncertainties of CH4, N2O and CO2 emissions in this study were (−57%, 124%), (−63%, 184%) and (−43%, 38%), respectively. But uncertainties of CH4 and N2O emissions calculated by the methodology of IPCC 2019 were (−91%, 189%) and (−99%, 184%). The minimum and maximum CH4 and N2O emissions calculated by IPCC 2019 were all outside of the shadow areas in Fig. 5(a,b), reflecting larger uncertainties than in our study.

Table 8 The combined uncertainty of GHG emissions from biological treatment.
Full size table
Fig. 5
figure 5

Trend and uncertainty of GHG emissions from WWTPs in China from 2006 to 2019 (in million tons CO2eq). (a) Trend and uncertainty of CH4 emissions from biological treatment. (b) Trend and uncertainty of N2O emissions from biological treatment. (c) Trend and uncertainty of CO2 emissions from biological treatment. (d) CO2 emissions from electricity consumption. (e) Trend and uncertainty of CH4 emissions from effluent. (f) Trend and uncertainty of N2O emissions from effluent. (g) Trend and uncertainty of CO2 emissions from effluent. (h) Trend and uncertainty of total CO2eq emissions. Bio: biologocal. Eff: effluent. Ele: electricity. The shadow areas indicate the 95% confidence interval of GHG emissions. The uncertainty of electricity consumption is not shown in (d) because of unavailable uncetainty of power grid baseline emission factors.

Full size image

The combined uncertainty of effluent GHG emissions is presented in Table 9 and Fig. 5(e–g). The overall uncertainties of the effluent N2O were very high (−33%, 1161%), mainly resulting from high uncertainty of the effluent N2O emission factor (−100%, 1394%). N2O emission factors vary substantially between WWTPs, due to different process designs and operational conditions46,47. Effluent CH4 and CO2 emission uncertainties were relatively low, with values of (−52%, 29%) and (−9%, 16%), respectively. The uncertainty of total GHG emissions of WWTPs are shown in Fig. 5(h) and Table S4. The uncertainties of total GHG emissions from WWTPs were about (−27%, 97%).

Table 9 The combined uncertainty of GHG emissions from effluent.
Full size table

Comparison with existing estimations

Several studies on CH4 or N2O emissions of WWTPs at the national level in China have been reported7,8,9,10,11,12,13. In Table S5, we list wastewater GHG estimations in the literature for comparison. In most cases, the current estimation results are not comparable. The use of different system boundaries across studies is one of the main reasons. For instance, CH4 emissions (76.2 Mt CO2eq) of wastewater from China’s second biennial update report on climate change50 in 2014 refer to emissions from both industrial and domestic wastewater at the national level and activity data was obtained from the Environmental Statistics Yearbook, while Zhao et al.10 considered CH4 emissions (29.2 Mt CO2eq) from 2019 WWTPs at the firm level in 229 cities in 2014 and the data was from the Urban Drainage Statistic Yearbook. Their results are not comparable, since 2019 WWTPs in Zhao et al.’s study contained mainly prefecture-level municipal WWTPs but excluded county-level and industrial WWTPs in China, and it is not clear how many WWTPs/wastewater treatment facilities are included in China’s second biennial update report. Therefore, the activity data and CH4 emissions were not comparable in these two studies, although they all used IPCC 2006 method for their inventories. In our paper, on-site CH4 emissions from 4455 WWTPs and 718 other treatment facilities were estimated to be 2.55 Mt CO2eq in 2014, which were about one tenth of Zhao et al.’s results. This discrepancy was caused by using different system boundaries and the use of different emission factors.

Most studies used emission factors from the IPCC, but even CH4 emission factors from IPCC 2006 and IPCC 2019 are quite different. The default methane correct factor (MCF) in IPCC 2019 is 0.03, while this value is 0.3 in IPCC 2006 for overloaded WWTPs, and it may differ by one order of magnitude for CH4 emissions. Our uncertainty analysis shows that CH4 emissions calculated by IPCC 2019 are about 20%–62% larger than our research, and uncertainties caused by IPCC 2019 were much higher than in this study. In other cases, emission factors from the literature without distinguishing different technologies were used to estimate GHG emissions. For example, the MCF of 0.165 was used to calculate CH4 emissions induced by domestic wastewater in several studies7,8,9,12. By using MCF 0.165, CH4 emissions from domestic wastewater were around 28 Mt CO2eq in 2014 in Du et al.7 While Yan et al.11 obtained that the estimated CH4 emissions were 0.77 Mt CO2eq in 2014 by using the emission factor of 2.3064 kg CH4/t COD removed. The discrepancy of CH4 estimations between Du et al.7 and Yan et al.11 in 2014 was nearly 36 times. In comparison, estimated CH4 emissions in our study are 2.6 Mt CO2eq. Comparing Guo et al.13 with our study, the main difference was that Guo et al. applied only one N2O emission factor (0.035 kg N2O-N/kg TN) to all treatment technologies and their CH4 emission factors were based on different provinces51. But our CH4 and N2O emission factors were based on the specific technology of each WWTP. Total CH4 and N2O emissions from biological wastewater treatment and CO2 emissions from electricity consumption in Guo et al. in 2016 were 31.4 Mt CO2eq, which are about twice of our result (15.9 Mt CO2eq).

Misuse of CH4 emission calculation formula in IPCC 2006 or IPCC 2019 is another reason leading to incomparability of CH4 emissions. Normally, CH4 emissions are equal to a CH4 emission factor times the difference between total influent COD (or BOD) mass and COD (or BOD) removed in the form of sludge. Total influent COD (or BOD) mass minus COD (or BOD) removed in the form of sludge means that organic components transferred to sludge do not generate direct CH4, but only the remaining organic matter in the wastewater has potential to emit CH4. Therefore, the unit (kg CH4/kg BOD or kg CH4/kg COD) of CH4 emission factor in IPCC indicates CH4 emissions per unit remaining organic mass in the influent after considering COD (or BOD) transferred to the sludge, rather than CH4 emissions per unit influent COD (or influent BOD) or per unit COD (or BOD) removed9. In addition, organic matter removed in the form of sludge was assumed as being zero for all treatment technologies7,8,9,10,12. The reasons for the incorrect assumption may be the lack of data on sludge generation, and the method to estimate organic components removed in the form of sludge is not mentioned in IPCC 2006, or the lack of background on wastewater treatment. The assumption may overestimate CH4 emissions as most aerobic biological treatment technologies generate sludge during wastewater treatment. However, IPCC 2019 updated the method to account CH4 emissions based on IPCC 2006, especially providing equations and detailed information to estimate COD (or BOD) transferred to sludge, which provides guidance for accurate CH4 accounting.

Limitations

We have four main limitations in this study. (1) A WWTP may have one or more wastewater treatment streams, and for each treatment stream, it may contain primary, secondary or tertiary treatment processes, while for each process (normally for a secondary treatment process), it has multiple treatment technologies. But to simplify GHG emissions estimation of biological treatment technologies of the secondary treatment process of a WWTP, the decision tree (Fig. 2) was applied to determine the main category of treatment technology and its corresponding emission factors, especially when a WWTP has several secondary treatment technologies. (2) Our emission factors of different biological treatment technologies were not based on the monitoring of each wastewater treatment plant. But we used emission factors in line with Chinese conditions. The emission factors were acquired from different references, such as on-site monitoring of specific biological technologies or modelling estimations in the literature, which was based on case studies of WWTPs in China. However, emission factors of some biological technologies, such as CH4 and CO2 emission factors of anaerobic technologies and constructed wetlands, were missing for China, thus we used IPCC emission factors for these technologies instead. On the other hand, given that emission factors of a specific biological treatment technology are greatly affected by operational conditions, different WWTPs with the same biological technology may have different emission factors. Therefore, GHG emission factors of a biological technology obtained from references are not representative for real emission factors of all WWTPs with the same technology. (3) GHG emissions from industrial WWTPs are not available and thus not included in our study although being important GHG emission sources of wastewater treatment systems52,53,54. For instance, Xing et al. reported that CH4 emissions from on-site industrial wastewater treatment were always higher than that of domestic wastewater treatment between 2003 and 2008 in China. CH4 emissions from industrial and domestic wastewater treatment were 0.95 Mt and 0.91 Mt respectively in 200854. (4) Anthropogenic CO2 emissions (or fossil CO2 emissions) from biological treatment processes and discharge pathways are of main concern compared with biogenic CO2 emissions, but we did not calculate fossil CO2 emissions separately, because the CO2 emission factors available in the literature are only reported as total CO2, rather than separate fossil and biogenic CO2.