Methane emissions from on-road vehicles in China: a case study in an urban tunnel

Reducing emissions of methane (CH4) in developed regions and urban areas is a practical way to curb the unexpected surge in global CH4 levels in recent decades. Traffic emissions are among the important anthropogenic CH4 emission sources in megacities, yet CH4 emissions from on-road vehicles are less characterized and not well addressed. Based on tunnel tests in an urban tunnel in south China, a real-world emission factor (EF) of CH4 was measured to be 0.26 ± 0.03 g·km−1 (mean ±95% C.I.) for on-road vehicle fleet which including gasoline vehicles, diesel vehicles, and liquefied petroleum gas vehicles, with an average CH4/CO2 mass ratio of 40.6E-5 g·g−1, and CH4 could account for 1.3% of vehicle CO2-equivalent emissions. Using the measured CH4/CO2 ratio and available automobile CO2 emission estimates, traffic CH4 emissions in 2014 could have reached 333 Gg and represented 0.6% of total anthropogenic CH4 emissions in China, approximately four times the previous reported value of 79 Gg. Our results indicate that improving energy efficiency would have co-benefits for reducing traffic emissions of CH4, as observed EFs of CH4 are positively correlated with that of CO2, and over 90% of traffic CH4 emissions in China could be avoided if the traffic CH4/CO2 ratio can be an order of magnitude lower as previously observed in a tunnel in Switzerland.


Introduction
Methane (CH 4 ) is the second-largest contributor to anthropogenic greenhouse forcing after carbon dioxide (CO 2 ) [1,2]. Apart from trapping the infrared energy emitted by the Earth and its atmosphere, CH 4 has extra warming effects affecting the abundance of other greenhouse gases, such as ozone (O 3 ), water vapor (H 2 O), and CO 2 [3]; becoming involved in gas-aerosol interactions [4]; and absorbing energy from the Sun at shorter wavelengths [5]. The radiative forcing attributed to methane emissions is very likely to be almost twice as large as that from its change in concentration [1]. Data from NOAA observation stations show that global mean atmospheric CH 4 started to rise in 2007 after a near-zero growth from 2000-2006 [6], with a sharper increase beginning in 2014 [7]. As a result, the global level of atmospheric methane climbed to 1.87 ppm in March, 2019 [8] from 1.78 ppm in 2006 [7]. Because CH 4 is a potent greenhouse gas with a 20-year Global Warming Potential (GWP 20 ) of 84 and a 100-year Global Warming Potential (GWP 100 ) of 28-32 [1,2,5], this unexpected surge in global methane levels presents a major challenge, and eliminating emissions of CH 4 would quickly provide a benefit for achieving the goals set out in the Paris Agreement to limit temperature increases to 2°C or, if possible, Any further distribution of this work must maintain attribution to the author(s) and the title of the work, journal citation and DOI. to 1.5°C above preindustrial levels [9,10]. In the near term, a weakening of aerosol cooling by a reduction in sulfur dioxide emissions would add to future warming, but can be tempered by reductions in methane emissions [1].
Anthropogenic methane emissions have dominated global methane growth since 1999 [11][12][13]. Although the drivers behind the surge in atmospheric methane in the recent decades are less well understood and hotly debated, there are limited opportunities to address agricultural emissions or natural emissions, and reducing methane emissions in the energy sector is the most practical option available to control global methane levels. In fact, the fractions of CH 4 emissions from the energy sector reached 44.6% in China in 2014 and 26.5% in the world in 2012 [14,15]. Considering ethics, social justice and equity issues, developed regions and urban areas should take on more responsibilities in emissions reduction efforts. Non-CO 2 forcers such as methane are emitted alongside CO 2 , particularly in the energy and transport sectors, and thus, reducing emissions of CO 2 and CH 4 in urban areas can be largely addressed as co-benefits of curbing emissions of air pollutants from the energy and transport sectors. In fact, CH 4 emissions from fossil fuel industries were found to be 20%-60% higher than previously estimated, implying a greater potential for energy efficiency improvements to mitigate anthropogenic climate forcing [16]. Transport is an important sector in fossil fuel consumption. Anthropogenic CH 4 emissions from mobile sources are estimated to contribute only 0.2%-0.5% to the global CH 4 budget [17,18], and this sector ranked 12th among all sources in 2017 in the US [19]. However, Nakagawa et al [20] revealed that up to 30% of CH 4 could come from automobile exhaust in Nagoya, Japan, suggesting substantial opportunities for reducing CH 4 from traffic emissions in megacities.
China has become the world's largest manufacturer and consumption market of motor vehicles; thus traffic emissions of CH 4 are of great concern, particularly in urban areas. However, as CH 4 is an unregulated pollutant, there are few studies concern its emissions from vehicle exhausts [21]. Emission factors (EFs) along with traffic statistics are needed to estimate traffic emissions of CH 4 [22], yet EFs of specific vehicle types are mainly derived from European or US vehicle emission databases for compiling China's CH 4 emission inventory using mobile source emission models such as the Computer Programme to Calculate Emissions from Road Transport (COPERT) [21,[23][24][25][26], MOBILE [27,28], International Vehicle Emissions (IVE) [29] and MOtor Vehicle Emission Simulator (MOVES) [30]. Due to differences in technological levels, emission control standards and driving conditions, simply borrowing EFs from developed countries will lead to significant uncertainty when estimating emissions in China [31,32]. There are only a few tests available that provide EFs for light-duty gasoline vehicles, gasoline vehicles, compressed natural gas vehicles, natural gas vehicles and passenger cars, respectively [33][34][35], or obtain the CH 4 /CO 2 ratio for a whole fleet [22,[36][37][38][39].
Tunnel tests have proven to be a useful method to estimate real-world fleet-wide EFs [22,40]. In this study, we conducted tests in 2014 in the underwater Zhujiang Tunnel with a daily traffic flow of approximately 40,000 vehicles in urban Guangzhou, South China, to provide a real-world EF of CH 4 for an on-road fleet for the first time, to recheck the status of traffic CH 4 emissions in China and to assess the potential of reducing CH 4 emissions in the transport sector.

Field work
Zhujiang Tunnel is an underwater tunnel crossing the Pearl River in west-central Guangzhou. Located in a relatively populated area, the traffic load in the tunnel reaches ∼40,000 vehicles per day. The tunnel is 1238.5 m long with a 721 m flat underwater section. It has two bores with two lanes in each bore. A schematic diagram of the tunnel was given by Liu et al [41] and Zhang et al [42]. During our test, the ventilation system of the tunnel was not operated. Zhujiang Tunnel has a speed limit of 50 km h −1 , and the traffic speed during our sampling (data from Guangzhou Transportation Administration Bureau, personal communication) ranged from 20 km·h −1 to 47 km·h −1 , averaging 35±5 km·h −1 (95% confidence intervals, 95% C.I.). Detailed descriptions of the Zhujiang Tunnel can be found in the supporting information text S1 is available online at stacks.iop.org/ ERC/2/061005/mmedia. Sampling was conducted from June 25th to July 1st in 2014. We placed the two sampling stations (outlet station and inlet station) 50 m from the ends of the 721-meter flat underwater section. Determined emission factors for air pollutants and halocarbons based on tunnel tests are reported elsewhere [41][42][43][44][45]. CO 2 and CH 4 in this study were measured using the same batch of air samples collected by canisters for the analysis of volatile organic compounds (VOCs). The VOC samples were collected during intervals of 02:00-3:00, 07:00-8:00, 08:00-9:00, 09:00-10:00, 10:30-11:30, 14:00-15:00, 17:00-18:00, 18:00-19:00, and 19:00-20:00 on June 25th, 26th, 28th and 29th. Two more samples were collected on June 27th (02:00-3:00) and June 30th (4:00-5:00) as supplements. The air samples were collected into pre-evacuated 2-liter electro-polished stainless steel canisters using a Model 910 pressurized canister sampler (Xonteck, Inc., Fremont, CA, USA). The sampler was set at a constant flow rate of 66.7 ml min −1 to guarantee that the air pressure of each canister was approximately 2 atm after sampling for 60 min. More details about quality control and quality assurance of canister samples were offered in Zhang et al [42].
An IRGASON eddy covariance system (Campbell, Inc., USA) with an integrated CO 2 and H 2 O open-path gas analyzer and a 3D sonic anemometer was used to measure in situ CO 2 , wind speed/direction and temperature. During the sampling period, the air temperature recorded by the IRGASON at the outlet station ranged from 27.9°C to 41.5°C with an average of 34.4±0.3°C (mean±95% C.I.).
Laboratory analysis CH 4 was analyzed by an Agilent Model 6890 GC equipped with an FID and a packed column (5 A Molecular Sieve 60/80 mesh, 3 m×1/8 inch). CO 2 was analyzed by the same equipment with a different packed column (10Ft 1/8 2 mm HayeSep Q 80/100 SS). CO 2 was first converted by a Ni-based catalyst to CH 4 and then detected by the FID after separation by the packed column [42]. Both CH 4 and CO 2 were quantified by an external calibration method. Working calibration curves were obtained by diluting ultra-pure CO 2 or CH 4 (>99.999%) to working standards with an Entech Model 4700 high-precision static dilution standards preparation system (Entech Instruments Inc., USA), running each of the working standard for three times and then plotting the average responses against the mixing ratios by linear regression. Re-calibration is needed: (1) if the R 2 for the linear dose-response correlation was below 0.99; and (2) the calibration curve is challenged with a NIST traceable standard (Spectra Gases, 398 ppm for CO 2 and 1.01 ppm for CH 4 ) each day before the analysis of air samples. If the determined mixing ratio is beyond ±0.5% the labelled value of the NIST traceable standard. The method detection limits (MDLs) of the analysis system for CH 4 and CO 2 are 20 ppb and 3 ppm, respectively.

Results and discussion
Emission factor (EF) The average EF for vehicles traveling through the tunnel during a time interval T can be calculated in the same way as in previous studies [46,47]: where EF fleet (mg·km −1 ) is the mean emission factor of CH 4 , Δc (mg·m −3 ) is the concentration difference between the inlet and outlet sampling stations inside the tunnel, V air (m·s −1 ) is the air speed velocity parallel to the tunnel sensed by the 3D sonic anemometer with an average of 3.54 m s −1 , A (m 2 ) is the tunnel cross section with value of 58.2 m 2 , N is the monitored traffic count passing the tunnel during the time interval T(s) (3600 s in this study), and l (m) is the distance between the two sampling points which is 0.621 km.
The results showed that the EFs of CH 4 measured in the tunnel ranged from 0.048 g·km −1 to 0.40 g·km −1 , with an average of 0.26±0.03 g·km −1 (mean±95% C.I.). The time series of EFs of CH 4 , as well as those of numbers of three fuel-type vehicles travelling through the tunnel, are shown in figure 1. During the time intervals of 2:00-3:00 or 3:00-4:00, the EFs of CH 4 became significantly lower, corresponding to a relatively lower amount (approximately 40%) of gasoline vehicles. The number of gasoline vehicles in these hours decreased by approximately 80% relative to that in rush hours. A highly significant (p<0.01) correlation was observed between the observed EFs of CH 4 and the percentages of gasoline vehicles (figure S1), implying that gasoline vehicles may dominate CH 4 emissions in the traffic fleet. Chassis dynamometer results from Koike and Odaka [33] have also showed that diesel vehicles have significantly lower emissions of CH 4 than gasoline vehicles. Even though there were fewer diesel vehicles on the weekends, no significant difference in the EFs of CH 4 was found between the workdays and weekends. Figure 2 shows a comparison of the EFs of CH 4 for vehicles measured in this study with those previously reported. The average EF of 0.26±0.03 g·km −1 from our study is approximately 50 times that of 0.0052 g·km −1 for gasoline vehicles measured using chassis dynamometer under New European Driving Cycle (NEDC) conditions [34] and over 20 times higher than that of 0.012±0.003 g·km −1 estimated for the US on-road fleet under the Federal Test Procedure −75 (FTP-75) conditions [17]. These large gaps cannot solely be explained by differences in vehicle operating conditions since NEDC and FTP-75 conditions include acceleration and idling modes during which more emissions would typically occur than during the cruising mode for vehicles in the tunnel; instead, they reflect the large gaps in fuel quality, engine performance, exhaust control and vehicle maintenance as well. As an example, during our study on-road vehicles in Guangzhou were mostly composed of Euro 3 (which equals to China 3) (39.4%) and Euro 4 (36.2%), and Euro 1 and Euro 2 still shared a notable proportion of 21.7%. Previous studies demonstrated that CH 4 emissions increase with the age of catalytic convertors [48][49][50][51]. Additionally, cars using premium grade gasoline fuel with higher aromatic hydrocarbon content and lower content of saturated hydrocarbons and olefins are found to have lower CH 4 emissions [48], and the notoriously higher amount of olefins in China' gasoline (a limit of 24% v/v even in the China 6 grade gasoline oil) would probably give rise to the CH 4 emissions from China's gasoline vehicles.
As shown in figure 2 and table S2, our measured EFs of CH 4 for the urban vehicle fleet were also substantially higher than those derived from the COPERT and MOBILE models [17,21,[23][24][25][26][27]30], but comparable to those derived from the IVE model [28,29], which was designed for mobile source emissions of developing countries by researchers at the International Sustainable Systems Research Center and the University of California at Riverside. In contrast to other models that use average speed to represent a driving cycle, the IVE model introduces parameters such as vehicle specific power and engine size to better represent driving conditions [28,52], making it more suitable for estimating vehicle CH 4 emissions in developing countries.
The CH 4 /CO 2 ratio The CH 4 /CO 2 ratio is proven to be a simple and practical way to estimate traffic emissions of CH 4 [17]. In order to better represent the emissions of CH 4 from on-road vehicles, the slope for the correlation of EFs between CH 4 and CO 2 was used as the average CH 4 /CO 2 ratio in this study. A significant (P<0.01) positive linear relationship was observed between the EFs of CH 4 and CO 2 measured in the time intervals in the Zhujiang Tunnel ( figure 3), with an average CH 4 /CO 2 mass ratio of 40.61E-5±7.21E-5 g·g −1 .  The CH 4 /CO 2 ratio in this study is consistent with that of 17.09E-5 to 50.91E-5 obtained in the Nanjing Yangtze River Tunnel in 2014 [39] by using a portable greenhouse gas analyzer to measure the concentration gradient between the inlet and outlet of the tunnel. It is worth noting that gasoline vehicles dominated the traffic composition in the Yangtze River Tunnel during their sampling time. However, tests in the Islisberg Tunnel in Switzerland in 2011 revealed a CH 4 /CO 2 ratio of 1.67E-5±0.07E-5 g·g −1 [22], which is approximately 25 times lower than the average ratio from this study. Nam et al [17] reported a ratio of 15E-5±4E-5 g·g −1 for the onroad fleet in the US in 2003 based on chassis dynamometer tests, while Herndon et al [36] used a 'Bus Chase' method to obtain a CH 4 /CO 2 (g·g −1 ) ratio of 21.82E-5±7.27E-5 for diesel vehicles and 443.64E-5±65.45E-5 for compressed natural gas (CNG) vehicles. Except for the difference in vehicle size (bus to light-duty vehicle), the equipped oxidation catalyst in cars for emission control may be the main reason for the differences in EFs [36]. Hu et al [37] also reported a relatively higher CH 4 /CO 2 (g·g −1 ) ratio of 254.55E-5±130.91E-5 for natural gas vehicles (NGV) in China in 2015.

Vehicle CH 4 emissions estimate in China
In 2014, annual traffic CO 2 emissions in China were 820 Tg while annual traffic CH 4 emissions were estimated to be 79 Gg, and annual anthropogenic CH 4 emissions were 5357 Gg [14]. However, using the CH 4 /CO 2 mass ratio of 40.61E-5±7.21E-5 from this study and China's traffic CO 2 emissions of 820 Tg yr −1 in 2014, we could roughly obtain an estimate of 333 Gg yr −1 for traffic CH 4 emissions in China in 2014, which is approximately 4 times that reported value of 79 Gg and accounts for approximately 0.6% of the total anthropogenic CH 4 emissions (5357.2 Gg) in China. This percentage is above the upper limit of<0.5% estimated by Metz [18], and comparatively higher than the values of<0.2% estimated by Nam et al [17] and 0.20% compiled for America in 2014 [19]. With a GWP100 of 32 [5] and a CH 4 /CO 2 ratio of (40.61E-5±7.21E-5) g·g −1 for the on-road fleet, CH 4 can account for approximately 1.3% of traffic CO 2 -equivalent emissions from our study in China, which is significantly higher than the range of 0.3%-0.4% estimated by Nam et al [17] for global CH 4 emissions and the value of 0.08% estimated for the U.S. on-road fleet in 2014 [19]. In upcoming decades, such as in a 20-year horizon, the global warming potential of CH 4 is 84 times that of CO 2 [1], indicating an even greater contribution to climate warming.

Conclusion
As the CH 4 /CO 2 mass ratios measured for an urban on-road fleet in this study in the Pearl River Delta were consistent with those obtained in the Yangtze River Delta in 2014 [39], so the much higher CH 4 /CO 2 mass ratios might be common for vehicle emissions in China. Using the CH 4 /CO 2 mass ratios obtained in this study, we can roughly estimate that on-road vehicles in China might have contributed ∼330 Gg CH 4 (∼1 Tg CO 2 -equivalent) in 2014, or 0.6% of China's total anthropogenic CH 4 emissions. Our results indicate that improving energy efficiency would have co-benefits for reducing traffic emissions of CH 4 , as the observed EFs of CH 4 are positively correlated with those of CO 2 . On the other hand, if the traffic CH 4 /CO 2 ratio can be lowered to the level of 1.67E-5, as observed in the Islisberg Tunnel [22], over 90% of traffic CH 4 emissions in China could be tempered, suggesting a large potential or space to reduce China's CH 4 emissions in the transport sector.
Why were the EFs of CH 4 or CH 4 /CO 2 mass ratios so high for on-road vehicle fleet in China?The answer is not clear. However, fuel quality, engine performance, exhaust after treatment facilities and/or maintenance might be among the factors inducing more CH 4 emissions. Our previous study in the same tunnel revealed that the global warming potentials associated with refrigerant leakage from on-road vehicles are equal to 1.4% of that of the directly emitted CO 2 [44], and now, we have demonstrated that the emissions of CH 4 from on-road vehicles are equal to 1.3% of the directly emitted CO 2 in terms of the global warming potentials. These findings raises concern about whether we need to consider overall environmental and climatic effects of traffic emissions in formulating emission control policies, such as the emissions of greenhouse gases (CO 2 , CH 4 and halocarbons) or secondary formation of ozone, which is also a greenhouse gas that can be formed from photochemical ageing of traffic-emitted VOCs and NO x .