Study of atmospheric CH4 mole fractions at three WMO/GAW stations in China



[1] Atmospheric CH4 mole fractions were continuously measured from 2009 to 2011 at three WMO/GAW stations in China (Lin'an, LAN; Longfengshan, LFS; and Waliguan, WLG) using three Cavity Ring Down Spectroscopy instruments. LAN and LFS are GAW regional measurement stations. LAN is located in China's most economically developed region, and LFS is in a rice production area (planting area > 40,000 km2). WLG is a global measurement station in remote northwest China. At LAN, high methane mole fractions are observed in all seasons. Surface winds from the northeast enhance CH4 values, with a maximum increase of 32 ± 15 ppb in summer. The peak to peak amplitude of the seasonal cycle is 77 ± 35 ppb. At LFS, the diurnal cycle amplitude is approximately constant throughout the year except summer, when a value of 196 ± 65 ppb is observed. CH4 values at LFS reach their peak in July, which is different from seasonal variations typically observed in the northern hemisphere. CH4 mole fractions at WLG show both the smallest values and the lowest variability. Maximum values occur during summer, which is different from other northern hemisphere WMO/GAW global stations. The seasonal cycle amplitude is 17 ± 11 ppb. The linear growth rates at LAN, LFS, and WLG are 8.0 ± 1.2, 7.9 ± 0.9, and 9.4 ± 0.2 ppb yr−1, respectively, which are all larger than the global mean over the same 3 year period. Results from this study attempt to improve our basic understanding of observed atmospheric CH4 in China.

1 Introduction

[2] Methane plays an important role in the Earth's radiative balance and atmospheric chemistry. The calculated radiative forcing due to methane accounts for about 18% of the total value of long-lived greenhouse gases, which is second only to CO2 in importance [AGGI, 2011]. With the continued development and expansion of human activities, the atmospheric CH4 mole fraction has increased more than 1000 ppb (parts per billion, 10−9 moles of CH4 per 1 mole of dry air) since the beginning of the industrial era. CH4 is emitted into the atmosphere by both anthropogenic and natural sources. A total of 60–70% of methane emissions are from anthropogenic sources, such as rice agriculture, ruminants, fossil fuel exploitation, landfills, and biomass burning (both natural and human induced) [Denman et al., 2007]. The destruction of CH4 by OH in the troposphere is the main sink and accounts for 90% of CH4 loss in the atmosphere, making the atmospheric CH4 budget sensitive to OH changes [Bousquet et al., 2011; Denman et al., 2007; Vaghjiani and Ravishankara, 1991].

[3] High-precision measurements are used to constrain global and regional inverse modeling efforts by requiring that estimated sources and sinks be consistent with atmospheric measurements [Hein et al., 1997; Houweling et al., 2000]. Current observing networks for methane have been used to constrain emissions at global [Crevoisier et al., 2012; Dlugokencky et al., 2011] and national [Bergamaschi et al., 2005] scales. Several shortcomings including the space and time sparsity of observations on regional scales, the transport model errors, and the multiplicity of sources make it difficult to quantify the year to year variations in observed methane. For example, global atmospheric CH4 mole fractions were almost constant during 1999 to 2006 (total increase of ~6ppb over the 8 year period) and then increased 29 ppb over the 5 year period from 2007 to 2011 [Rigby et al., 2008; WMO Greenhouse Gas Bulletin, 2011, 2012]. The reasons for the renewed increase are not yet fully understood [Bousquet et al., 2011; Dlugokencky et al., 2009; Simpson et al., 2012]. This lack of understanding is due, in part, to inadequate coverage of atmospheric CH4 observations, especially in regions with large natural emissions (e.g., South America) and with rapid economic growth (e.g., Asia). To improve our understanding of the CH4 budget, it is imperative that we establish more extensive observing networks. With the rapid development of its economy, China has become the largest CO2 and CH4 emitter in the world [Gregg et al., 2008], although its emission per inhabitant remains much lower than previously developed countries [Tsutsumi et al., 2006]. Estimations of total CH4 emissions in China are variable and have large uncertainties. For example, Zhang and Chen [2010] report the total CH4 emissions by the Chinese economy in 2007 were 39.5 Tg. EDGAR [2011] reports a total emissions of 73 Tg in 2008, which almost doubles the value reported by Zhang and Chen.

[4] Previous short-term CH4 measurement campaigns and research programs in China have focused primarily on emissions from agricultural fields or urban areas [Cai et al., 2000; Wang et al., 2001; Wang et al., 2009; Zou et al., 2005]. Long-term atmospheric observations in China have been relatively sparse. In 1983 and 1991, the China Meteorological Administration (CMA) established atmospheric observing stations at Lin'an (LAN) in Zhejiang province and Longfengshan (LFS) in Heilongjiang province. Both stations have been designated as World Meteorological Organization (WMO) Global Atmosphere Watch (GAW) surface-based regional measurement stations. Initially, LAN and LFS were set up as automatic weather stations. It was not until 2009 that in situ CH4 measurement systems were installed at both sites. The Mt. Waliguan (WLG) station in Qinghai province is the first site in China to operate a long-term greenhouse gas sampling and measurement program. Since May 1991, CMA personnel have collected weekly air samples in glass flasks. These samples are shipped to the National Oceanic and Atmospheric Administration (NOAA) Earth System Research Laboratory (ESRL) in Boulder, Colorado, United States, and analyzed for a suite of greenhouse gases. In 1994, with support from the WMO/GAW program, WLG was established as a WMO/GAW surface-based global measurement station. In August 1994, an in situ CH4 measurement system (HP-5890, Hewlett-Packard, CA, USA) was installed at WLG with assistance from Environment Canada. This original instrument produced nearly 13 years of high quality data, which comprises the longest continuous atmospheric methane record in China [Zhou et al., 2004]. In 2008, the aging instrument was decommissioned. In January 2009, to improve and expand the greenhouse gas measurements in China, a Cavity Ring Down Spectroscopy instrument (G1301, Picarro, CA, USA) was installed at each of the three WMO/GAW stations. In this study, we present and analyze the first 3 years of measurements from these stations.

2 Experiment

2.1 Sampling Sites

[5] The three sampling stations are Lin'an (LAN), Longfengshan (LFS), and Mt. Waliguan (WLG). The locations of the stations are shown in Figure 1. The LAN is located in the center of the Yangzi River Delta area and is 50 km west of Hangzhou (the capital of Zhejiang province) and 200 km southwest of Shanghai (the largest economic center in China). The station is approximately 6 km northeast of Lin'an which has a population of ~100,000. Lin'an is a tourist town and has no industries with significant CH4 emissions. North of the LAN station (1.4 km away) is a small factory where activated charcoal is manufactured from burning bamboo. In addition, there is a very small village (200 inhabitants) 1 km to the north. The LFS station is 140 km southeast of Harbin city (the capital of Heilongjiang province). Wuchang, the nearest city, has ~200,000 inhabitants and is 40 km northwest of the site. The station is located on the northwest edge of the Longfengshan reservoir, which has an area of 20 km2. The reservoir's dam is located 0.1 km north of LFS. Beyond the dam to the north, there is a small area of paddy rice field and several small villages (of 100 inhabitants each). There are no big cities or large industries within a 40 km radius of the two regional stations. WLG station is situated in remote western China and relatively isolated from industrial and populated centers. Greenhouse gases measurements from WLG provides essential information on sources and sinks from within the Eurasian continent because of its unique location [Zhou et al., 2003, 2005]. Descriptions of the three WMO/GAW stations are listed in Table 1 [Liu et al., 2009].

Figure 1.

Location of the three WMO/GAW stations in China.

Table 1. Descriptions of the Three WMO/GAW Stations
StationStation IDLongitudeLatitudeAltitudeIntake HeightRepresentation AreaVegetation Canopy
(m asl)(m agl)
LongfengshanLFS127.6E44.73N330.510, 80Northeastern PlainPaddy, Forest
Lin'anLAN119.44E30.18N138.610, 50Yangtze River Delta Economic ZonePaddy, Wheat field, Shrub
Mt. WaliguanWLG100.90E36.29N381680Northeastern Tibetan PlateauAlpine pasture, Plateau

[6] At WLG station, the air sampling inlet is fixed at 80 m on an 89 m sampling tower, located 15 m from the lab. At LFS and LAN, in 2009, the sampling inlets are fixed at the top of wind poles (10 m above the ground) located in the observation fields. Near each sampling port, a wind direction and speed sensor are also installed. The distances from the observation fields to the labs are 65 and 25 m for LFS and LAN, respectively, which ensure that collected air samples are minimally affected by human activities at the lab. In 2010, new sampling towers were set up at LAN (50 m) and LFS (80 m). At that time, another sampling port was installed at the top level of each tower. The Picarro system then switched between the 10 m and the top level intake every 5 min at the two stations. Because of the short records from the highest level at LAN and LFS, we mainly discuss the results from the 10 m intake.

2.2 Measurement System

[7] A Cavity Ring Down Spectroscopy system (G1301, Picarro Inc.) is used for continuous measurement of atmospheric methane. This instrument has been proven to be an excellent solution for making precise measurements of atmospheric methane since its response is both highly linear and very stable [Crosson, 2008]. Schematic of the analytical system is illustrated in Figure 2. Air samples are delivered to the instrument at approximately 5 L min−1 by a vacuum pump (UN022, KNF Neuberger, Freiburg-Munzingen, Germany) via a dedicated 10 mm o.d. sampling line (Synflex 1300 tubing, Eaton, OH, USA). The ambient air first passes through a 7 µm stainless steel membrane filter (SS-4F-7, Swagelok, OH, USA) located upstream of the pump and then (after the pump) is bypassed by a pressure relief valve (RL3, Swagelok, OH, USA) set at 1 atm gauge pressure to release excess air pressure. It is well understood that the water vapor in the sample can affect the results of CH4 measurement via dilution and spectroscopic effects [Chen et al., 2010; Rella et al., 2012]. To reduce the interference of water vapor on the CH4 results, the ambient air is dried to a dew point of approximately −60°C by passing it through a glass trap submerged into a −70°C methanol bath (MC480D1, SP Industries, PA, USA). The outflow gas stream from the trap is set to 300 mL min−1 by a mass flow controller (1179A, MKS, MA, USA). The flow rate of G1301 is about 240 mL min−1. Excess gas is vented to ambient using a stainless steel “T” type three-way connector, which ensures that the sample fed to G1301 is at near-ambient pressure. The residence time for sample air running through the system at each station is less than 60 s.

Figure 2.

Schematic of the in-situ system. WH: Working High standard, WL: Working Low standard, T: Target gas.

[8] Methane mole fractions are referenced to a Working High standard (WH) and a Working Low standard (WL). Additionally, a calibrated cylinder filled with compressed ambient air is used as a Target gas (T) to regularly check the precision and stability of the system. All standard gases are pressurized in 29.5 L treated aluminum alloy cylinders (150A, Scott-Marrin, CA, USA) fitted with high-purity, two-stage gas regulators (CGA-590, Scott Specialty Gases, PA, USA). The standard gases are linked to the WMO CH4 standard (NOAA 04 scale) maintained by NOAA ESRL [Dlugokencky et al., 2005]. The automated sampling module equipped with an eight-port multi-position valve (EMT2SD8MWE, Valco, TX, USA) to deliver various gas sources (ambient and standards) to the instrument. The two standards and the target cylinder are analyzed by the system every 12 h. The analytical precision of the measurement systems are better than 0.01% of reading based on the continuous analysis of standard gases, and the accuracies are better than 0.1%. To further evaluate the performance of the analytical system, CMA participated in the methane reference gas inter-comparison for Asia from 2008 to 2009 and 2010 to 2011 sponsored by the WMO World Calibration Centre (WCC, Japan Meteorological Agency). Comparison results show that the WCC results and the CMA measurements by CRDS-based system are compatible to within 2.1 ppb [JMA, 2012].

[9] The discrete samples collected weekly at WLG have been measured by Carbon Cycle Greenhouse Gases group (CCGG) of NOAA ESRL in Boulder, CO, USA. Two samples were collected in series using glass flasks and a portable battery powered sampling apparatus with a 5 m (agl) intake height [Dlugokencky et al., 1994]. The samples were measured at NOAA using an Agilent 6890 GC, with a repeatability of approximately 1 ppb. All of the CH4 measurements were referenced to the WMO CH4 scale. The co-located NOAA flask program at WLG ensures that the long-term CMA in-situ measurements can be routinely compared with an independent record.

2.3 Data Processing

[10] Because of dead volume in the plumbing, the reported mole fractions were not stable until 2 min after switching between different gas streams. Every standard gas cylinder was connected to the system for 5 min in a sequence. Consequently, the raw ambient data were separated into 5 min segments. The data processing routine used the last 3 min of each 5 min segments to compute the methane mole fractions. Ambient measurements were calibrated using a linear two-point fit through the most recent standard gases measurements (WH and WL). Ambient CH4 values were accepted for analysis only during periods when the measurement of the most recent target gas (T) using this same calibration procedure was within 2 ppb of its assigned value. More than 98% of all data points met this criterion for the three stations.

[11] After computing mole fractions, the data were manually inspected. Occasionally, information in the station logbook identified periods when measurements were affected by analytical or sampling problems including poor instrument performance or local influences such as nearby fires, vehicles, and cattle. More than 97% of the data from each station remained after this step. All 5 min data were then combined into hourly averages. To evaluate the seasonal cycle and peak to peak amplitude, we adopted the same method used by Zhou et al. [2004]. The mean annual CH4 cycle was calculated based on monthly mole fractions. To obtain each monthly mean value, we subtracted a 12 month running mean and then averaged all the January values, February values, and so on. The annual mean mole fractions reported in this paper include their standard deviations. Other average values are reported with 95% confidence intervals (CI).

3 Results and Discussion

3.1 Methane Data

[12] Figure 3 shows the valid hourly CH4 mole fractions (after manually inspection) at LAN, LFS, and WLG. The observed results from the top level of the sampling tower at LAN and LFS are also presented (LAN station began to monitor from the top level from 15 July 2010, LFS station began to monitor from the top level from 10 August 2010). Valid weekly flask sample measurements from WLG during the period of this study are also displayed. Data gaps in 2011 at LFS and WLG are mainly due to the malfunctions of the instruments.

Figure 3.

Seasonal variations of hourly CH4 mole fractions at the three stations. LAN, LFS and WLG represent Lin'an station, Longfengshan station and Mt. Waliguan station, respectively. (a) Dots represent the hourly values from 10 m at LAN. Open circles represent the hourly data from 50 m. (b) Dots represent the hourly data from 10 m at LFS. Open circles represent values from 80 m. (c) Dots represent the hourly data from 80 m at WLG. Open circles denote discrete CH4 flask measurements from the NOAA ESRL flask air-sampling program.

[13] Figure 4 shows the difference between the discrete flask sample measurements made by NOAA ESRL and the corresponding hourly CH4 values observed by the CMA CRDS instrument at WLG. During most of the time, differences between the two systems vary by ± 4 ppb. The mean difference (flask minus in situ) is 2.0 ± 0.6 ppb (95% CI, the same as below) for 170 flask sampling events. This result suggests that NOAA flask and CMA in situ measurements at WLG are compatible to within ± 2 ppb, which meet the recommendations outlined by WMO/GAW [Brailsford, 2012]. Differences between the two data sets are likely due to the sampling methods. The NOAA flask sampler collects air samples almost instantaneously (sampling period < 1 min), whereas the G1301 system generally has twelve 5 min data points in one hour. The fluctuation of atmospheric CH4 mole fractions during the 1 h time period obviously contributes to the difference. Moreover, the different sampling intake heights may also contribute to the observed differences. A relatively positive tail of the distribution (Figure 4) indicates that occasionally the flask samples might be affected by very local sources at WLG.

Figure 4.

Comparison between the flask sample values and G1301 measurements at Waliguan from 2009 to 2011. The distribution represents the difference between the NOAA/ESRL flask results and the corresponding G1301 hourly values (340 data points).

3.2 Average Diurnal Variations

[14] Mean diurnal CH4 variations in April, July, October, and January are used to represent the average variations in spring, summer, autumn, and winter. Only days containing twenty-four 1 h average values are considered because the day to day variations of CH4 can be quite large. The mean diurnal variations are shown in Figure 5. Diurnal patterns at LAN are similar in the four seasons with the highest value occurring at 5–7 local time (LT) and the lowest at 14 LT. These variations are consistent with the diurnal cycle of the boundary layer that reaches its maximum height near the middle of the day. At night, radiation loss at the ground level leads to a shallow stable boundary layer [Worthy et al., 1998]. The peak to peak amplitude of the diurnal cycle is 45 ± 26 ppb in spring, 95 ± 59 ppb in summer, 37 ± 27 ppb in autumn, and 27 ± 26 ppb in winter and is due, in part, to the strong industrial and agricultural sources in the Yangtze Delta area. The large amplitude in summer may also be ascribed to the relatively deep boundary layer and the high temperatures which stimulate greater emissions from the agricultural area and larger consumption by OH radicals. Worthy et al. [1998] also observed a diurnal cycle with amplitude of ~150 ppb during summer when CH4 emissions from a local wetland at Fraserdale, Canada, were very strong.

Figure 5.

Mean diurnal variations of CH4 mole fractions in January, April, July and October at the three stations. Error bars indicate 95% confidence intervals.

[15] Similarly large CH4 diurnal variations at LFS are also observed in summer. The diurnal amplitude is 196 ± 65 ppb, which is the largest among the three stations. It is mainly due to the high CH4 emissions from waterlogged paddy rice fields throughout summer [Cai et al., 2000]. The LFS station is located in the Heilongjiang province, which is the most important rice production area (rice planting area > 40,000 km2) in China. Dramatically smaller diurnal cycles are observed in the other seasons in which rice production is not active. The daily amplitudes are less than 27 ± 18 ppb, which suggests that atmospheric CH4 levels are relatively stable. In spring and autumn, daily CH4 variations are similar with the lowest value occurring from 14 to 16 LT. These trends are also consistent with the change in the boundary layer height suggesting that the CH4 mole fraction at the two regional stations (LFS and LAN) is affected by local sources.

[16] Variability in the daily CH4 mole fractions at WMO/GAW global stations is generally small and sometimes diurnal variations cannot be observed [Aoki et al., 1992]. At WLG, similar results are observed. The observed daily amplitudes are small with values of 5 ± 5 ppb in spring, 7 ± 6 ppb in summer, 10 ± 9 in autumn, and 7 ± 6 ppb in winter, which are not statistically significant and similar to the observed results presented by Zhou et al. [2004]. The small diurnal patterns indicate the weak local sources at this station.

3.3 Comparison of CH4 Mole Fractions Between Different Levels

[17] To understand the effect of local sources on the observed mole fractions, the hourly differences between 10 m and the top levels (50 m for LAN and 80 m for LFS) are shown in Figure 6. Note that the results are hourly mole fractions from 10 m minus those from the top level. Generally, CH4 mole fractions at 10 m are higher than the top levels, which indicate that the measurements at 10 m are affected by local sources. The differences show obvious diurnal variations with larger values at night and smaller values during daytime, which is also consistent with boundary layer variations. In daytime, the boundary layer extends higher and vertical mixing is rapid. CH4 emitted at or near the surface is rapidly mixed throughout the boundary layer. At night, an opposite situation occurs, with the stable boundary layer trapping surface emissions, and the difference are generally larger. The maximum differences are 7 ± 1 ppb at 2–6 LT for LAN and 43 ± 11 ppb at 1–3 LT for LFS. The difference at LFS is larger than LAN which is due, in part, to strong sources near the LFS station. It is also possible that the shorter tower at LAN contributes to the observed difference. During midday, differences at the two regional stations are small and relatively stable. At LAN, the differences are smaller than 2 ppb (from 0.7 to 1.9 ppb) from 9 to 17 LT. At LFS, differences are smaller than 5 ppb (from 3.1 to 4.4 ppb) from 9 to 15 LT. These small differences are indicative of the degree to which mixing occurs in the boundary layer. Given the diurnal variations of CH4 at the two regional stations, the observed data during these periods should be the least influenced by local sources.

Figure 6.

Diurnal differences of hourly CH4 mole fractions between the 10 m and the top level at the two regional stations. (a): LAN station, 10 m results minus 50 m values. (b): LFS station, 10 m results minus 80 m values. Error bars denote 95% confidence intervals.

3.4 Impact of Local Surface Wind

[18] It has been suggested that the studying of CH4 mole fractions in conjunction with meteorological data could help to understand CH4 emissions and transport [Dlugokencky et al., 1995]. To understand the influence of local surface wind on the observed CH4 mole fractions, in Figures 7-9, we plot wind rose distributions for the mean CH4 values observed from different wind sectors in the four seasons (spring: March to May; summer: June to August, autumn; September to November; winter: October to February).

Figure 7.

Seasonal hourly CH4 mole fractions (ppb) from the 16 horizontal wind directions at LAN. The line segments across the data points (error bars) on each direction represent 95% confidence intervals. CH4 mole fractions from N-NNE-NE-ENE-E-ESE and from SSE sectors in spring, from the N-NNE-NE-ENE-E-ESE sectors in summer, from the N-NNE-NE-ENE-E-ESE-SE sectors in autumn, and from the NNE-NE-ENE-E-ESE-SE, and S sectors in winter are excluded as local events.

Figure 8.

Seasonal hourly CH4 mole fractions (ppb) from the 16 horizontal wind directions at LFS. The line segments across the data points (error bars) on each direction represent 95% confidence intervals. CH4 mole fractions from the N, SSW-SW-WSW-W and SE sectors in spring, from the E-ESE, SSE, and NNW-N sectors in summer, from the N, E, SSE, and WSW-SW sectors in autumn and from the NNW-N sectors in winter are excluded as local events.

Figure 9.

Seasonal hourly CH4 mole fractions (ppb) from the 16 horizontal wind directions at WLG. The line segments across the data points (error bars) on each direction represent 95% confidence intervals. CH4 mole fractions from the NNE-NE-ENE-E-ESE sectors are all excluded as local events in all seasons.

[19] At LAN, winds from the N-NNE-NE-ENE-E-ESE-SE sectors result in higher CH4 mole fractions in all seasons. Especially in summer, the maximum enhancement relative to the seasonal average (2001 ± 3 ppb) is 32 ± 15 ppb when winds are from the NE sector. This is mainly because surface winds bring large amounts of CH4 emitted from Shanghai and Hangzhou which are northeast of LAN. The average values from the north are 1975 ± 5 ppb in spring, 2024 ± 5 ppb in summer, and 2025 ± 7 ppb in autumn, which are significantly higher than the seasonal averages (1969 ± 1 ppb in spring, 2001 ± 3 ppb in summer, and 2005 ± 2 ppb in autumn). This observation may be due to the activated charcoal factory located 1.4 km to the north of the station. In winter, the CH4 enhancement from the N sector is not obvious because the factory is generally shut down during this period. CH4 mole fractions from the S sector are also 5 ± 5 ppb higher than the average value (1986 ± 2 ppb), which may be due to the emissions from Lin'an (in winter, most of inhabitants burn coal to heat their homes). During the four seasons, CH4 mole fractions from the W-…-SSW sectors are lower than the seasonal averages and are less affected by local sources. The lowest mole fractions are from the WSW sector with values of 1952 ± 6 ppb in spring, 1949 ± 8 ppb in summer, 1986 ± 10 ppb in autumn, and 1968 ± 6 ppb in winter.

[20] At LFS, wind rose distributions display different patterns in the four seasons. In the spring, winds originating from the N, W-WSW-SW-SSW, and SE sectors bring higher CH4 mole fractions (Figure 8). The maximum enhancement is from the SSW sector with a value of 9 ± 4 ppb relative to the seasonal average (1918 ± 1 ppb). In summer, CH4 mole fractions from the SSE, ESE-E, and NNW-N sectors are higher than the other sectors by 0 ~ 146 ppb. The Longfengshan reservoir, located to the E-ESE-SE-SSE of the station, is a strong CH4 source [IPCC, 2001] and contributes to the observed maximum enhancement of 92 ± 29 ppb relative to the seasonal average (2020 ± 4 ppb) from the E sector. Because of the small area of paddy rice fields and small villages to the north of LFS, the average values from the N sector are 1924 ± 3 ppb in spring, 2037 ± 15 ppb in summer, 1949 ± 6 ppb in autumn, and 2026 ± 16 ppb in winter, which are higher than the respective seasonal averages (1918 ± 1 ppb in spring, 2020 ± 4 ppb in summer, 1941 ± 1 ppb in autumn and 1981 ± 2 ppb in winter). Thus air from the NNW-N sectors is potentially polluted and the observed results do not likely represent a large mixed volume. In autumn, in addition to winds from the north, surface winds from the E, SSE, and WSW-SW sectors also induce higher CH4 mole fractions. Maximum enhancements are from the E and SSE sectors with values of 15 ± 10 ppb. In winter, anthropogenic activities (consumption of coal for heating etc.) may be the main reason for higher CH4 mole fractions. CH4 values are 2026 ± 16 ppb from the N and 2020 ± 11 ppb from NNW sector, which are much higher than the other sectors. However, from the ENE-E-ESE-SE-SSE-S sectors, where the Longfengshan reservoir is located, CH4 mole fractions are lower with the lowest value of 1958 ± 6 ppb from the E sector. This is because the reservoir is frozen during winter and there is little or no CH4 emission from these sectors.

[21] A previous study of WLG showed that CH4 mole fractions were from 10 to 20 ppb higher than the weighted average mole fractions when surface winds were from the ENE-E-ESE-SE sectors [Zhou et al., 2004]. Similar patterns are also observed from the NNE-NE-ENE-E-ESE sectors during all seasons with values ranging from 1 to 25 ppb higher than the seasonal averages (Figure 9). These sectors contain more highland barley plantations and higher population density. Furthermore, the capital of Qinghai province (Xining city) is located 150 km to the northeast and may also contribute to the higher CH4 values. It is interesting to note that the maximum mole fractions are all from the NE sector in all seasons with values of 1862 ± 2 ppb in spring, 1874 ± 2 ppb in summer, 1887 ± 4 ppb in autumn, and 1880 ± 7 ppb in winter, which are significant higher than the seasonal averages (1852 ± 1 ppb in spring, 1866 ± 1 ppb in summer, 1867 ± 1 ppb in autumn and 1854 ± 1 ppb in winter). Tang et al. [1999] studied the relationship of black carbon (BC) at WLG with surface winds and long-range transport and showed that the highest BC concentrations occur with air mass originating from the NNE and NE sectors. They concluded the BC enhancements were due to emissions from the Yellow River Canyon industrial area northeast of Waliguan more than 500 km away. According to their study, the higher CH4 values from these sectors may also be attributed to emissions within this area.

[22] Hourly CH4 mole fractions during the three years were also calculated under different surface wind scales (Beaufort wind scale): 0 scale: <0.3 m s−1, first scale: 0.3 to 1.5 m s−1, second scale: 1.6 to 3.3 m s−1, third scale: 3.4 to 5.4 m s−1, fourth scale: 5.5 to 7.9 m s−1, fifth scale: 8.0 to 10.7 m s−1, sixth scale: 10.8 to 13.8 m s−1, seventh scale: 13.9 to 17.1 m s−1. Results show that observed CH4 mole fractions are strongly influenced by the wind force at the three stations. At LAN and LFS, higher wind speeds result in lower CH4 values during all seasons. During calm conditions, average CH4 values are 2007 ± 5 (LAN) and 1980 ± 10 ppb (LFS). When the wind force is on the fifth scale, the average CH4 mole fractions are reduced to 1984 ± 14 at LAN and 1918 ± 7 ppb at LFS. These results also indicate that higher wind speeds result in more diluted CH4 mole fractions. At WLG, in spring, autumn and winter, lower CH4 levels are also accompanied by greater surface wind speeds. In summer, the opposite is observed. The methane values increase with enhanced wind force. The average CH4 mole fraction is 1852 ± 5 ppb during calm conditions (zeroth scale) and is 1869 ± 11 ppb at the seventh scale. Although the larger wind speed may dilute CH4 from very local sources at WLG, the higher wind speed may also transport emissions from the broader region, which may overwhelm the effect of dilution resulting in the higher observed CH4 mole fractions.

3.5 Influence of Regional Transport in Summer

[23] In summer, the large amplitude of the diurnal cycle at the two regional stations suggests that the observed CH4 mole fractions are strongly affected by local sources. To understand the contribution of regional transport on observed CH4 values, we computed 3-day backward trajectories coincident with hourly methane mole fractions using the Hybrid Single - Particle Lagrangian Integrated Trajectory (HYSPLIT) dispersion model [Draxler and Rolph, 2003]. The model is based upon NCEP/NCAR reanalysis data. The trajectories were calculated for every hour (1, 2, 3 LT …) in summer. Because of the data gap at LFS in the summer of 2011, only trajectories from 2009 to 2010 were calculated for all the three stations. The analysis results are shown in Figure 10.

Figure 10.

Cluster analysis of 72 hrs backward trajectories for every hour (0, 1, 2, 3 LT …) in the summer and the corresponding average CH4 mole fractions for LAN, LFS and WLG. The CH4 mole fraction with confidence interval on each cluster is the average value from all the trajectories in the cluster.

[24] There are 3 main clusters for LAN. Clusters 1 and 2 account for 73% of the total trajectories; the average transport speeds (107 ± 7 km day−1 for cluster 1 and 75 ± 7 km day−1 for cluster 2) are faster than cluster 3 (20 ± 3 km day−1). Cluster 1 comes from the east and the average mole fraction is 1976 ± 7 ppb, which is smaller than the average values (1991 ± 5 ppb) in summer. As discussed above, the average CH4 mole fraction is higher (> 2021 ± 21 ppb) when surface winds are from NE-ENE-E sectors. The difference suggests that higher values from these sectors are mainly due to transport from very local sources. Cluster 3 moves slowly and CH4 accumulates in the local area. Higher values are also observed when surface winds are from the NNW-N sectors (Figure 7). Thus at LAN, local sources are the leading factor affecting the observed CH4 values in summer.

[25] The three main clusters at LFS pass through cities and agricultural areas. Besides the very local sources, regional transport in summer may also contribute to the high CH4 values. The third cluster at LFS moves slowly (25 ± 2 km day−1) and the average mole fraction (2089 ± 15 ppb) is the highest among the three clusters. This is due to the strong emissions from paddy fields and forest in the regional area.

[26] There are two main clusters for WLG. Cluster 1 account for 81% of the total trajectories. All of the trajectories in cluster 1 are from the NE-ENE-E-ESE-SE sectors. Since the mean trajectory passes through Xining and Lanzhou (capital of Gansu province), which have strong industrial CH4 sources, the average CH4 mole fraction is 5 ± 2 ppb higher than the average value in summer (1866 ± 1 ppb). Considered with the increase of CH4 mole fractions when surface winds are from the NE-ENE-E-ESE sectors, transports of regional and local emissions may all contribute to the higher CH4 mole fractions. Cluster 2 transects the northwest, which is a poorly developed area consisting primarily of sand dunes. Air mass dilutes the local CH4 mole fractions to produces a lower mean value of 1856 ± 1 ppb.

3.6 Evaluation of “Regional” Mole Fractions

[27] As discussed above, the observed CH4 mole fractions at LFS and LAN are strongly affected by local sources. To understand the CH4 level at a larger scale, we filter the data into “regional” and “local” events. The “regional” events represent CH4 mole fractions in a large scale (> 10 km2), while the “local” events represent values in surrounding area of the station (≤ 10 km2). Comparing measurements from the 10 m intake and the top level of the sampling tower shows that during midday, atmospheric CH4 is the least influenced by local sources and the differences between the two levels are at a minimum. The analysis of the wind rose distribution pattern further indicates the main CH4 pollution sectors at each station. Thus two steps are used to subset the observed data into “regional” events.

[28] At LAN, results from 0 to 8 LT and 18 to 23 LT are not considered. At LFS, data from 0 to 8 LT and from 16 to 23 LT are excluded. Local meteorological data are then used to filter the remaining data. Based on the discussion in section 3.4, CH4 values influenced by local events (where observed CH4 mole fractions are higher than the seasonal average) are also excluded. At LAN, local events occur when winds are from the N-NNE-NE-ENE-E-ESE sectors, and SSE sector in spring, from the N-NNE-NE-ENE-E-ESE sectors in summer, from the N-NNE-NE-ENE-E-ESE-SE sectors in autumn, and from the NNE-NE-ENE-E-ESE-SE sectors, and the S sector in winter (Figure 7). At LFS, local events occur when winds are from the N and SSW-SW-WSW-W sectors and from the SE sector in spring, from the E-ESE, SSE, and NNW-N sectors in summer, from the N, E, SSE, and WSW-SW sectors in autumn and from the NNW-N sectors in winter (Figure 8). The remaining data are considered to be minimally influenced by local sources. For LAN and LFS, 16% and 19% of hourly CH4 mole fractions are considered to be regionally representative.

[29] At WLG, we use criteria similar to Zhou et al. [2004] to filter the observed data into “background” events. In this study, hourly CH4 values at WLG associated with surface wind originating from the NNE-NE-ENE-E-ESE sectors are excluded. In summer, measurements when wind speeds are ≥ 10 m s−1 are excluded. Finally, measurements exceeding 3 standard deviations from a curve fitted to all valid hourly averaged data are excluded to eliminate instrument noise. It should be noted that Zhou et al. [2004] used vertical wind speed to further filter the data (~5% of the total data are flagged). Vertical wind speed sensor data were not available during the period of this study. Approximately 53% of the 2009 ~ 2011 data are categorized as “background”, which is similar to the result (~50%) obtained by Zhou et al. [2004].

3.7 Long-Term Trends

[30] Linear fits to all “regional” or “background” hourly data from 2009 to 2011 are used to roughly estimate the long-term trends for each station. The calculated trends in regional/background CH4 mole fractions at the three stations are all positive during the observation period and are 8.0 ± 1.2 (R = 0.11, P < 0.05), 7.9 ± 0.9 (R = 0.14, P < 0.05), and 9.4 ± 0.2 (R = 0.54, P < 0.05) ppb yr−1 for LAN, LFS and WLG, respectively. The WMO Greenhouse Gas Bulletin, 2010, 2011, 2012 report the global atmospheric CH4 is increasing at a rate of ~ 5 ppb yr−1 from 2009 to 2011. The rates of increase observed at the three stations in China are obviously higher (≥ 2.9 ± 0.9 ppb) than the global average. This is because China is a net source of global CH4 budget. It should be noted that only three years of data are used to calculate the growth rate. The relatively short period of measurements may result in inaccurate estimates of the observed trends.

3.8 Seasonal Cycle

[31] Figure 11 shows the monthly CH4 mole fractions during the observation period at the three stations. For comparison, we also overlay the monthly values from Mauna Loa in United States (MLO), Jungfraujoch in Switzerland (JFJ), Anmyeon-do (AMY) in the Republic of Korea, and Ryori in Japan (RYO). MLO and JFJ are WMO GAW global measurement stations. Both AMY and RYO are GAW regional measurement stations. In the northern hemisphere, atmospheric methane mole fraction generally reaches a minimum in summer and a maximum in late fall or mid-winter. This is due to the destruction of methane by OH radicals in troposphere, which is the strongest in summer. The observed atmospheric CH4 levels are due to a combination of factors at the local, regional, and global scales [Bousquet et al., 2011; Denman et al., 2007; Warwick et al., 2002]. As a result, observed CH4 generally displays complicated variations and trends [Frankenberg et al., 2006; Houweling et al., 2000]. At very remote sites such as MLO, with very few local sources, this trend is simple and can be easily captured. However, the observations in this study were all made in the Chinese inland, and is thus influenced by a variety of sources such as natural wetlands, agricultural fields (e.g., rice, corn, winter-wheat), landfills, coal, fuel, and animals.

Figure 11.

Average monthly CH4 mole fractions at the three stations and other WMO/GAW stations from 2009 to 2011. RYO: Ryori in Japan, AMY: Anmyeon-do in Republic of Korea, MLO: Mauna Loa in the United States (Hawaii), JFJ: Jungfraujoch in Switzerland. Error bars denotethe standard deviation of monthly means from 2009 to 2011.

[32] At LAN, regional CH4 is at a minimum in July and a maximum in December, which roughly consists with most sites in northern hemisphere (such as JFJ, MLO, and RYO). This variation is mainly due to the abundant OH radicals in the Yangtze Delta area in summer (generated by the precursors such as ozone, CO and NOX) [Cheung and Wang, 2001] and the sparse OH radicals inducing CH4 accumulation in winter. The highest CH4 mole fractions at LFS occur in January, which is mainly due to the consumption of coal for heating and biomass burning which tends to begin in October in the northeastern China plain. In addition, the daytime planetary boundary layer and the night time surface layer are both at lower height in the winter, leading to significant signal enhancements for the same emissions levels. The CH4 levels at LFS decrease after January and start to increase in May, which is likely caused by the very strong CH4 emissions from the waterlogged paddy rice fields started from May [Yue et al., 2005]. These emissions are highest from July to August. Consequently, atmospheric CH4 reaches a peak value in July (1961 ± 6 ppb). The seasonal variation at LFS shows a “W” pattern with two peaks (July and January) and differs from the typical pattern observed in the northern hemisphere.

[33] It is interesting to point out that although LAN and LFS are both located in a highly agricultural active areas, the CH4 mole fractions in summer display opposite trends (with valley for LAN and peak for LFS). LAN is located in the Yangtze Delta area, which is the most economically developed region in China. There are many strong industrial CH4 sources here [Wang et al., 2009]. Additionally, the area is a highly productive region of paddy rice and winter-wheat in China [Yan et al., 2003]. Because LAN is at a low latitude, the sunshine duration and temperature (yearly average: 14 °C) are higher than LFS and WLG (yearly average are 5 °C for LFS and - 1 °C for WLG). As a result, emissions from both industries and terrestrial ecosystem are strong during the four seasons. The diurnal variability and large amplitudes during all seasons at LAN also indicates that emissions are active during the entire year. The relatively smaller CH4 mole fraction in summer is consistent with variation in the northern hemisphere. In contrast, the LFS station is at a more northerly latitude with no strong industrial CH4 sources nearby. In summer, the diurnal variations are also very clear and the amplitudes are much larger than in other seasons, indicative of strong regional sources. Although in summer, consumption by OH radicals is the strongest in summer, methane emissions from the terrestrial ecosystem (paddy rice fields, forest etc.) at LFS may overweight the consumption and consequently lead a significant increase in “regional” CH4 mole fractions.

[34] Atmospheric CH4 variations at WLG differ from variations observed at other global stations such as MLO, and JFJ (Figure 11). Peak values at WLG occur from July to August, consistent with the findings of Zhou et al. [2004]. The relatively higher CH4 mole fractions during summer are probably due to emissions from ruminants, the weak sink of atmospheric CH4, and regional transports. Waliguan is on the Tibetan plateau (alpine pasture) and the ruminants' emissions should be the strongest in summer when their grass consumption is the greatest. Ma et al. [2002] showed that photochemical capacities in both free troposphere and boundary layer air mass conditions at Waliguan were very weak in summer. In addition, Xiong et al. [2009] concluded that the local maxima of water vapor, CH4 and CO in summer on the Tibetan plateau were likely driven by the dynamic transport of the monsoon. The CH4 emissions from the terrestrial ecosystems (and probably from regional transport) in this area may overweight the destruction by OH radicals and induce higher CH4 mole fractions. It can also be concluded that CH4 levels at WLG during this period are subject to CH4 sources, as well as subsequent regional transport. Table 2 summarizes the yearly average CH4 mole fractions from 2009 to 2011 at the three stations and other WMO/GAW stations. Yearly CH4 values at WLG are 1843 ± 2 ppb (1 σ, the same as the follows) in 2009 and 1853 ± 6 ppb 2010, which are lower than those at LAN and LFS by ~ 90 ppb. The observed seasonal amplitude at WLG is 17 ± 11 ppb. The seasonal amplitudes at LAN and LFS are 77 ± 35 and 73 ± 8 ppb, respectively. Annual averaged CH4 values at these two stations (> 1931 ± 17 ppb) are at least 123 ± 17 ppb higher than the global values (1808 ppb in 2010, 1813 ± 2 ppb in 2011) [WMO Greenhouse Gas Bulletin, 2011, 2012] and annual values from RYO, AMY, JFJ, and MLO (Table 2). The higher annual values at LAN and LFS stations are likely due to strong regional CH4 sources. Annual average values are also relatively high (> 1906 ± 8 ppb) at AMY in the Republic of Korea, which is influenced, in part, by cities in Korea and Chinese mainland.

Table 2. Yearly Average Mole Fractions and Peak to Peak CH4 Amplitude From the Background Data Sets
  • ---

    The years containing less than 12 - month data are not calculated.

  • a

    Seasonal amplitudes are calculated from the detrended data.

Altitude (m asl)13933138162604633973580
Annual mean in 2009 (ppb)1935 ± 131931 ± 171843 ± 21878 ± 81906 ± 81811 ± 5---
Annual mean in 2010 (ppb)1947 ± 171942 ± 171853 ± 61883 ± 101914 ± 141817 ± 61869 ± 7
Annual mean in 2011 (ppb)1961 ± 16---------1929 ± 91817 ± 6---
Seasonal amplitude (ppb)a77 ± 3573 ± 817 ± 1150 ± 1441 ± 1529 ± 629 ± 10

[35] Data gaps from March 26th to July 1st, 2011 at LFS, and from October 15th to December 31st, 2011 at WLG, may bias our estimates of the monthly seasonal means as well as the long-term trends. To evaluate the potential bias due these gaps, we created the same artificial gaps in all 3 data sets and repeated our analysis. We found no change in the seasonal patterns but the seasonal amplitude at LAN in 2011 was reduced to 64 ± 32 ppb.

4 Conclusions

[36] Atmospheric CH4 measurements made at three WMO/GAW stations have been reported. The diurnal patterns, surface wind effects, influence of regional transport, atmospheric trends, and seasonal variations at each site were discussed. The results indicate that CH4 mole fractions from 10 m above the ground at the two regional stations (LAN and LFS) are affected by local and regional sources, especially in summer. Measurements from WLG are representative of the Tibetan Plateau.

[37] The calculated regional/background CH4 growth rates are all higher than the global average. Methane levels at LAN are representative of the Yangtze Delta area, China and are the highest among the three stations. Variations of CH4 at LFS are large in summer, which is mainly due to the strong emissions from local terrestrial ecosystems (paddy rice field). The seasonal variations at WLG are different from other GAW global stations (such as MLO in United States, and JFJ in Switzerland) with peak values occurring in summer.

[38] In this study, we present atmospheric CH4 measurements from 2009 to 2011. The relatively short 3-year record may introduce bias in our analysis of the seasonal variations and estimates of trends. To better understand atmospheric CH4 in China, a more extensive CH4 observing network and longer period of in situ measurements is required. To this end, the CMA has already installed CH4 measurement systems at Shangdianzi near Beijing (40.65ºN, 117.12ºE) and Shangri-La in Yunnan province (27.48ºN, 99ºE). By the end of 2013, CH4 measurement systems will be installed at three new stations [Heyuan in Guangdong province (23.69ºN, 114.6ºE), Jinsha in Hubei province (29.63ºN, 114.2ºE), Akedala in Xinjiang province (47.1ºN, 87.97 º E)]. The eight CH4 observatories in the near future will help us to better understand the CH4 character and the emissions in China.


[39] We express our great thanks to the staff at Lin'an, Longfengshan and Waliguan station who have contributed to the system installation and maintenance at the stations. This work is supported by National Natural Science Foundation of China (No. 41175116), the National Key Basic Research Program (No. 2010CB950601), the International S&T Cooperation Program of the MOST (No. 2011DFA21090), and the CMA operational fund. The data used in this study from Chinese stations will be available for public within the China Meteorological Administration (CMA) policy.

[40] We also thank Ryori in Japan, Anmyeon-do in the Republic of Korea, Mauna Loa in the United States (Hawaii), and Jungfraujoch in Switzerland for providing monthly CH4 data from 2009 to 2011. The monthly data are downloaded from World Data Centre for Greenhouse Gases (WDCGG) and are permitted to be used by the data holder.