Observations and emission constraints of trichlorofluoromethane (CFC-11) in southeastern China: first-year results from a new AGAGE station

The recovery of the ozone layer relies on decreasing atmospheric mixing ratios of ozone-depleting substances (ODSs), including chlorofluorocarbons (CFCs). A significant decline in the mixing ratio of trichlorofluoromethane (CFC-11 or CCl3F ), the second most abundant CFC, has been observed since the mid-1990s. However, a slowdown in the decline after 2012 indicates a rise in emissions, particularly in Eastern Asia. Ground-based observations are lacking in southeastern China, limiting a thorough evaluation of CFC-11 levels and emissions in this region. A new Advanced Global Atmospheric Gases Experiment background station was established at Xichong (XCG), Shenzhen, China, to provide high-frequency continuous in situ observations. The annual mean CFC-11 mixing ratio, recorded from May 2022 to April 2023, is 221.64 ± 2.29 ppt. When compared with a monthly (MHD) or daily (MLO) observation, this value is found to be 0.45% to 5.36% higher than the northern hemispheric background. With the inverse modeling and interspecies correlation method, we estimate CFC-11 emissions in southeastern China between 1.23 ± 0.25 Gg yr−1 and 1.58 ± 0.21 Gg yr−1, in line with the bottom-up estimation of 1.50 Gg yr−1. Results indicate that CFC-11 emissions in the Pearl River Delta region have returned to levels before 2010, aligning with regional and global trends. Observations from XCG would compensate for the deficiency of CFC-11 measurements in southeastern China, paving the road for ODS studies in this region and beyond.


Introduction
Trichlorofluoromethane (CFC-11 or CFCl 3 ) is one of the most significant anthropogenic ozone-depleting substances (ODSs) in the atmosphere and a powerful greenhouse gas (GHG) [1,2].The global production and consumption of CFC-11 for dispersive uses were mandated to be phased out by 2010 in response to the Montreal Protocol and its amendments (MPA) [3].Before the phase-out, CFC-11 was used in various industrial applications, such as refrigerant in chillers, propellant gas in aerosol cans, and polyurethane foam blowing agents in buildings [4,5].As remaining CFC-11 in foams and chillers (banks of CFC-11) gradually leak into the atmosphere, a steady decline in CFC-11 emissions is expected to continue for decades.However, measurements of ambient CFC-11 levels remain insufficient, especially in China and its southeastern part [6].This study reports CFC-11 mixing ratios and estimates CFC-11 emissions considering new high-frequency continuous in situ observations from the first regional background station in southeastern China.
Contrary to the expectations of declining emissions from banks and effective implementation of the MPA agreement, recent studies reveal an unexpected increase in global and regional emissions of CFC-11 since 2013 [7,8].Inverse modeling [9,10] and tracerbased estimates [11,12] identify unexpected production and use of CFC-11 from eastern China, which contributes 64 ± 32% of the total global CFC-11 emission increases estimated by the Advanced Global Atmospheric Gases Experiment (AGAGE) and 41 ± 20% of that by the National Oceanic and Atmospheric Administration (NOAA).Other regions in China or other countries may be responsible for the remaining increase in emissions.[10].
Given the long atmospheric lifetime of CFC-11 (∼55 years [13]), new CFC-11 production and consumption significantly influence its decline rate of atmospheric levels, slowing down the recovery process of the ozone layer [14].Consequently, conducting in situ, observations in China are crucial to fill the observational gaps, especially in its fast-growing developing regions.Southeastern China has undergone significant economic and industrial growth over the past decades, thus holding great importance in investigating CFC-11 emissions.However, ambient levels of CFC-11 in southeastern China have been poorly quantified [12,[15][16][17][18][19][20][21].Previous studies allow emission estimates only based on short-term campaigns [18,22].For example, Lin et al (2019) [22] measured elevated CFC-11 at a remote mountaintop station from August to November 2017.
Here, This study reports CFC-11 mixing ratios in southeastern China based on the quasi-continuous high-frequency in situ observations between May 2022 and April 2023 from the Xichong station (hereafter referred to as XCG), a new AGAGE station in Shenzhen, China.Emissions of CFC-11 from southeastern China are estimated based on Bayesian inverse modeling and the interspecies correlation (ISC) method.This study enhances the understanding of temporal variations of CFC-11, focusing on the emission hotspots in a region characterized by rapid economic and industrial development.In situ observations at XCG largely compensate for the previous lack of CFC-11 observations in southeastern China, thus providing a unique data set for future investigations of halogenated gases.

XCG station and study domain
The station, XCG is located at 22.48 • N, 114.56 • E and an elevation of 137 m above sea level, is one of the atmospheric regional background stations in the AGAGE network (http://agage.mit.edu/stations/xichong, last accessed on Mar 22, 2024 [23]).The station is on the Dapeng Peninsula in Shenzhen in the southern Pearl River Delta (PRD) region, Guangdong, China.XCG is situated away from industrial or densely populated areas to minimize influence from nearby sources.

Experimental instruments
High-frequency continuous in situ observations are performed with the custom-made 'ODS5-pro' gas chromatographic/mass spectrometry (GC-MS) system (Beijing Huanaco Innovation Co., Ltd) installed in October 2021.ODS5-pro comprises an automatic sampling module, an analysis system, standard gases, auxiliary gases (helium and nitrogen), and Sensitivities, i.e., how sensitive the observations at XCG are to emissions from each 0.5 • × 0.5 • grid, are from the lower 100 m of the FLEXPART model output [29].Results are at the natural logarithm scale.data processing systems.The analysis system is composed of a refrigerant-free preconcentration module with a Stirling cooler (TC4189, Lihan Cryogenics Co., Ltd Guangdong, China), a gas chromatography (GC) instrument (A91P, Panna Instruments Co., Ltd Jiangsu, China), and a quadrupole mass spectrometry (MS) detector (7700B, Suzhou Anyeep Instrument Co., Ltd Jiangsu, China).The principles and processes for the preconcentration module are based on the state-of-the-art Medusa GC-MS system, widely used worldwide within the AGAGE network [30].
ODS5-pro measures CFC-11, dichlorodifluoromethane (CFC-12 or CCl 2 F 2 ), chlorodifluoromethane (HCFC-22 or CHClF 2 ), chloromethane (CH 3 Cl), dichloromethane (CH 2 Cl 2 ) and approximately 40 other halogenated compounds [31].The system automatically and simultaneously collects ambient air samples, calibrates standards using cryogenic preconcentration, and then analyzes and quantifies them using GC-MS.ODS5-pro makes observations for ambient air at regular 140-minute intervals and is bracketed with a standard before and after the air sample to correct for instrumental drift in calibration.
The precision of these observations, determined by repeating analysis of a calibration standard, is typically within 1%, 1%, 1%, 5%, and 5% for CFC-11, CFC-12, HCFC-22, CH 3 Cl, and CH 2 Cl 2 , respectively.The accuracy of the system is within the precision for each compound, and better than ± 0.5%.The sampling methodology for determining halogenated gases in the background atmosphere is comparable with methods reported by AGAGE and NOAA [30].For calibration of ambient air samples, a working standard gas (compressed ambient air) is regularly calibrated by a tertiary standard gas supplied by the AGAGE network.The observations follow AGAGE standard scales, traceable to the Scripps Institution of Oceanography (SIO) calibration scales SIO-05 (CFC-11, CFC-12, HCFC-22, and CH 3 Cl) and SIO-14 (CH 2 Cl 2 ).
The linear response curve is used to illustrate the relationship between the sampling volume and the MS spectrum response under the condition of standard injection volume (2 L), by analyzing samples with ambient mole fraction levels across a range of volumes (0.1-6.0 L).The ODS5-pro system, as demonstrated in figure S2, shows good linear responses with the injection volume of 0.1-6.0L for all target compounds.The linear fitting coefficients R 2 for CFC-11, CFC-12, HCFC-22, CH 2 Cl 2 , and CH 3 Cl are all above 0.999.Yi et al (2023) [31] provide detailed information regarding the ODS5pro system and the sampling and analyzing methods.

Baseline
To determine the background levels of CFC-11 and tracers, a robust extraction of baseline signal (REBS) filter [32] is applied.The REBS filter employs iterative processes to fit a local regression curve to the data, incorporating robustness weights for values exceeding the baseline.It iteratively excludes data points outside a specified range around the current baseline and marks those points as pollutant mixing ratios.In this study, the fitting process converges after five iterations, and we exclude data points falling out of ±1 standard deviation from the baseline to obtain the best fit for the mole fraction enhancements of trace gases (figure S3).

Inverse modeling
The air parcels transported to XCG are analyzed using the Lagrangian particle model FLEXPART version 10.4 [25].The model simulates advection, random diffusion, and atmospheric turbulence, according to a preset time interval and the number of hypothetical particles.Backward simulations are conducted driven by global meteorological fields from the National Centers for Environmental Prediction's Climate Forecast System Reanalysis model, with a spatial resolution of 0.5 • × 0.5 • and a temporal resolution of 6 hours [33,34].In backward simulations, 50 000 virtual particles were released in each 3 hour interval and tracked backward in time for 20 days in a regional domain.
The output of the FLEXPART backward simulation is source-receptor sensitivity (SRS) matrices (H), also known as 'footprint' .The emissions of CFC-11 (E CFC−11 ) from southeastern China are calculated combining the simulated SRS with a Bayesian inversion algorithm and CFC-11 observations.The cost function for optimizing CFC-11 emissions in the grid is: where x is the emission state vector (Gg yr −1 ), x a is its prior estimate (Gg yr −1 ), y obs is the observed CFC-11 enhancements over baseline (ppt), S a is the prior error covariance matrix (Gg yr −1 ), and S o is the observation error covariance matrix (ppt).Following previous studies [35,36], ∇ x J(x) = 0 yields the minimum of the cost function: Here, S b is the posterior error covariance matrix (Gg yr −1 ).The prior emissions vector (x a ) is from the bottom-up inventory of Chinese provincial CFC-11 emissions, disaggregated into grids based on GDP [37].The prior emission for southeastern China is 1.50 Gg yr −1 during the study period.
In the conducted study, three different prior emissions fields were utilized, including scaling by 0.5 ('low'), 1 ('base'), and 2 ('high') for CFC-11 emissions.To these prior magnitudes, distinct prior emission errors of 150%, 100%, and 50% were assigned for the low, base, and high priors, respectively [36].The square of the corresponding value was used as the diagonal element of S a .It is chosen so that the relatively higher or lower priors are tested to ensure that the inversion is not biased by the high prior magnitudes.The three prior emissions were scaled appropriately to cover the posterior emissions.Importantly, the findings indicated that the magnitude of the prior emission and its error did not have a significant impact on the posterior emission results.The observational uncertainties, as derived from the XCG observations, stand at 2.32 ppt for elevated CFC-11 mixing ratios.The spatial distribution of CFC-11 emissions and the associated uncertainties are further discussed in section 3.2.

ISC method
The ISC method was applied to estimate CFC-11 emissions (E CFC−11 ) from southeastern China [11,12,17,22,31,[38][39][40][41][42][43][44][45][46][47][48].This method derives the emission of a trace gas of interest by correlating its enhanced mixing ratios above the baseline with that of a reference compound (tracer).This empirical ratio approach estimates regional emissions of various substances in a simple and robust manner, compared with inverse methods that require complex computational processes in combination with chemical transport models.Selecting the appropriate tracers follows three criteria: (1) a long lifetime and thus low chemical reactivity during transport from the source to the receptor, (2) well-quantified tracer emissions, and (3) approximately co-located source regions for target species and tracer to ensure significant correlations.
The emission of CFC-11 in southeastern China is estimated as: where E Tracer (Gg yr −1 ) is the emission of the tracer, M CFC−11 (g mol −1 ) and M Tracer (g mol −1 ) is the molecular weight of CFC-11 and the selected tracer, respectively, S (unitless) is the slope of the orthogonal regression between the CFC-11 enhancement over the baseline (∆C CFC−11 ) and tracer enhancement (∆C Tracer ).The weighted Deming regression (WDR) method was used to compute the ∆C CFC−11 /∆C Tracer ratio S [49].Previous studies using the WDR method derived a more accurate slope and intercept and generated best fits representing the overall correlation trends [44,49].Following error propagation, the error of CFC-11 emission (σ ECFC−11 ) is: where σ ETracer and σ s is the error of the tracer emission E Tracer , and ∆C CFC−11 /∆C Tracer ratio S, respectively.S1 complies with all observations in the figure .ratios at XCG are approximately 0.45% to 5.36% higher than those measured by MHD and MLO, respectively.Compared with previous observations made in urban and regional background stations (251.09± 9.14 ppt), much lower mixing ratios (221.64 ± 2.29 ppt) were obtained during the study period, arguing for a recent reduction in production and emissions of CFC-11 in southeastern China.

Results and discussion
Despite a continuous decrease in background levels, clear pollution signals of CFC-11 are evident in figure S3a, suggesting consistent inflows of air masses at XCG as influenced by regional sources.Short-term elevated CFC-11 are seen in September 2022 and January 2023.Meanwhile, enhanced CFC-12 (figure S3b), which is known to be likely related to CFC-11 production, was observed.The ratio of observed CFC-12 to CFC-11 is highly variable (0.43 to 2.33), depending on whether the CFC-12 is vented to the atmosphere, destroyed, stockpiled, or banked in new equipment [57].The moderate correlation between enhanced CFC-12 and CFC-11 (r = 0.49, figure S4a) suggests potential minor CFC-12 and CFC-11 emissions in southeastern China.

CFC-11 Emissions in southeastern China
Inverse modeling was used in conjunction with CFC-11 observations to constrain CFC-11 emissions in each grid, as presented in figure 3. The three inversions (base, high, low prior magnitudes in case, table S3) yield total posterior emissions from 1.45 ± 0.08 Gg yr −1 to 1.70 ± 0.12 Gg yr −1 (on average 1.58 ± 0.21 Gg yr −1 ) in southeastern China.Thus, magnitudes of prior emissions and errors do not significantly alter the posterior emissions.The results highlight the Yangtze River Delta (YRD) as the major source region, with a small portion from the PRD.Some regions around the YRD, including Shanghai, northern Zhejiang, and southern Jiangsu, contribute the most to total emissions.In the Guangdong-Hong Kong-Macao Greater Bay Area, the most urbanized region in China, our constraints lead to 18.4% reduction (75.17 Mg yr −1 to 61.34 Mg yr −1 ) in posterior emissions compared to the prior emissions, indicating that CFC-11 emissions in this region are lower than the bottom-up inventory.For the YRD region, additional sites are needed to achieve more accurate constraints.[37], inverse modeling, and the interspecies correlation (ISC) methods.The ISC method uses observations of CFC-11 and several tracers, including HCFC-22, CH2Cl2, and CH3Cl, with respective tracer emissions from bottom-up inventories [58], merged inversion and bottom-up results [59], and inversion results [60].Error bars show the ± 1 standard deviations of the mean.aligning with the bottom-up provincial inventories of 1.50 Gg yr −1 [37].Table S4 lists detailed results from the ISC method.

Changes in CFC-11 emissions in southeastern China
CFC-11 emissions are expected to have decreased since 2012 due to the diminishing size of the banks.Previous studies have projected future CFC-11 emissions in China using bottom-up estimates of reported productions, the size of the CFC-11 banks, and assumed emission rates [37,61,62].In 2008-2011, the bottom-up estimates were consistent with the top-down estimates [16,39,42,45,48].However, a significant divergence was observed between the two during 2014-2018 [7,11,13], which markedly decreased in the recent years of 2022-2023, suggesting that either there may be no accidental CFC-11 emissions in southeastern China, or that the unintended CFC-11 emissions have likely been effectively managed to return to levels that are closely in line with the projected estimates (1.50 Gg yr −1 in Wu et al (2023) [37], 1.03 Gg yr −1 in Fang et al (2018) [62] ).The results also show that emissions of CFC-11 in the PRD have returned to pre-2010 levels, with the estimates being 0.60 Gg yr −1 to 0.69 Gg yr −1 lower than estimates conducted for urban regions in the PRD during September 2008 (1.00 ± 0.20 Gg yr −1 ) [17] and November 2002 (1.00 ± 0.30 Gg yr −1 ) [47].This might imply the success of MPA implementation in controlling the production and emission of CFC-11 in this region.However, due to the lack of historical emission data for the YRD, it is unable to ascertain the changes in emissions in the YRD.

Conclusion
This study reports CFC-11 mixing ratios and emissions in southeastern China.In situ observations and subsequent modeling of ambient air parcels at XCG reveal possible sources of CFC-11 emissions in southeastern China, particularly near the YRD.The findings suggest that CFC-11 emissions in southeastern China from both top-down (1.23 Gg yr −1 to 1.58 Gg yr −1 ) and previous bottomup (1.50 Gg yr −1 ) estimations have returned to pre-2010 level (1.00 Gg yr −1 ) in recent years, likely due to the successful implementation of the MPA.More research monitoring CFC-11 emissions is needed to confirm this trend and investigate potential remaining sources in other parts of China.
Compared with previous studies, the most direct and accurate estimates are presented by applying the inverse modeling and ISC methods based on highfrequency continuous in situ observations from a new regional background station, XCG, which will contribute continuously to global observations of CFC-11, as well as other key ODSs and fluoridated GHGs, with a focus on southeastern China and Southeast Asia in the future.To enhance the accuracy of halogenated gas emission estimates in mainland China, observations from more stations, especially those in western and northern China, are needed to compensate for the low sensitivity of observations at this new station in parts of the China region in the future.

Figure 1 .
Figure 1.The location of the Xichong station (XCG; light blue triangle) and its source-receptor sensitivity (SRS).The bold black line outlines the study area, referred to as southeastern China, including the provinces of Anhui (AH), Fujian (FJ), Guangdong (GD), Hainan (HN), Jiangsu (JS), Jiangxi (JX), Shanghai (SH), and Zhejiang (ZJ).Sensitivities, i.e., how sensitive the observations at XCG are to emissions from each 0.5 • × 0.5 • grid, are from the lower 100 m of the FLEXPART model output[29].Results are at the natural logarithm scale.

Figure 2 .
Figure 2. CFC-11 observations in China from 2004 to 2023.Averaged mixing ratios of CFC-11 in urban (hollow symbols) and rural (filled symbols) sites are presented with the sampling duration (horizontal solid lines) and ±1 standard deviation (vertical error bars).The abbreviation Mt. refers to mountain sites.High-frequency observations at the Xichong station (XCG, black) from May 2022 to April 2023 are included.Data from two northern hemispheric background stations, Mace Head (MHD, dark blue, Ireland; monthly) and Mauna Loa Observatory (MLO, light blue, the United States; daily), are also included.TableS1complies with all observations in the figure.

Figure 3 .
Figure 3. Averaged CFC-11 emissions (Mg yr −1 ) constrained with observation at the Xichong station (XCG; light blue triangle) from May 2022 to April 2023 as well as the spatial distribution of uncertainties.The emissions outside the study area (bold black line) were exclude.Figure S5 shows the prior emissions and the difference between the prior and averaged posterior emissions.

Figure 4 .
Figure 4. CFC-11 emissions in southeastern China.Results are based on bottom-up inventories[37], inverse modeling, and the interspecies correlation (ISC) methods.The ISC method uses observations of CFC-11 and several tracers, including HCFC-22, CH2Cl2, and CH3Cl, with respective tracer emissions from bottom-up inventories[58], merged inversion and bottom-up results[59], and inversion results[60].Error bars show the ± 1 standard deviations of the mean.

Furthermore
, the ISC method was applied to estimate CFC-11 emissions with HCFC-22, CH 2 Cl 2 , and CH 3 Cl as To minimize autocorrelation among observations, a 5-day aggregation was conducted before performing the regression analysis of the polluted mixing ratios [41].Statistically significant (p − value < 0.01) correlations were found between enhanced CFC-11 and tracers at XCG.The slope of ∆C CFC−11 /∆C Tracer (ppt/ppt) is 1.90 × 10 −2 ± 0.35 × 10 −2 for HCFC-22, 2.30 × 10 −3 ± 0.51 × 10 −3 for CH 2 Cl 2 , and 8.50 × 10 −3 ± 1.40 × 10 −3 for CH 3 Cl.Tracers with reported Chinese emissions were allocated based on the 2022 provincial GDP.It is worth noting that additional uncertainty was introduced by using GDP to allocate the tracer emissions in southeastern China.This is due to the distribution of different tracers is inconsistent to varying degrees with the distribution of GDP in 2022.HCFC-22 emissions were from the provincial inventories [58], CH 3 Cl was from Hu et al (2022) [60], and CH 2 Cl 2 was from An et al (2021) [59].Figure 4 compares estimated emissions in southeastern China using different tracers, ranging from 1.23 ± 0.26 Gg yr −1 to 1.58 ± 0.21 Gg yr −1 ,