Observed decade-long improvement of combustion efficiency in the Yangtze River Delta region in China

The ΔCO/ΔCO2 ratio is a good indicator of the combustion efficiency of carbon-containing fuels, and can be useful to assess the combustion efficiency on a regional scale. In this study, we analyzed in-situ observations of CO2 and CO concentrations from 2011 to 2021 at the Station for Observing Regional Processes of the Earth System (SORPES), in the Yangtze River Delta (YRD) region of eastern China, and calculated the ΔCO/ΔCO2 ratio to investigate the combustion efficiency in the YRD region. Furthermore, we used a Lagrangian particle dispersion model WRF-FLEXPART to evaluate the contribution of each emission sources to the observed ΔCO/ΔCO2 ratio. We found that the observed ΔCO/ΔCO2 ratio showed a persistent decreasing trend of 1.0 ppb/ppm per year and decreased ∼47.9% during this period, illustrating an evident improvement in the combustion efficiency in the YRD region. The improvement of the combustion efficiency is a result of China’s Air Pollution Prevention and Control Action Plan announced in 2013. However, the decrease of ΔCO/ΔCO2 ratio slowed down from 1.3 ppb ppm−1 per year during 2011–2016 to 0.6 ppb ppm−1 per year during 2017–2021. The simulation results reveal that the slowdown of the decrease in the ΔCO/ΔCO2 ratios can be explained by the slowing improvement of combustion efficiency in steel source in the industry sector. Our results verify the effectiveness of emission reduction efforts in the YRD region and highlight the necessity of long-term observations of CO2 and CO.


Introduction
As the most important anthropogenic greenhouse gas in the atmosphere, the concentration of carbon dioxide (CO 2 ) has increased ∼50% relative to its preindustrial level (Wu et al 2024), accounting for ∼64% of the radiative forcing by long-lived greenhouse gases (Montzka 2023).Human activities, especially fossil fuel use and deforestation, are the main drivers of the CO 2 rise, with more than 70% of the total fossil fuel CO 2 emissions from cities and urban areas (IEA 2008, Friedlingstein et al 2022).Atmospheric carbon monoxide (CO) is a trace gas that has an important impact on human health and atmospheric chemistry.As an indirect greenhouse gas, CO has its main sink of the reaction with hydroxyl radical (OH), thus influencing the oxidation capacity of the atmosphere and the concentrations of CO 2 and CH 4 (Daniel andSolomon 1998, Petrenko et al 2013).Moreover, CO is also one of the precursors of surface ozone (O 3 ), and can result in an indirect radiative forcing of ∼0.2 Wm −2 (IPCC 2007, Myhre et al 2013).
In addition to oxidation of hydrocarbons in the atmosphere, atmospheric CO is mainly produced from incomplete combustion of carbon-containing fuels, such as fossil fuels and biomass.Since fossil fuel burning and biomass burning are the major sources of CO in most urban areas, the ratio between variations of atmospheric CO and CO 2 , i.e. ∆CO/∆CO 2 , can reflect the combustion efficiency and be used to identify the sources.For example, the ∆CO/∆CO 2 ratios are relatively high (∼100 ppb ppm −1 ) in inefficient burning like biomass burning, whereas the ratio from power plant plumes may be two orders of magnitude smaller (∼1 ppb ppm −1 ) (Andreae and Merlet 2001, Wang et al 2010, Zhao et al 2012, Tohjima et al 2014).
In addition to source identification, the longterm trend of ∆CO/∆CO 2 ratio can reflect the changes in combustion efficiency of fossil fuels.Based on in situ observations of CO 2 and CO concentrations at Hateruma Island during 1998-2010, Tohjima et al (2014) found a continuing decrease of the ∆CO/∆CO 2 ratio from 45 ppb/ppm in 1998 to 30 ppb/ppm in 2010, indicating an improvement of fossil fuel combustion efficiency.Measurements of the ∆CO/∆CO 2 ratios from vehicles emissions in a tunnel in Pennsylvania (Bishop et al 1996) showed that the ratio was 95 ppb/ppm for pre-1975 vehicles and 18.9 for those in 1983-1993.An emission reduction of ∆CO/∆CO 2 from on-road vehicles was also observed during 1996-2006(Bishop and Stedman 2008), owing to emission control technologies and vehicle maintenance.
However, the ∆CO/∆CO 2 ratios can also be influenced by biospheric CO 2 fluxes.To more accurately represent the fossil fuel combustion efficiency, the ratio of CO to fossil fuel-derived CO 2 (CO 2ff ), i.e. ∆CO/∆CO 2ff , has been adopted, which requires the separation of fossil fuel CO 2 from natural sources and sinks.The ∆CO 2ff signal can be obtained by measuring ∆ 14 CO 2 (Turnbull et al 2011a, 2011b, 2015, Djuricin et al 2012, Miller et al 2012, Niu et al 2016, Berhanu et al 2017), based on the fact that fossil fuel CO 2 is 14 C-free and dilutes the atmospheric ∆ 14 CO 2 (Suess 1955).For example, Djuricin et al (2012) used tree ring radiocarbon measurement in several sampling sites to calculate the ratio between CO and CO 2ff during 1980-2008, and showed a ∆CO/∆CO 2ff ratio decrease of ∼20 ppb/ppm from 56-97 ppb ppm −1 at all sites over the study period, as a result of the reduction in CO emissions from onroad mobile gasoline combustion sources.
The improvement of combustion efficiency can lead to energy savings and air pollution reduction.Therefore, combustion efficiency monitoring is of great importance for urban areas.China is the world's largest CO 2 emitter with rapid economic growth fueled by fossil energy (Gregg et al 2008).However, long-term in-situ measurements of atmospheric CO 2 are still sparse in China, especially in urban areas with ∼85% of China's carbon emissions (Dhakal 2010, Wang et al 2010).Studies of combustion efficiency using the observed atmospheric ∆CO/∆CO 2 ratio, especially long-term ones, are even more limited in China.The majority of ∆CO/∆CO 2 ratio studies are concentrated in the areas around Beijing (Han et al 2009, Wang et al 2010, Turnbull et al 2011b, Liu et al 2018, Che et al 2022, Li et al 2022).Our study area, the Yangtze River Delta (YRD) region, covers the alluvial plains surrounding the Yangtze River estuary with an area of around 350 000 square kilometers and is home to over 240 million people.Although the YRD region is one of the fastest economic growing and the most densely populated regions in the world, only a few studies on the combustion efficiency have been performed so far (Huang et al 2015, Liu et al 2018).
Our study focuses on the long-term trend of the combustion efficiency from anthropogenic emissions in the YRD region.Firstly, we analyzed the continuous observation of ∆CO/∆CO 2 ratios during 2011-2021 to quantify the variation and the trend of the ratios.We then simulated ∆CO/∆CO 2 ratios during 2013-2020 using WRF-FLEXPART (Stohl et al 2005, Stein et al 2015) to assess the contributions of different emission sources to the observed changes.Our study assessed the effect of emission control policies on fossil fuel combustion efficiency, which would be a reference for the policy makers in CO 2 and air pollution reduction in the future.

Atmospheric measurements of CO and CO 2 2.1.1. Site and measurements
Continuous CO 2 and CO concentration measurements were conducted at the Station for Observing Regional Processes of the Earth System (SORPES), which is located at the Xianlin campus of Nanjing University (118 • 57 ′ 10 ′′ E, 32 • 07 ′ 14 ′′ N), at a height of ∼40 m above sea level.The SORPES station focuses on investigating the impacts of human activities in the rapidly urbanized and industrialized East China, and is suitable to capture plumes from fossil fuel combustion in the YRD city cluster (Ding et al 2016).
Measurements of CO 2 and CO from August 2011 to October 2021 were analyzed in this study.Atmospheric concentrations of CO 2 and CO were measured by two non-dispersive infrared (NDIR) analyzers (Thermo Fisher Scientific, model 410i and 48i), respectively.The measurements had a time resolution of 5 min, and were averaged to obtain hourly mean values.The Model 410i uses an internally stored calibration curve to linearize the instrument output over a range up to a concentration of either 10 000 ppm (Standard) or 25% (High Level), and for the Model 48i the range is up to a concentration of 10 000 ppm.The sampling intake was on the roof of a two-floor building on the site.

Observational data analysis
The original time series of CO and CO 2 concentrations were divided into short segments according to each of the two adjacent troughs of smoothed CO 2 data that was obtained by applying a Gaussianweighted moving average filter over a sequence of 30 h window.Note that the smoothed data was only used to obtain the segments, and the ∆CO/∆CO 2 ratios were calculated using unsmoothed hourly averaged data for each segment.As CO and CO 2 are co-emitted from fossil fuel burning, we expect a strong correlation between ∆CO and ∆CO 2 when observed plumes were dominated by fossil fuel burning.In practice, we selected the segments with an R-squared value larger than 0.6 for further analysis (see figure S2).The trend of ∆CO/∆CO 2 , i.e. the fossil fuel combustion efficiency, over the study period, was then calculated.We assessed the uncertainty of the calculated ratios of ∆CO/∆CO 2 by using CO 2 measurements made by both Thermo Fisher Scientific analyzer and Wisdominc GGA-311 greenhouse gas high-precision analyzer.

Emissions and fluxes data 2.2.1. MEIC inventory emissions of CO and CO 2
The emission inventory used in this study is Multiresolution Emission Inventory model for Climate and air pollution research (MEIC) v1.4 Chinese CO 2 and CO emission inventories (http://meicmodel.org.cn, last access: 29 September 2023).The inventories contain power, industry, residential, transportation and agriculture sectors and are subdivided into 22 sources (Li et al 2017, Liu et al 2015, Tong et al 2018, Liu et al 2021, Peng et al 2019, Zheng et al 2014).
In our study, we first used the total and sectoral CO and CO 2 monthly emission inventories in China.The spatial resolution of the inventories is 0.25 • × 0.25 • .We distribute the emissions equally on 0.1 • × 0.1 • grids to match the resolution of the calculated footprints described below, and then combine the inventories with the footprints to investigate the influence of the total anthropogenic emissions and each emission sector on the ∆CO/∆CO 2 ratio.To understand how the emissions change in a particular source influences the ∆CO/∆CO 2 ratios, we recalculated the ∆CO/∆CO 2 ratios by letting emissions from one single source change while keeping the emissions from other sources consistent (at their initial emission levels in 2013).

GCAS CO 2 flux dataset
We obtained a global monthly 1x1-degree surface CO 2 fluxes dataset generated by the Global Carbon Assimilation System, version 2 (GCASv2), which is named as GCAS2021 (Jiang et al 2022, https://doi.org/10.5281/zenodo.5829774,last access: 28 January 2024).The CO 2 fluxes contain the components of net ecosystem exchange (NEE) and ocean (OCN) fluxes, and wildfire (FIRE) and fossil fuel and cement (FFC) emissions.Similar to the approach introduced in section 2.2.1, we calculated the influence of biosphere NEE on ∆CO 2 and then the biosphere-derived ∆CO 2 , which was then compared with fossil fuelderived ∆CO 2 to assess the influence of biosphere on the observed ∆CO/∆CO 2 ratios.

WRF-FLEXPART simulations of ∆CO and ∆CO 2
The WRF-FLEXPART model (Brioude et al 2013) was implemented to establish the link between surface emissions and atmospheric concentrations.We ran the model in winter seasons (January, February and December) during the period of 2013-2017.The detailed configurations of WRF-FLEXPART are presented in table S1.
The meteorological fields for WRF-FLEXPART were generated by the WRF-CHEM model, which considered the effect of aerosols' radiative forcing on suppressing the development of the planetary boundary layer, and thus would be suitable on polluted days (Jia et al 2021).The model was driven by the National Centers for Environmental Prediction (NCEP) Final Operational Global Analysis (FNL) data every 6 h to generate output every hour.The spatial resolution was 20 km × 20 km, and the simulations covered a range of 160 × 180 grids (indicated by the fan-shaped area in figure 1) with the center coordinate of (35 • N, 110 • E).A total of 27 vertical levels were used from ground to the top pressure of 50 hPa.
The Lagrangian particle dispersion model (LPDM) FLEXPART simulates trajectories of a large number of particles to describe the atmospheric transport and dispersion processes.It can determine the spatial distribution of potential source contributions for the given receptors when running backward in time.In our study, 10 000 virtual particles were released at a height of 10 m above ground level.The domain of the simulations was from 105 • E to 126 • E, 25.5 • N to 43.5 • N with resolution of 0.1 • × 0.1 • , covering most area of eastern China as shown in the red rectangular frame in figure 1.The residence time for a thickness of 100 m above the surface was calculated and considered the 'footprint' retroplume (Ding et al 2009).The FELXPART model output two results per day, one at 2:00 and the other at 14:00 (BJT).Then the simulated footprint [s −1 ] would be multiplied with CO and CO 2 emission inventories to get anthropogenic emissions' contribution on CO and CO 2 concentrations at SORPES.The calculation of simulated ∆CO and ∆CO 2 is defined as: Here the CO 2 _emission and CO_emission represents CO and CO 2 emissions (or fluxes) in the inventories [gm −2 s −1 ], height represents the footprint level in FLEXPART [m], and k is a flexible coefficient considering the existence of unit transformation for different emission inventories.
We multiplied sectoral CO and CO 2 grid emission inventories and the footprint to discuss the contribution of changes in combustion efficiency in each emission sector to the variation of total ∆CO/∆CO 2 ratio.For the purpose of distinguishing individual emission sector from total emission, we designed five different emissions scenarios.In the first scenario, we keep all emission sources at their 2013 emission level as a baseline; and in the rest scenarios, we separately let the emissions from one emission sector change overtime while the remaining emission sectors keep their emissions unchanged from 2013.In this research we used MEIC sectoral CO and CO 2 monthly emission inventories, and emissions changes of industry, power, residential and transportation sectors were discussed.

Observed ∆CO/∆CO 2 ratios
Our observations reveal a persistent decrease of the ∆CO/∆CO ratio; however, the rate of decrease has slowed down since 2017.The observed ∆CO/∆CO 2 ratios at the SORPES show that the ∆CO/∆CO 2 ratios decreased by an average of 1.0 ppb/ppm per year, from about 21.5 ppb ppm −1 in early 2011 to 11.2 ppb ppm −1 at the end of 2021 (see figure 2(a)).Furthermore, we found that the decrease of the ∆CO/∆CO 2 ratios slowed down from 2017, with a decrease of 1.3 ppb ppm −1 per year during 2011-2016 and 0.6 ppb ppm −1 per year during 2017-2021.
A clear seasonal cycle can be seen in the ∆CO/∆CO 2 ratio.The magnitude of winter ∆CO/∆CO 2 is nearly twice as high as that of summer, and the decrease rate in winter is also much faster, with an average of 1.4 ppb ppm −1 per year for winter and 0.8 ppb/ppm per year for summer.Regardless of the seasonal difference, the slowdown of the decrease can be found in both summer and winter, from 1.4 ppb ppm −1 to 0.9 ppb ppm −1 in winter (considering the production reduction caused by the shutdown of industrial enterprises during the Spring Festival, we excluded the Spring Festival) and from 1.5 ppb ppm −1 to 0.2 ppb ppm −1 in summer, respectively.In addition, the correlation between CO and CO 2 is less significant in summer than in winter; only 22.2% of the data passes the R 2 > 0.6 filters in summer while the proportion in winter is 65.1%, which is likely caused by non-fossil related sources of CO and CO 2 , and vegetation sinks of CO 2 .Similar seasonal variations have been reported in previous studies (Tohjima et   The observed ∆CO/∆CO 2 ratio of 21.5-11.2ppb ppm −1 in the YRD region for the period of 2011-2021 is much higher than the reported ∆CO/∆CO 2 ratios in Europe for the same period.For example, the ratio measured in Switzerland by Berhanu et al (2017) in the winter of 2012 was 7.3 ppb ppm −1 .The measurement conducted in the Netherlands by Super et al (2017) found the mean ratios in different cities varying from 2.0-3.9 ppb ppm −1 in 2014-2015, and a ratio of 3.01-6.80ppb ppm −1 in France from 2013 to 2014 was measured by Ammoura et al (2016).Moreover, the ∆CO/∆CO 2 ratio is in the range of 10.5-60.2 for other cities in China (Liu et al 2018, Xia et al 2020, Che et al 2022, Li et al 2022, Wu et al 2022).Therefore, there is still a long way to go for China to improve its combustion efficiency.To improve this situation, it is important to attribute the ∆CO/∆CO 2 ratios to specific emission sources, where emissions reduction efforts can be focused.In the following section, we employed LPDM modeling to investigate the main causes of the observed ∆CO/∆CO 2 ratios.

Contributions of the decreasing of ∆CO/∆CO 2 ratios 3.2.1. Comparison of simulations and observation
Considering that atmospheric CO 2 and CO concentrations are mainly affected by fossil fuel emissions in winter, we performed our simulation for winter only.We calculated ∆CO and ∆CO 2 by multiplying the anthropogenic emissions of CO and CO 2 by footprint, then dividing them to get the ∆CO/∆CO 2 ratio and taking the monthly average.
The ∆CO/∆CO 2 ratio simulated by WRF-FLEXPART from 2013 to 2020 is compared with the observation at SORPES (figure 3).Although the simulated ∆CO/∆CO 2 ratio is on average about 2.3 ppb ppm −1 lower than the observation during 2013-2020, the simulated and observed downward trends are consistent.The monthly average observation of the decreasing rate from 2013-2020 is about 1.3 ppb ppm −1 per year, while the simulated rate is 1.2 ppb ppm −1 per year.Previous researchers have assessed the uncertainty of China's CO emission inventories and found a large uncertainty of 50%-86% (Zhang et al 2009, Kurokawa et al 2013).Feng et al (2020) further confirmed that the CO emissions in the YRD region were underestimated in the MEIC inventory, which may explain the underestimate of ∆CO/∆CO 2 ratio in WRF-FLEXPART simulation.However, the emission changes on the grid scale are larger than the emission uncertainties in the YRD region.Therefore, the overall trend of ∆CO/∆CO 2 ratio may not be much affected by the uncertainty.In the following step, we aim to find out how temporal changes of emissions per individual sector have affected the ∆CO/∆CO 2 ratios.

Contribution from individual emission sector
The contribution from each individual sector to the variations of the ∆CO/∆CO 2 ratio from the 2013 level is shown in figure 4   the decrease of the ∆CO/∆CO 2 ratio.The improvement of combustion efficiency in the industry sector resulted in a total decrease of 3.9 ppb ppm −1 in the ∆CO/∆CO 2 ratio from 2014 to 2020, explaining an average of 63.3% of the total decrease over this period.Each of the transportation and residential sectors caused a decrease of about 0.9 ppb ppm −1 by 2020, together accounting for 46.6% of the trend.The power sector's contribution is the smallest, less than 10% for all these years.The ∆CO/∆CO 2 ratio caused by the power sector only in 2020 has even increased 1.6 ppb/ppm compared to 2013, due to the emissions changes of the power sector, implying a decrease in combustion efficiency of the power sector.
In addition, we find out that the slowdown of the decrease rate after 2017 in the ∆CO/∆CO 2 ratio can likely be attributed to the industry sector.Although the decrease rate in the ∆CO/∆CO 2 ratio for all sectors has slowed down after 2017; the decrease rate for the industry sector is much larger compared with the other three sectors.The total ∆CO/∆CO 2 ratio decreased 4.3 ppb ppm −1 from 2013 to 2016 when only the industry sector is considered, while the reduction of the ratio dropped to 3.9 ppb ppm −1 by 2020.This finding raises our concern about more specific changes of combustion efficiency in emissions from the industry sector, so in next section we have made a further investigation to assess its detailed categories.

Emission changes in sources in the industry sector
The 22-source provincial emission inventories are applied in our study to explore the emissions changes in industry sector.We use the sum of the emissions from Jiangsu, Zhejiang, Anhui and Shanghai to represent the total emissions in the YRD region.The emissions of CO 2 and CO in the sector are divided into industrial boiler, cement, coking, steel, Among all the emission sources in industry sector, the steel source has made dominant contributions to the decrease in ∆CO/∆CO 2 ratio in industrial emissions, and the slowdown in the rate of decrease is also mainly due to the change in steel source emissions.The ∆CO/∆CO 2 ratio variations caused by each source up to 2017 and 2020 are shown in table 1.Compared to the initial ratio of 27.1 ppb ppm −1 in 2013, the improvement of combustion efficiency in emissions from the steel source has led to a decrease of 7.6 ppb ppm −1 by 2017 and 9.9 ppb ppm −1 by 2020.Previous researches have also proved the obvious improvement of the combustion efficiency in steel sources during this period in China (Xia et al 2016, Zheng et al 2018b).The ∆CO/∆CO 2 ratio changes caused by other sources are at least an order of magnitude lower than steel sources.The emissions changes in industrial boilers, cement, coking and petrochemical industry sources have a slight increase in the ∆CO/∆CO 2 ratio by 2020.The increase of ∆CO/∆CO 2 ratios came with a decrease of ∼10% in the total emissions in industry and power sector, which was caused by the lockdown of COVID-19.We speculate that the increase of the ratios was also influenced by changing activities during COVID-19, which reduced the emissions from the sources with relatively small CO to CO 2 emission ratios (i.e.power and high combustion efficiency boilers and kilns).Considering the ∆CO/∆CO 2 ratios have tended to be flattening at a low level recently, the small fluctuations in the contributions of different source sectors will lead to inter-annual variations in the measured ∆CO/∆CO 2 ratio.
The improvement of combustion efficiency in anthropogenic emissions, especially in industry sector and steel source, is largely a result of the issue of China's Air Pollution Prevention and Control Action Plan in 2013.The Action Plan formulated ten measures for controlling air pollution, including controlling industries with high energy consumption and high pollution, eliminating backward production capacity in industries, speeding up the adjustment of the energy mix, increasing the supply of clean energy such as natural gas and coal-to-methane.Previous studies have found the Action Plan effective in improving the air quality and promoting the energy efficiency (Feng et al 2019, Liu et al 2022).Our study reveals the variations of ∆CO/∆CO 2 ratios in the YRD region as a result of the Action Plan, suggesting an improvement of combustion efficiency.However, the improvement of anthropogenic combustion efficiency brought by the Action Plan has become weaker after 2017, which means new measures may be taken as a step forward.

The uncertainty from biospheric influence
The observed ∆CO/∆CO 2 ratios contain signatures from biospheric signals, although we tried to limit the biospheric influence by using a threshold for the correlation coefficient and focusing on winter when the vegetation activities are weak.We compared the ∆CO 2 caused by biosphere NEE and fossil fuel emissions in winter during 2013-2020 (figure S5), the CO 2 emitted from respiration of vegetation is on average 15.5% of that from fossil fuel, and has little change in the trend (∼−2%) during the study period.We then applied monthly averaged ratios of simulated biospheric and fossil fuel ∆CO 2 to estimate the ∆CO 2ff and calculated the ∆CO/∆CO 2ff (figure S6).The ratio of ∆CO/∆CO 2ff is on average 2.8 ppb ppm −1 (15.6%) higher than the observed ∆CO/∆CO 2 ratio during the period of 2013-2020, which will not influence the main conclusions of this study.Due to the high uncertainty in the simulated biospheric CO 2 , we did not use it to assess the influence of biosphere on the overall trend of the ∆CO/∆CO 2 ratio.

Conclusion
Long-term observations of ∆CO/∆CO 2 ratios at the SORPES station are analyzed in our study to investigate the combustion efficiency in the YRD region.We found that the observed ∆CO/∆CO 2 ratios showed a persistent decreasing trend of 1.0 ppb ppm −1 per year and decreased ∼47.9% during the studied period, illustrating an evident improvement in the combustion efficiency in the YRD region.However, the decrease of ∆CO/∆CO 2 ratios slowed down from 1.3 ppb ppm −1 per year during 2011-2016 to 0.6 ppb/ppm per year during 2017-2021.Through the simulations of WRF-FLEXPART, we found that the largest contribution to the decrease of ∆CO/∆CO 2 ratios from 2013 to 2020 was caused by the industry sector, which accounts for more than 60% among all the major emissions sources in the inventories.Meanwhile, the slowdown of the decrease in ∆CO/∆CO 2 ratios is also mainly due to the emissions changes in the industry sector.Further investigation of 22-source inventories reveals that the steel source is the dominating cause of the combustion efficiency improvement in the industry sector.The improvement of the combustion efficiency is attributed to China's Air Pollution Prevention and Control Action Plan since 2013.The Action Plan has led to a significant improvement of the combustion efficiency up to 2017, but is not so effective as before after 2017.The slower decrease of ∆CO/∆CO 2 ratios denotes that the improvement slows down with the completion of the rectification in production capacity, thus new technologies and policies are needed for further improvement in the future.This study sheds light on the effectiveness of China's Air Pollution Prevention and Control Action Plan to reduce emission of air pollutants, and the results could be helpful for policy makers to develop further measures for sustained environmental improvement in the future.

Figure 1 .
Figure 1.Domain settings in WRF-FLEXPART and topographic height of the field (m).
al 2014, Ammoura et al 2016, Chandra et al 2016, Liu et al 2018).The reason for the higher ratio of ∆CO/∆CO 2 in winter may owe to the higher CO/CO 2 emission ratio in transportation and industry (Zheng et al 2018a), and the lower correlation of CO and CO 2 variations in summer can be explained by the enhanced signals of CO 2 from biosphere (Wang et al 2010, Tohjima et al 2014) and of CO from oxidation of VOCs (Seiler 1974, Vimont et al 2019).Several studies have been done in the surrounding areas in the YRD region, Liu et al (2018) measured the ∆CO/∆CO 2 ratios of 20.1 ppb/ppm at the Lin'an station (40 • 39 ′ N, 117 • 07 ′ E, 293 m asl) between 2009
(a), with the contribution proportion shown in figure 4(b).As shown in figure 4(a), the industry sector contributed the most to

Figure 3 .
Figure 3.Comparison of the monthly averaged time series of observed (rose red dots) and simulated (green squares) ∆CO/∆CO2 ratios in winter between 2013-2020.The error bar is the 95% confidence interval.

Figure 4 .
Figure 4. (a) ∆CO/∆CO2 ratio changes made by each emission sector and (b) the contribution proportion of them.

Table 1 .
Contributions of each industry source in ∆CO/∆CO2 ratio changes (unit: ppb ppm −1 ).The positive value means a reduction effect on total ∆CO/∆CO2 ratio and the negative value means an increased effect.