The impact of COVID-19 lockdown on atmospheric CO2 in Xi'an, China.

Lockdown measures to control the spread of the novel coronavirus disease (COVID-19) sharply limited energy consumption and carbon emissions. The lockdown effect on carbon emissions has been studied by many researchers using statistical approaches. However, the lockdown effect on atmospheric carbon dioxide (CO2) on an urban scale remains unclear. Here we present CO2 concentration and carbon isotopic (δ13C) measurements to assess the impact of COVID-19 control measures on atmospheric CO2 in Xi'an, China. We find that CO2 concentrations during the lockdown period were 7.5% lower than during the normal period (prior to the Spring Festival, Jan 25 to Feb 4, 2020). The observed CO2excess (total CO2 minus background CO2) during the lockdown period was 52.3% lower than that during the normal period, and 35.7% lower than the estimated CO2excess with the effect of weather removed. A Keeling plot shows that in contrast CO2 concentrations and δ13C were weakly correlated (R2 = 0.18) during the lockdown period, reflecting a change in CO2 sources imposed by the curtailment of traffic and industrial emissions. Our study also show that the sharp reduction in atmospheric CO2 during lockdown were short-lived, and returned to normal levels within months after lockdown measures were lifted.

Introduction

The novel coronavirus disease (COVID-19) epidemic is a major public health emergency, which spread fast, caused extensive infections and proved difficult to contain. In order to block the spread of COVID-19 and protect people's health, the Chinese government adopted strong lockdown measures in Wuhan on January 23, 2020. In the next few days, a number of provinces and cities across the country also began to adopt similar measures to control the spread of the disease. These included strict traffic controls, restrictions on residents' going out, and closures of market gatherings and various businesses. The reopening of schools in China following the Spring Festival (Jan 25, 2020 to Feb 4, 2020) was also delayed. Only business entities that provided people with daily necessities, such as health care and food supplies, remained open. As a result of these COVID-19 prevention measures, the number of vehicles on the road declined dramatically and manufacturing production dropped sharply (Wang et al., 2020a).

Fossil fuel combustion produces both carbon dioxide (CO2) emissions and air pollutants. Energy consumption (including fossil fuels) has increased sharply in recent years as a result of urbanization and modern lifestyle changes (Hosseini et al., 2019). Fossil fuels are responsible for 85% of CO2 emissions and 64% of total greenhouse gas emissions (Razzaq et al., 2020). As a consequence of increasing energy consumption, CO2-dominated greenhouse gases have also increased in recent years (Adeniyi et al., 2019), and on a global scale are attributed to the following sectors: electricity and heat generation (44%), transportation (26%), and industry (19%), according to the International Energy Agency (IEA, 2020). The consumption of fossil fuels causes ecological and environmental problems (Al-Juboori et al., 2020a), such as climate warming and urban smog (Sher et al., 2020). CO2 emissions must be curtailed to mitigate global warming and enhance sustainable development (Lelieveld et al., 2019). This strategy hinges on alternative energy technologies, which include: solar and wind power (Qazi et al., 2019); sustainable hydrocarbon fuel production from CO2 (Al-Juboori et al., 2020b); hydrogen fuel cells (Al-Shara et al., 2019); and biofuels (Razzaq et al., 2020). Additionally, carbon capture and sequestration techniques such as post-combustion CO2 capture using fast adsorbents derived from biomass (Sher et al., 2020) can reduce CO2 emissions. The transition from fossil fuels to renewable energy will play an essential role in CO2 emissions reduction, but it is not happening fast enough (Gielen et al., 2019). The COVID-19 pandemic provided an opportunity to test how fast carbon emissions can be reduced by sharply curtailing emissions.

Control measures to check the spread of COVID-19 led to a reduction in road and air traffic, a temporary closure of businesses, and a decrease in industrial productivity (Ding et al., 2020). A number of factors that contributed to CO2 emissions reductions (Wang et al., 2020b) include: 1) reduced power demands due to the delayed resumption of work after the Spring Festival; 2) a reduced demand for steel and blast furnace operation time; 3) a reduced energy demand from a variety of business enterprises; and 4) COVID-19 restrictions required people to stay at home. This latter factor markedly reduced road and air traffic, with immediate reductions in transportation-related CO2 emissions. Several studies have performed statistical analyses to estimate national and global reductions in carbon emissions that resulted from pandemic prevention measures. For example, compared with data from the same time periods in 2019, daily global CO2 emissions decreased by 17.0% by early April 2020 (Le Quéré et al., 2020) and decreased by 8.8% in the first half of 2020 (Liu et al., 2020a). In China, carbon emissions fell by 11.0% over the first quarter of 2019 (Han et al., 2020). There are also a number of observational studies that shows declines of atmospheric pollutants. Bauwens et al. (2020) found that NO2 concentrations decreased rapidly following the COVID-19 lockdown both in China and Italy. Xu et al. (2020) reported that submicron aerosol mass concentrations were reduced by 50% during the COVID-19 lockdown in Lanzhou, China. Surface measurements made at more than 800 monitoring stations show that mean levels of PM2.5 and NO2 in northern China decreased by approximately 35% and 60%, respectively, after the COVID-19 lockdown (Shi and Brasseur, 2020). Sharma et al. (2020) reported that a PM2.5 concentration decrease of 43% in India during the COVID-19 lockdown period compared to the previous 4 years.

Cities play an important role in the effort to reduce carbon emissions (Xu et al., 2021), as they account for about 70% of global carbon emissions (Churkina, 2016). The COVID-19 pandemic represents an experiment that can be used to test how much urban-scale atmospheric CO2 concentrations are lowered when anthropogenic CO2 emissions are sharply curtailed. Liu et al. (2020b) reported decreases in on-road CO2 concentrations in Beijing during COVID-19 with six on-road observations using mobile platforms. Turner et al. (2020) observed a 5–50 ppm decrease in midweek CO2 concentrations during rush hour monitoring in the San Francisco Bay Area. However, the response of averaged atmospheric CO2 concentration to the lockdown on an urban scale is still unknown (Pigliautile et al., 2020).

Xi'an is the largest city in northwestern China and all of its residential complexes were locked down from February 5 to February 21, 2020, due to COVID-19 control measures. Here we study the urban-scale effect of the lockdown in Xi'an on atmospheric CO2 concentrations and stable carbon isotope compositions (δ13C) in the first quarter of 2020. The objective of this study is to detect the averaged CO2 concentration change during the 2020 COVID-19 lockdown period relative to 2019 levels and relative to meteorological corrected levels. Quantifying the impact of the COVID-19 lockdown on atmospheric CO2 concentrations on a city scale is important to future carbon emission measures for sustainable development. It can also provide useful information for modeling studies.

Methods

Study site

Xi'an is currently the capital of Shaanxi Province and in historical times, was the capital of China during thirteen dynasties. Its population reached 10 million in 2018. Xi'an is located in the south-central part of the Guanzhong Basin, bordered by the loess plateau to the north and the Qinling Mountains to the south. This basinal configuration, and the mild winds typical of Xi'an most of the year, inhibit the removal of air pollutants (Yang, 2003). The observations made for this study (Fig. 1 ) were carried out on the main building of the Institute of Earth Environment, Chinese Academy of Sciences (IEECAS) in southeast Xi'an from Jan 1 to March 31, 2019 and 2020.

Fig. 1

The location of the study site (IEECAS) in Xi'an.

Timeline of COVID-19 responses in Xi'an during the study period

We divided the first quarter of 2020 into five stages according to the different measures taken in response to COVID-19. Stage 1 (January 1 to January 24, 2020) represents a normal period before the Spring Festival holiday (the Chinese Lunar New Year). Stage 2 (January 25 to February 4) is during the Spring Festival holiday of 2020. Spring Festival is the largest holiday in China. In normal years, people travel to visit their relatives and friends from the first day of the Chinese New Year. But in 2020, millions of people were asked to stay at home in an effort to stop the spread of the new coronavirus. Stage 3 (February 5 to February 21) was the lockdown period, with the strictest control measures enforced. Only one person per family was allowed to venture out to purchase daily necessities such as food and medicine, once every two days. The reopening of industries and schools in Xi'an was also delayed throughout this period. Transportation was largely restricted with few vehicles on the road during that time. Stage 4 (February 22 to February 28) was a transition period as the lockdown measures were relaxed. Businesses began to reopen and restrictions on residents began to be lifted. During stage 5 (February 29 to March 31) normal patterns were reestablished, approaching Stage 1 conditions. However, schools and cinemas did not reopen immediately, and group tours that crossed provincial borders were still restricted.

Experimental set-up

Atmospheric CO2 concentration and its δ13C were measured using a Picarro G2131-i carbon isotopic analyzer (Picarro, Inc., USA). The Picarro analyzer measures CO2 concentration, δ13C in CO2, CH4 and H2O. The precision for CO2 is better than 0.2 ppm and for δ13C is better than 0.2‰. Air samples were pumped directly into the Picarro analyzer at a flow rate of 25 ml/min. The CO2 concentration was derived from the sum of dry air concentrations of 12CO2 and 13CO2. The instrument was calibrated by two standard gases (cylinder 1 with CO2 395.49 ± 0.02 ppm, δ13C in CO2 −8.980 ± 0.008‰, CH4 1993.5 ± 0.2 ppb, and cylinder 2 with CO2 491.43 ± 0.02 ppm, δ13C in CO2 −10.395 ± 0.024‰, CH4 3029.3 ± 0.5 ppb) obtained from the Chinese Academy of Meteorological Sciences. Each standard gas is pressurized in a 29.5-L treated aluminum alloy cylinder (Scott-Marrin, Inc., California) fitted with a high-purity, two-stage gas regulator, and calibrated with cylinders assigned by the WMO/GAW CO2 Central Calibration Laboratory operated by NOAA/ESRL.

Reconstruction of missing CO2 data

To study the atmospheric CO2 response to the lockdown, we took January to March 2019, as a reference period. However, 6 days during the period (January 27 to February 1, 2019) were missing due to instrument failure. A study in Shanghai, China found that atmospheric CO2 and CO correlate well with each other (Wei et al., 2020). Daily averaged CO data for Xi'an were obtained from the Chinese Air Quality online Monitoring and Analysis Platform (CAQMAP, 2020), and we found that the daily averaged CO2 and CO have a highly significant (p < 0.001) linear relationship (Fig. 2 ). Based on this, we reconstructed the daily average CO2 concentrations to fill the gap from the 6 missing days in our record.

Fig. 2

Linear regressions between CO2 and CO for the first quarter of 2019 (a) and 2020 (b) in Xi'an. The CO data are from CAQMAP (2020).

Results and discussion

Analysis of the atmospheric CO2 concentration in Xi'an

Daily CO2 concentration variations in the first quarters of 2019 and 2020 are shown in Fig. 3 . CO2 concentrations decreased during the 2020 Spring Festival because the Xi'an city government began to limit outdoor activities from the beginning of the Chinese New Year. The Spring Festival trend in 2020 is much different from the typical pattern represented by the 2019 data. In the latter, CO2 concentration began to increase since the first day of the holiday as people started to travel to visit relatives and friends. This resulted in the enhancement of vehicle emissions. In 2020, CO2 concentrations decreased steadily from the start of the Spring Festival to the end of the lockdown. In the transition period, CO2 concentrations followed an increasing trend, as lockdown measures in Xi'an were relaxed and people returned to work. After the reestablished normal period in the city, CO2 concentrations became stable, with weak daily fluctuations.

Fig. 3

Atmospheric CO2 concentration in Xi'an City in the first quarter of 2019 (a) and 2020 (b). The blue line shows daily average CO2 concentrations; black dashed line shows monthly average CO2 concentrations; the red dotted line in Fig. 3(b) shows average CO2 concentrations in each stage. (For interpretation of the references to colour in this figure legend, the reader is referred to the Web version of this article.)

The red dotted line in Fig. 3(b) shows the atmospheric CO2 concentrations in the five stages in 2020: 1) 484.5 ± 21.4 ppm, 2) 458.6 ± 12.8 ppm, 3) 448.0 ± 15.7 ppm, 4) 456.1 ± 9.8 ppm and 5) 442.4 ± 9.6 ppm. This trend correlates well with the timing of measures taken to control the spread of COVID-19. The CO2 concentration in the lockdown period is significantly lower (p < 0.001) than that in the normal period before the Spring Festival by 7.5% and is also lower (p = 0.083) than the Spring Festival period and lower (p = 0.235) than the transition period.

Because the natural variability of CO2 caused by the carbon cycle (Peters et al., 2017) and meteorological conditions (Ballantyne et al., 2012) in the short term are large, they may mask anthropogenic variability. Hence, we used monthly-averaged CO2 concentrations to check whether an anomaly could be distinguished in February (the height of the lockdown). The horizontal black dashed lines in Fig. 3 show the monthly average CO2 concentrations. The values are 479.2 ± 21.9 ppm, 451.0 ± 14.2 ppm, and 442.0 ± 9.4 ppm for January, February and March 2020, respectively; and as a reference they are 472.5 ± 20.9 ppm, 457.3 ± 16.2 ppm, and 439.8.0 ± 10.1 ppm for January, February and March 2019, respectively. In January and March the CO2 concentrations are slightly higher in 2020 than that in 2019, but in February, CO2 concentrations are significantly (p < 0.01) lower (1.4% or 6.3 ppm) in 2020 than in 2019. The changes in monthly average CO2 concentrations were the same for the first quarters of 2019 and 2020, which decreased in February (as compared to January) and decreased again in March (as compared to February). CO2 declines in March were significant in both 2019 (p < 0.01) and 2020 (p < 0.01). The decrease in February compared with January is not significant in 2019 (p = 0.108), while it is significant in 2020 (p < 0.01), suggesting that the lockdown influenced fossil fuel CO2 emissions. The CO2 concentration declined 3.2% (or 15.2 ppm) in February compared with January in 2019, but in 2020 it declined 5.9% (or 28.2 ppm). These all indicate the influence of the lockdown on CO2 concentrations.

Prediction of CO2 from meteorological data

Although the comparison of the five stages in the first quarter of 2020, and the month-to-month comparison between 2020 and 2019 show an apparent influence of COVID-19 measures, these results include both natural carbon cycle variability and meteorological conditions. Quantifying and attributing changes in CO2 concentrations requires accounting for meteorological effects in addition to direct emissions (Turner et al., 2020). We did this by determining the difference between CO2 and CO2 , as explained next.

We adopted the method of Venter et al. (2020), that predicts air pollution proxies (PM2.5, O3, and NO2) from meteorological parameters, to estimate first quarter 2020 CO2 concentrations (CO2 ). We first established the relationship between daily CO2 concentration and weather parameters (temperature, relative humidity, wind speed and precipitation, at Xi'an Jinghe National Meteorological Station, No. 57131, RP5, 2020) in 2019 using a multiple linear regression model. The resulting R2 and p-value for the relationship are 0.573 and < 0.001, respectively. The regression equation we derived is as follows:

with the parameters T (temperature, °C), RH (relative humidity, %), PP (precipitation, mm) and WS (wind speed, m/s). Then the daily CO2 concentrations in the first quarter of 2020 were estimated by inputting the daily meteorological parameters of 2020. Considering the background CO2 concentration, the difference between the observed CO2 (observed CO2 concentration minus the background CO2 concentration from Mauna Loa, Hawaii, NOAA, 2020), and the estimated CO2 (estimated CO2 concentration minus the background CO2 concentration) is defined as the lockdown effect. In fact, the CO2 includes fossil fuel CO2 (CO2 ) and biogenic CO2 (CO2 ) (Levin et al., 2003). Former studies in Xi'an showed that in winter months, the CO2 comes predominately from fossil fuel emissions (Wang et al., 2018), which can account for more than 90% (Zhou et al., 2020). Thus the CO2 in our study mainly reflects fossil fuel CO2 variations.

Fig. 4 shows that the CO2 is significantly (p = 0.013) lower than the CO2 by 35.7% (or 11.7 ppm) during the lockdown period, which is close to the reduction of fossil fuel emissions in China, by 32.0 ± 12% in February 2020 (Tohjima et al., 2020). However, during the normal period before the Spring Festival the CO2 is significantly (p = 0.005) higher than the CO2 by 24.0% (or 13.8 ppm). In the other three stages there are no obvious differences between the observed and estimated values. This result indicates a clear COVID-19 lockdown effect on CO2. The CO2 during the lockdown period is 20.4% lower than that during the normal period before the Spring Festival. Note that the amplitude of this decline is significantly smaller than the uncorrected CO2 value, which declined 52.3%. In the study of Liu et al. (2020b), a higher CO2 concentration was observed during the lockdown period in 2020 than the same period in 2019. This higher CO2 should be attributed to the weather conditions, rather than COVID-19 control measures since they observed a significant decline of on-road CO2 concentration for the same period. In the transition period the CO2 concentration was higher (p = 0.738) than CO2 again, indicating a return to pre-lockdown conditions.

Fig. 4

CO2 time series of observed (green) and estimated (red) values, from weather parameters. (For interpretation of the references to colour in this figure legend, the reader is referred to the Web version of this article.)

The variation of atmospheric CO2 concentrations in the first quarter of 2020 indicates that quickly lowering atmospheric CO2 concentrations is possible. However, the impact of the COVID-19 lockdown measures on atmospheric CO2 concentrations is short-term. To maintain low atmospheric CO2 concentrations, emissions must continue to be suppressed, perhaps through green technologies to produce hydrogen (Al-Shara et al., 2021) and hydrogen fuel cells (Khzouz et al., 2020).

Evidence of the COVID-19 lockdown effect on CO2 from isotopic measurements

Stable isotopes (δ13C) in atmospheric CO2 provide a valuable means to distinguish between different CO2 sources in air because different sources can have very different δ13C values. In order to investigate whether CO2 sources changed significantly during the lockdown period, the Keeling-plot method (Keeling, 1958) was used to determine δ13C values for each time period. The observed CO2 can be divided into background CO2 and source CO2. According to the mass balance of CO2 and its stable carbon isotopes, we can write the following (Keeling, 1958):

Combining equations (2), (3) we can obtain:

By plotting δ13C and 1/CO2 , the mean isotopic signature of the CO2 can be obtained as the y intercept of the Keeling-plot curve. We applied this method to the morning rush hour for each period we divided in 2020 to study the effect of lockdown on vehicle emissions (Fig. 5 ).

Fig. 5

Keeling-plots morning rush hour in each stage of the first quarter of 2020.

The results show that the δ13C values for the normal period before Spring Festival, during Spring Festival, during the lockdown period, in the transition period, and the reestablished normal period are: 26.8‰, −25.6‰, −17.5‰, −27.3‰ and −24.9‰, respectively. The δ13C in the lockdown period is obviously different from the other four stages. We note that the very low R2 value (0.18) for the lockdown period makes it impossible for us to obtain a reliable δ13C CO2 source value. The Keeling plot method assumes that the background and source are constant during the period investigated (Keeling, 1958). The CO2 concentrations and associated δ13C show negligible variations at the background site in Mauna Loa (NOAA, 2020), thus a low R2 may reflect daily CO2 source changes during the lockdown period, and so the assumptions of the Keeling plot method are not satisfied. The daily δ13C can be affected by different weather conditions resulting in advection of different air masses or changes in the CO2 source distribution (Sturm et al., 2006), and energy usage patterns (Górka and Lewicka-Szczebak, 2013) due to the adjustment of the epidemic prevention policies from strict to loose during the lockdown period. Another possible reason for the low R2 might result from the low CO2 range (Zobitz et al., 2006), which is only 41.6 ppm in the lockdown period. The δ13C values in the other four periods are close to those observed in Wroclaw, Poland, which are −25.7‰ and −27.3‰, during two heating seasons (Górka and Lewicka-Szczebak, 2013), indicating fossil fuel combustion CO2 sources during the winter season.

Conclusions

The impact of the COVID-19 lockdown on atmospheric CO2 concentrations in Xi'an was assessed using ground observations corrected for the influence of weather. The results show that during the lockdown period, observed CO2 concentrations were 7.5% lower than normal (before the Spring Festival). Daily CO2 sources changed during the lockdown, as reflected by the low correlation (R2 value) observed using the Keeling-plot method applied before, during, and after the lockdown period. Although the impact of the lockdown on atmospheric CO2 concentration in Xi'an was large, its impact was short-lived. Following the relaxation of the pandemic prevention measures, CO2 concentrations increased again to similar levels as observed in 2019. This study quantifies to some extent the rate and magnitude of changes that can occur by sharply curtailing anthropogenic CO2 emissions in an urban environment. In practice, we expect that such reductions can be achieved through the implementation of green technologies. Our monitoring approach can be used in the future to assess the efficacy of such technologies.

Declaration of competing interest

The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.