Skip to main content

Thank you for visiting nature.com. You are using a browser version with limited support for CSS. To obtain the best experience, we recommend you use a more up to date browser (or turn off compatibility mode in Internet Explorer). In the meantime, to ensure continued support, we are displaying the site without styles and JavaScript.

Detection of fossil-fuel CO2 plummet in China due to COVID-19 by observation at Hateruma

Abstract

The COVID-19 pandemic caused drastic reductions in carbon dioxide (CO2) emissions, but due to its large atmospheric reservoir and long lifetime, no detectable signal has been observed in the atmospheric CO2 growth rate. Using the variabilities in CO2 (ΔCO2) and methane (ΔCH4) observed at Hateruma Island, Japan during 1997–2020, we show a traceable CO2 emission reduction in China during February–March 2020. The monitoring station at Hateruma Island observes the outflow of Chinese emissions during winter and spring. A systematic increase in the ΔCO2/ΔCH4 ratio, governed by synoptic wind variability, well corroborated the increase in China’s fossil-fuel CO2 (FFCO2) emissions during 1997–2019. However, the ΔCO2/ΔCH4 ratios showed significant decreases of 29 ± 11 and 16 ± 11 mol mol−1 in February and March 2020, respectively, relative to the 2011–2019 average of 131 ± 11 mol mol−1. By projecting these observed ΔCO2/ΔCH4 ratios on transport model simulations, we estimated reductions of 32 ± 12% and 19 ± 15% in the FFCO2 emissions in China for February and March 2020, respectively, compared to the expected emissions. Our data are consistent with the abrupt decrease in the economic activity in February, a slight recovery in March, and return to normal in April, which was calculated based on the COVID-19 lockdowns and mobility restriction datasets.

Introduction

The outbreak of the new coronavirus (COVID-19) was first identified in Wuhan, China, in December 2019. The government of China took a range of measures including a lockdown of Wuhan, shut-down of the inter-city transportation, and reduction of socioeconomic activity at the end of January 2020 to prevent the spread of COVID-19 within China and to the outside world. As a result of these measures, it was estimated that the emissions of fossil-fuel-derived CO2 (FFCO2) in China decreased by about 25% during January–February1,2. Significant reductions in nitrogen dioxide (NO2), which is an atmospheric pollutant mainly produced by fossil fuel combustion in engines, were also detected over China by remote sensing satellites3,4. However, no observational evidence has been reported about the detection of China’s emission reduction in atmospheric CO2 concentrations. It is also unclear whether atmospheric observations can support the direct estimation of reductions in FFCO2 emissions in near-real-time.

A global observation network for atmospheric greenhouse gases has been developed since the 1960s5 not only to evaluate the global anthropogenic and natural flux budgets but also to quantitatively estimate the regional/country-scale emission changes6,7,8,9. Therefore, detecting the COVID-19 influence on atmospheric CO2 from observations and, if possible, quantitatively evaluating the emission change in China is crucially important to test the capacity of our observation networks. National restrictions owing to the COVID-19 situation give us a unique opportunity to validate some of the hypotheses that are needed to successfully implement the Paris Agreement.

The National Institute for Environmental Studies (NIES)/Center for Global Environment Research (CGER) has been conducting global monitoring of atmospheric greenhouse gases by using a variety of platforms, including ground sites10, commercial cargo ships11,12, aircraft13,14, and satellites15,16. As a part of the global monitoring effort, NIES/CGER operates an observation station at Hateruma Island (HAT), Japan, to carry out comprehensive atmospheric measurements. Since the island is located in a marginal region of continental East Asia, the outflow of the continental air masses with elevated greenhouse gas concentrations is often captured at HAT. Previous studies revealed that the synoptic-scale variations during the winter season can be used to constrain the emissions from continental East Asia17. Here, we examine whether the reduced economic activity in China has caused any detectable change in the CO2 synoptic variations at HAT. Synoptic variations are defined as the hourly to weekly variations in the CO2 and CH4 time series, and their variabilities are termed as ΔCO2 and ΔCH4, respectively (see “Methods” section).

Results

Anomalous behavior of CO2 in February 2020

In Fig. 1, daily means of the detrended and deseasonalized atmospheric CO2 and CH4 mole fractions (see “Methods” section) observed at HAT in January-March 2020 are compared with those of the previous 9-year (2011–2019) average. Closely investigating these data, we found that February 2020 was the only occasion when CO2 was systematically lower than the long-term (2011–2019) mean for 22 days out of the 29 days of the month, and 9 days in the 2020 values fell outside the 1-σ standard deviation range of the 2011–2019 average. A large fraction of the variabilities in CO2 (ΔCO2) at HAT are affected commonly by the movement of emission signals from China by air mass transport, which can be analyzed effectively by air mass trajectories (Fig. S1 in Supplementary Material). Due to the common high emissions of CO2 and CH4 over China and air mass trajectories between China and HAT, similar variabilities were observed also for CH4 (ΔCH4) in February 2020, although the frequency of lower value cases was fewer than those for CO2. Such anomalies were emphasized when we compared the 7-day moving averages between the daily means and the 9-year averages (Fig. 1).

Figure 1
figure1

Variabilities in atmospheric CO2 and CH4 mole fractions observed at HAT. The detrended and deseasonalized CO2 (top) and CH4 (bottom) mole fractions observed at HAT from January to March 2020 are compared with the corresponding 9-year averages. The red triangles and red bars are the daily means and the standard deviation (1-σ) for 2020, the black triangles and black bars are the 9-year average of the daily mean and daily standard deviation for the corresponding day of the year, and the grey shade is the standard deviation (1-σ) of the 9-year data. The red and black solid lines represent the 7-day moving averages of the daily means in 2020 and the 9-year averages, respectively.

However, it is quite difficult to relate such anomalies to the change in FFCO2 emissions. The atmospheric CO2 signal due to changes in FFCO2 weakens by atmospheric transport as the distance between the source and observation site (or receptor), i.e., HAT, increases. Yet the rather simple air mass trajectories and the suppressed terrestrial biospheric exchange in the winter months allow us to evaluate the FFCO2 sources in China from the synoptic-scale variabilities at HAT17,18. In the rest of the analysis, we discuss the ratios of ΔCO2 and ΔCH4 for removing the first-order effect of atmospheric transport on the CO2 synoptic variability. This study was based on the assumption that the CO2 and CH4 flux signals had similar spatial distributions in the outflow region of China during the winter through spring, and thus produced coinciding peaks and troughs in atmospheric variability observed at HAT17,18.

25-year change in the ΔCO2/ΔCH4 ratio in winter

The temporal variations in the monthly average of the ΔCO2/ΔCH4 ratio for January, February, and March after 1998 are shown in Fig. 2. The average ΔCO2/ΔCH4 ratios for the three months show an increasing trend during 2001–2011 and reach a plateau after 2011. An analysis using the Lagrangian particle dispersion model (LPDM) revealed that emissions from the northeastern and eastern parts of China predominantly contributed to the increase of the ΔCO2/ΔCH4 ratio at HAT when averaged over November through March17,18. For comparison, we have also plotted the FFCO2 inventory emission estimates for China using data from the International Energy Agency (IEA)19, Emission Database for Global Atmospheric Research (EDGARv5.0)20, and Global Carbon Project (GCP)21. The trend of the observed ΔCO2/ΔCH4 ratio is similar to the increasing trend of the FFCO2 estimates. Therefore, the rapid increase in the ΔCO2/ΔCH4 ratio during the 2000s has been attributed to the rapid increase in fossil fuel consumption associated with the unprecedented economic growth in China. Although a steady increase in anthropogenic CH4 emissions was estimated from China during the 2000s22,23,24, the increasing trend of ΔCO2/ΔCH4 indicates that the relative growth rate of the CO2 emissions has exceeded that of the CH4 emissions. The rather stable ΔCO2/ΔCH4 ratio after 2011 suggests a consistent and slower increase of both FFCO2 and CH4 emissions from China (Fig. 2). Thus, we used the period of 2011–2019 as the reference period for analyzing the 2020 emission change.

Figure 2
figure2

Temporal variations in the monthly average ΔCO2/ΔCH4 ratios for January, February, and March since 1996. The grey thick line represents the trend curve of the ΔCO2/ΔCH4 ratio based on a digital filtering technique25 with a cut-off period of five years and the grey-shaded area represents the 95% range of the variations from the trend curve. Thin lines are the estimation of FFCO2 emissions from China based on IEA, EDGARv5.0, and GCP. The position of the right y-axis is adjusted so that the FFCO2 temporal variations visually fit the trend curve of the ΔCO2/ΔCH4 ratio. The vertical bars represent the standard deviations (1σ) for the monthly values.

The ΔCO2/ΔCH4 ratio for February 2020 shows a significant decrease beyond the 95% confidence interval. The histograms of the individual ΔCO2/ΔCH4 ratios of the 24-h time windows for February show that the histogram for 2020 shifts slightly toward a smaller ΔCO2/ΔCH4 ratio compared to the recent decade (2011–2019) (Fig. S2 in Supplementary Material). The ΔCO2/ΔCH4 ratio of February 2013 also shows a significant decrease. Investigating the back trajectories arriving at HAT in February during the recent decade (Fig. S1 in Supplementary Material), we found that air masses were occasionally transported from Southeast Asia in February 2013, a region with lower CO2/CH4 molar emission ratios than the East Asian countries. Thus, the low ΔCO2/ΔCH4 ratio in February 2013 could be attributed to irregular air mass transport. Since such an irregularity was not confirmed in February 2020, we consider that a reduction in the FFCO2 emissions from China caused the observed decrease.

Temporal variation in the ΔCO2/ΔCH4 ratio from December 2019 through April 2020

The temporal changes in the 30-day moving-window average of ΔCO2/ΔCH4 ratio are shown in Fig. 3. The 9-year (2011–2019) average ratios show a slight decline from January to February. As the period between late January and early February corresponds to the Chinese New Year, the decline might be attributed to the decrease in FFCO2 emissions from China associated with the seasonal decrease in economic activity. The 9-year average also shows a decreasing trend in April, due mainly to a gradual increase in biogenic CH4 emissions (Fig. S3 in Supplementary Material). Although the 30-day-averaged standard deviations for CO2 and CH4 (ΔCO2 and ΔCH4) show basically similar patterns of temporal variations (Fig. S3 in Supplementary Material), the 30-day-averaged ΔCO2/ΔCH4 ratios show a sharp decrease from January to February 2020. The January anomaly exceeds the range of the 9-year average, reaches a minimum in mid-February, and then gradually moves toward the 9-year average in March. The observed results suggest a rapid recovery in the FFCO2 emissions from China by the end of March, which are consistent with recent FFCO2 emission estimates based on the activity data from power generation and industry (https://carbonmonitor.org/).

Figure 3
figure3

Temporal variations in ΔCO2/ΔCH4 ratios and estimated FFCO2 emissions. (a) The red line represents the 30-day moving average of the ΔCO2/ΔCH4 ratios from December 2019 to April 2020, the black line with grey shade represents the 9-year (2011–2019) average of the 30-day moving average and the range of the variation (± 1σ), and the orange lines represent the results of the NICAM-TM simulation, corresponding to the control emission case (no symbols) in comparison with the weakest (circles), moderate (triangles) and strongest (squares) FFCO2 reduction cases. The red and black open squares represent the monthly means of the ΔCO2/ΔCH4 ratios and the 9-year (2011–2019) average of the monthly means. (b) The blue lines represent the temporal variations of estimated FFCO2 emission change in China based on the restriction levels for the 30 Chinese provinces given by Le Quéré et al.2.

We examined whether the observed change in the ΔCO2/ΔCH4 ratio reflected the FFCO2 emission change in China on a day-to-month time scale with respect to the pre-pandemic emission level (Fig. 3). The temporal change in the relative FFCO2 emissions in China is shown in Fig. 3, which was calculated from the time series of the levels of measures against the virus for the Chinese provinces given by Le Quere et al.2 (see “Methods” section). The observed ΔCO2/ΔCH4 ratio traces amazingly well the change in the relative FFCO2 emissions from China. This result validates convincingly the hypothesis that the ΔCO2/ΔCH4 ratio at HAT can track changes in FFCO2 emissions from China on a shorter time scale during the winter months when the terrestrial biosphere is in hibernation.

To further investigate whether the estimated FFCO2 change could explain the observed changes in the ΔCO2/ΔCH4 ratio, an atmospheric transport model named NICAM-TM26 was used to estimate the changes in the ratios (see “Methods” section). Globally distributed CO2 fluxes included emissions from fossil-fuel combustion and cement production, terrestrial biospheric exchange and air-sea exchange, and CH4 fluxes comprised of both anthropogenic and natural emissions. In addition to the control simulation, three sensitivity simulations were performed with modified FFCO2 emissions in China, in which the emissions were reduced to 70%, 80%, and 89% following the lock-down intensity time series at daily intervals with the strongest, modest, and weakest assumptions, respectively. In March, these reductions were relaxed to 87%, 92%, and 98%, respectively. The temporal changes in the ΔCO2/ΔCH4 ratios, which were computed from the simulated data at an hourly time interval, are also shown in Fig. 3. Although being more variable than the observations, the differences in the simulated ratios for the moderate and strongest reduction cases from that of the control case reproduced well the rapid decrease between January and February 2020. Note that the rather large variability in the simulated ΔCO2/ΔCH4 ratio is in part attributed to the relatively small variability in the simulated CH4. Among the simulation results of the three FFCO2 reduction cases, which correspond to the bottom, middle, and top lines of the FFCO2 emission estimates shown in Fig. 3, the strongest reduction (70% reduction) case explains best the observed reduction in February 2020.

In the NICAM-TM simulations, we used climatological CO2 fluxes from the terrestrial biosphere to calculate the ΔCO2/ΔCH4 ratios (see “Methods” section). Thus, the temperature-dependent interannual variability in heterotrophic respiration was ignored in the model simulation. An analysis of the surface air temperature anomaly did not show any large change over the East China region in February–March 2020 (Fig. S4 in Supplementary Material). The simulated monthly ΔCO2/ΔCH4 ratios for February and March based on the time-dependent FFCO2 emissions and the climatological biospheric CO2 fluxes generally well reproduced the observed temporal variations over the period of 2000–2019 (see Fig. S5 in Supplementary Material). From these facts, we conclude that the observed decrease in the ΔCO2/ΔCH4 ratios at Hateruma was caused predominantly by the change in FFCO2 emissions rather than the change in the terrestrial biosphere.

Estimation of monthly FFCO2 emissions from China

We estimated the FFCO2 emission decreases caused by the influence of the COVID-19 outbreak in China in February and March by using the observed differences between the monthly-mean ΔCO2/ΔCH4 ratios in 2020 and the 9-year (2011–2019) averages (Fig. 3, Table 1). The observed decreases in the ΔCO2/ΔCH4 ratios were 29 ± 11 mol mol−1 (from 129 ± 11 to 100 ± 2 mol mol−1) in February 2020 and 16 ± 11 mol mol−1 (from 133 ± 11 to 117 mol mol−1) in March 2020. The influence of the FFCO2 reductions on the ΔCO2/ΔCH4 ratio can be evaluated using NICAM-TM simulations as well (see “Methods” section and Table S1 in Supplementary Material). Applying the observed decreases to the linear relationships between the decreases in the FFCO2 emissions and the simulated changes in the ΔCO2/ΔCH4 ratio for the individual months, we obtained the estimates of the relative FFCO2 emission decrease of 32 ± 12% in February and 19 ± 15% in March (Fig. 4, Table 1). These estimates are close to the upper limits for the reduction of the activity-based estimations2, which are 20% (11 to 30%) in February and 8% (2 to 13%) in March.

Table 1 Change in the observed ΔCO2/ΔCH4 ratio and estimated FFCO2 decrease.
Figure 4
figure4

Estimation of the decrease in FFCO2 emissions in China. The relationship between the relative decrease in FFCO2 emissions from China and the decrease in the simulated ΔCO2/ΔCH4 ratios is plotted as open circles (blue: February; red: March). The blue and red lines represent the linear regression lines of the simulated data in February and March, respectively. The decreases in the observed ΔCO2/ΔCH4 ratios are applied to the linear regression lines to obtain the estimates for the FFCO2 decrease.

In the calculations, we assumed that the regional pattern of the distribution of FFCO2 emissions was stable, therefore, we used a single scaling factor to adjust the FFCO2 emissions from China. However, regional variations in the distribution of the FFCO2 emissions could also affect the ΔCO2/ΔCH4 ratios at HAT as was discussed in detail in a previous study17, which showed that different FFCO2 emission maps caused significant differences in the ΔCH4/ΔCO2 ratios. Furthermore, as was described in the previous section, the observation at HAT is not sensitive to the FFCO2 emissions from central and western China. Therefore, it should be noted that these limitations would introduce additional errors in the estimated FFCO2 emissions, which are discounted in this study.

For the NICAM-TM simulations, we have assumed that CH4 emissions from China have not changed due to the COVID-19-related economic slowdown. However, there is a possibility that among the major sources including coal exploitation (32%), wastewater handling (13%), enteric fermentation (13%), oil and natural gas exploitation (5%), and solid waste disposal (5%), CH4 emissions from fossil fuel exploitation were also reduced following the reduction in fossil fuel consumption. In a sensitivity simulation, the fossil-fuel-related CH4 emissions, which account for about 37% of the total CH4 emissions from China in winter, were reduced at the same rate as the FFCO2 reduction due to the COVID-19 influence. Such a reduction in CH4 emissions corresponding to the moderate case of FFCO2 reduction in China resulted in increases in the ΔCO2/ΔCH4 ratios in February and March by 28% and 26%, respectively. Taking into account these results, the decrease in the observed ΔCO2/ΔCH4 ratios may suggest that either the decrease in the CH4 emissions was not large or the fossil-fuel-derived CO2 reduction was much larger than the estimations. Nevertheless, we consider that the former is a more plausible explanation. This is because (1) FFCO2 emissions are not directly associated with the above-mentioned anthropogenic CH4 emissions and (2) it is difficult to rapidly (within a few months) change the CH4 seepage from inundated coal mines as well as the biogenic CH4 emissions from wastewater, solid waste and enteric fermentation.

Recently, the UN environment programme discussed the effect of COVID-19 on the growth rate of the global CO2 concentrations27. It is quite difficult to forecast the annual decrease in the global FFCO2 emissions in 2020 because it depends on a variety of factors: the duration of the COVID-19 outbreak, the extent of the restrictions on socioeconomic activity, and the recovery track of the economic activity after the outbreak. A recent estimate based on the plausible range of possibilities shows that the decrease in estimated annual emissions in 2020 ranges from − 5.3 to − 7.5%. Considering the present level of the global FFCO2 emissions (about 10 PgC year−1), even a 10% reduction of the global emissions results in an atmospheric decrease of only about − 0.5 ppm. Given that CO2 increases by about 2 ppm per year with a large interannual variability (up to 70%)28 due to the biosphere-climate feedbacks, we will not be able to distinguish easily the effect of FFCO2 emission reduction on the atmospheric CO2 growth rate in 2020 without a very highly accurate estimation of the land and oceanic uptakes. Our model sensitivity simulation using the control and the largest reduction scenario (− 30% FFCO2 emissions in February) cases suggested a maximum of 0.8 ppm in the total column CO2 (XCO2) change over China and the signal from emission reductions spread to southwest and northeast directions by the atmospheric transport (Fig. S6 in Supplementary Material). Given the small difference of XCO2, the existing remote sensing satellites, the GOSAT and OCO series with single-shot precision of about 2 ppm16,29, will face challenges in detecting any COVID-19 effect. However, our method can detect large signals from the emission reduction from any specific region (China) in near-real-time using continuous, simultaneous, and high-precision measurements of CO2 and CH4.

Methods

Atmospheric observation at HAT

Continuous monitoring of atmospheric CO2 and CH4 has been carried out at HAT since 1993 and 1996, respectively. In this study, the sample air was drawn from the top of a tower at the height of 36.5 m (46.5 m above sea level), dried by passing through dehumidifiers, and then introduced into the individual measurement systems. Atmospheric CO2 was continuously measured by using a nondispersive infrared spectroscopic analyzer (NDIR), whereas atmospheric CH4 was semi-continuously measured by a gas chromatograph (GC) equipped with a flame ionization detector (FID). These measurement systems were calibrated against several standard gases supplied from high-pressure cylinders, and CO2 and CH4 mole fractions were carefully determined against NIES’s original mole fraction scales11,30. The precisions of CO2 and CH4 were ~ 0.1 μmol mol−1 (ppm) and ~ 2 nmol mol−1 (ppb), respectively. The details of the measurements have been reported elsewhere9,10. Note that the analytical interval of the GC/FID system before December 1997 was 1.5 times longer than that after December 1997. Thus, the data after December 1997 were used in this study as was done in a previous study17.

Since 2013, a cavity ring-down spectroscopic analyzer (CRDS, Picarro G-2401) has also been deployed at the monitoring station at HAT to back up the atmospheric observation of CO2, CH4, and CO. The CO2 mole fractions determined by CRDS and NDIR show considerable compatibility; the mean and the standard deviation of the difference in the hourly CO2 mole fractions (NDIR–CRDS) for the period of January 2019–March 2020 was 0.01 ± 0.3 ppm. Since the NDIR system was broken during December 19–30, 2019, and March 4–24, 2020, the data by CRDS were used to fill the data gaps in this study.

An example of the observed synoptic variations of atmospheric CO2 and CH4 at HAT is shown in Fig. S7 in Supplementary Material, where the detrended and deseasonalized time series computed based on a digital filtering technique25 between January and March 2018, 2019, and 2020 are depicted. Note that the right y-axis for CH4 is upside down. It should also be noted that the mean CO2 variations at HAT didn’t show a significant diurnal cycle because of the relatively small influences from local sources or local meteorology during winter as well as CH410. The figure clearly shows that there exists an excellent similarity in both temporal variations. During winter, air masses arriving at HAT are frequently transported from continental East Asia due to the winter East Asian Monsoon. Additionally, there is a rough similarity in the source distributions of the FFCO2 and CH4 in the East Asian region (e.g. Tohjima et al.18). These conditions resulted in excellent correlative synoptic-scale variations in the atmospheric CO2 and CH4 at HAT.

Methods of ΔCO2/ΔCH4 ratio analysis

We didn’t use either baselines or smooth-curve fits to the time series of the atmospheric CO2 and CH4 to examine synoptic-scale variations because we had difficulty in determining appropriate ones31. Instead, we investigated the relative variation ratio by adopting an approach taken by Tohjima et al.17. First, we calculated the standard deviations (ΔCO2 and ΔCH4) and correlation coefficient (R) between CO2 and CH4 for the data within a 24-h time window. Then, we obtained the ΔCO2/ΔCH4 ratio from the individual standard deviations. Note that the ΔCO2/ΔCH4 ratio corresponds to the linear regression slope of the CO2 and CH4 scatter plot based on the Reduced Major Axis method (RMA)32, which takes into account errors in both the independent (x-axis) and dependent (y-axis) variables. This procedure was repeated for the entire data set by shifting the 24-h time window by one hour. Then, the ΔCO2/ΔCH4 ratios with corresponding correlation coefficients larger than 0.7 (R > 0.7) were used to compute the monthly averages or 30-day moving averages shown in Figs. 2, 3, and 4.

To investigate the individual contributions of ΔCO2 and ΔCH4 to the temporal variations in the 9-year-averaged ΔCO2/ΔCH4 ratio shown in Fig. 3, the temporal variations of the 9-year-averaged 30-day moving averages of the ΔCO2 and ΔCH4 were examined (Fig. S3 in Supplementary Material). In general, ΔCO2 and ΔCH4 show similar temporal variations, and their slight differences caused the temporal variation in the ΔCO2/ΔCH4 ratio. The gradually decreasing trend in the ΔCO2/ΔCH4 ratio after March may be attributed to the larger increase in ΔCH4. The 30-day moving averages of the ΔCO2 and ΔCH4 from December 2019 to April 2020 are also depicted in Fig. S3 in Supplementary Material. Again, there is a considerable similarity between ΔCO2 and ΔCH4, suggesting that the atmospheric mixing predominantly caused the temporal variability in those atmospheric components.

In this study, we used the value of 0.7 as the criteria of the correlation coefficient (R) and 24 h for the time window (TW) of the correlation analysis. The results shown in this study are slightly influenced by the selection of these values. The estimated FFCO2 emissions from China based on the observed monthly mean ΔCO2/ΔCH4 ratio for the ranges of 0.5 < R < 0.8 and 12 < TW < 48 (hour) are plotted in Fig. S8 in Supplementary Material. On average, there is a slight tendency that the estimated monthly FFCO2 emissions increase with increasing values of R and TW. However, the standard deviations of the estimated values for the individual months, being less than 6%, are smaller than the uncertainties associated with the individual estimations (Table 1). Therefore, we ignored the effect of the selection of these values in this study.

Estimation of the change in daily FFCO2 emissions from China based on activity data

The influence of measures against COVID-19 on China’s FFCO2 emissions was evaluated based on the approach taken by Le Quere et al.2. They categorized the restrictions to the normal economic activity into 4 levels by introducing a confinement index (CI, where CI = 0 is no restriction and CI = 3 is the highest level of restriction). Then, the change in activity for six economic sectors (power, industry, surface transport, public, residential, and aviation) was estimated as a function of the CI level. Note that the mean and the range (lower and upper levels) of the individual activity changes were given for all combinations of the three CI levels and the six economic sectors2. We evaluated the change in the FFCO2 emissions during January-April 2020 by using the time series of the CI levels for 30 Chinese provinces2. As for the FFCO2 emissions for individual provinces, we used the data summarized by Shan et al.33. We assumed that the proportions of the six economic sectors for the 30 provinces were the same as those for China’s total emissions taken from IEA (2019). The temporal change in the contributions of the individual provinces to the total FFCO2 emissions for the moderate case is depicted in Fig. S9 in Supplementary Material. The temporal change in the total FFCO2 emissions for the moderate case is also plotted in the figure (right y-axis).

Simulation of the change in atmospheric CO2 and CH4 at HAT

Atmospheric CO2 and CH4 mole fractions at HAT were simulated by using a three-dimensional atmospheric transport model, NICAM-TM26, and a comprehensive set of global CO2 and CH4 fluxes. NICAM-TM was developed from the Nonhydrostatic ICosahedral Atmospheric Model (NICAM)34 to examine the atmospheric transport and flux inversion studies of greenhouse gases35. The model with a horizontal resolution (mean grid interval) of about 112 km used in this study was driven by nudging horizontal winds towards the data of the Japanese 55-year Reanalysis (JRA-55)36. For the FFCO2 flux maps, we used the Open-source Data Inventory for Anthropogenic CO2 (ODIAC) of version 2019 (ODIAC2019), which is a global high-resolution FFCO2 data product37,38. We used monthly mean air-sea CO2 flux maps prepared by the Japan Meteorological Agency (JMA)39,40 and monthly biomass burning CO2 flux maps from the Global Fire Emission Database version 4 s (GFED4s, van der Werf et al. 2017)41. Note that we used the latest flux maps from ODIAC (2018), JMA (2018), and GFED4s (2016) to simulate the atmospheric CO2 mole fractions in 2019 and 2020. In addition, we used averaged monthly biospheric CO2 flux maps based on the inversion flux dataset during 2006–2008. As for the CH4 flux maps, we used an inversion flux dataset derived from the NICAM-TM 4D-var system42,43. These inversion flux maps were prepared to participate in a multi-disciplinary study aimed at global CH4 budget estimation under the Global Carbon Project (GCP)44. Similarly, the monthly CH4 flux maps for 2017 were used to simulate the atmospheric CH4 variation in 2020 because of the limitation of the inversion duration. The simulated synoptic-scale variations in CO2 and CH4 are compared with the observed variations in Fig. S10 in Supplementary Material, where detrended data are plotted.

Estimation of the relationship between FFCO2 emissions and ΔCO2/ΔCH4 ratio

The relationship between the FFCO2 emissions from China and the ΔCO2/ΔCH4 ratios at HAT was estimated based on the NICAM-TM simulation. In addition to the simulation described in the above section (control case), we conducted three simulations for the three reduction cases (lower, moderate, and upper cases) of the FFCO2 emissions from China2. Furthermore, to cover the range of the observed ΔCO2/ΔCH4 changes, we conducted another two simulations for extreme FFCO2 reduction cases, in which the FFCO2 reduction rates were set to 125% and 150% of those for the upper case. Table S1 in Supplementary Material lists the monthly averages of the decreases in the FFCO2 emissions and the corresponding ΔCO2/ΔCH4 ratios based on the simulations for February and March 2020.

Data availability

Continuous observations of CO2 and CH4 time series are available through the NIES database (website). https://db.cger.nies.go.jp/portal/geds/atmosphericAndOceanicMonitoring?lang=eng.

References

  1. 1.

    Myllyvirta, L. Analysis: Coronavirus temporarily reduced China’s CO2 emissions by a quarter. CarbonBrief (2020). https://www.carbonbrief.org/analysis-coronavirus-has-temporarily-reduced-chinas-co2-emissions-by-a-quarter. Accessed 19 Feb 2020.

  2. 2.

    Le Quéré, C. et al. Temporary reduction in daily global CO2 emissions during the COVID-19 forced confinement. Nat. Clim. Change https://doi.org/10.1038/s41558-020-0797-x (2020).

    Article  Google Scholar 

  3. 3.

    National Aeronautics and Space Administration. Airborne nitrogen dioxide plummets over China. NASA: Earth Observatory 1–5 (2020). https://www.earthobservatory.nasa.gov/images/146362/airborne-nitrogen-dioxide-plummets-over-china. Accessed 15 May 2020.

  4. 4.

    ESA. COVID-19: Nitrogen Dioxide Over China (2020).

  5. 5.

    Keeling, C. D. The concentration and isotopic abundances of carbon dioxide in the atmosphere. Tellus 12, 200–203 (1960).

    ADS  Article  Google Scholar 

  6. 6.

    Dlugokencky, E. J. NOAA/ESRL. NOAA/ESRL (2019). ftp://aftp.cmdl.noaa.gov/data/greenhouse_gases/ch4.

  7. 7.

    Keeling, C. D., Piper, S. C., Whorf, T. P. & Keeling, R. F. Evolution of natural and anthropogenic fluxes of atmospheric CO2 from 1957 to 2003. Tellus B Chem. Phys. Meteorol. 63, 1–22 (2011).

    ADS  CAS  Article  Google Scholar 

  8. 8.

    Francey, R. J., Frederiksen, J. S., Paul Steele, L. & Langenfelds, R. L. Variability in a four-network composite of atmospheric CO2 differences between three primary baseline sites. Atmos. Chem. Phys. 19, 14741–14754 (2019).

    ADS  CAS  Article  Google Scholar 

  9. 9.

    Mukai, H. et al. Characterization of atmospheric CO2 observed at two-background air monitoring stations (Hateruma and Ochi-ishi) in Japan. in Sixth International Carbon Dioxide Conference (ed. Nakazawa, T.) (2001).

  10. 10.

    Tohjima, Y. Analysis and presentation of in situ atmospheric methane measurements from Cape Ochi-ishi and Hateruma Island. J. Geophys. Res. 107, 4148 (2002).

    Article  Google Scholar 

  11. 11.

    Terao, Y. et al. Interannual variability and trends in atmospheric methane over the western Pacific from 1994 to 2010. J. Geophys. Res. 116, D14303 (2011).

    ADS  Article  Google Scholar 

  12. 12.

    Tohjima, Y. et al. Analysis of seasonality and annual mean distribution of atmospheric potential oxygen (APO) in the Pacific region. Glob. Biogeochem. Cycles https://doi.org/10.1029/2011GB004110 (2012).

    Article  Google Scholar 

  13. 13.

    Machida, T. et al. Worldwide measurements of atmospheric CO2 and other trace gas species using commercial airlines. J. Atmos. Ocean. Technol. 25, 1744–1754 (2008).

    ADS  Article  Google Scholar 

  14. 14.

    Umezawa, T. et al. Statistical characterization of urban CO2 emission signals observed by commercial airliner measurements. Sci. Rep. 10, 7963 (2020).

    ADS  CAS  PubMed  PubMed Central  Article  Google Scholar 

  15. 15.

    Yokota, T. et al. Global concentrations of CO2 and CH4 retrieved from GOSAT: First preliminary results. SOLA 5, 160–163 (2009).

    ADS  Article  Google Scholar 

  16. 16.

    Yoshida, Y. et al. Improvement of the retrieval algorithm for GOSAT SWIR XCO2 and XCH4 and their validation using TCCON data. Atmos. Meas. Tech. 6, 1533–1547 (2013).

    Article  Google Scholar 

  17. 17.

    Tohjima, Y. et al. Temporal changes in the emissions of CH4 and CO from China estimated from CH4/CO2 and CO/CO2 correlations observed at Hateruma Island. Atmos. Chem. Phys. 14, 1663–1677 (2014).

    ADS  Article  Google Scholar 

  18. 18.

    Tohjima, Y., Mukai, H., Hashimoto, S. & Patra, P. K. Increasing synoptic scale variability in atmospheric CO2 at Hateruma Island associated with increasing East-Asian emissions. Atmos. Chem. Phys. 10, 453–462 (2010).

    ADS  CAS  Article  Google Scholar 

  19. 19.

    IEA. CO2 emissions from fuel combustion. (2020). https://webstore.iea.org/co2-emissions-from-fuel-combustion-2019-highlights. Accessed 15 May 2020.

  20. 20.

    Crippa, M. et al. High resolution temporal profiles in the emissions database for global atmospheric research. Sci. Data 7, 121 (2020).

    PubMed  PubMed Central  Article  Google Scholar 

  21. 21.

    Friedlingstein, P. et al. Global carbon budget 2019. Earth Syst. Sci. Data 11, 1783–1838 (2019).

    ADS  Article  Google Scholar 

  22. 22.

    Bergamaschi, P. et al. Atmospheric CH4 in the first decade of the 21st century: Inverse modeling analysis using SCIAMACHY satellite retrievals and NOAA surface measurements. J. Geophys. Res. Atmos. 118, 7350–7369 (2013).

    ADS  CAS  Article  Google Scholar 

  23. 23.

    Thompson, R. L. et al. Methane emissions in East Asia for 2000–2011 estimated using an atmospheric Bayesian inversion. J. Geophys. Res. Atmos. 120, 4352–4369 (2015).

    ADS  CAS  Article  Google Scholar 

  24. 24.

    Saeki, T. & Patra, P. K. Implications of overestimated anthropogenic CO2 emissions on East Asian and global land CO2 flux inversion. Geosci. Lett. 4, 9 (2017).

    ADS  Article  Google Scholar 

  25. 25.

    Thoning, K. W., Tans, P. P. & Komhyr, W. D. Atmospheric carbon dioxide at Mauna Loa observatory: 2. Analysis of the NOAA GMCC data, 1974–1985. J. Geophys. Res. Atmos. 94, 8549–8565 (1989).

    ADS  CAS  Article  Google Scholar 

  26. 26.

    Niwa, Y., Tomita, H., Satoh, M. & Imasu, R. A three-dimensional icosahedral grid advection scheme preserving monotonicity and consistency with continuity for atmospheric tracer transport. J. Meteorol. Soc. Japan 89, 255–268 (2011).

    Article  Google Scholar 

  27. 27.

    UN. Record global carbon dioxide concentrations despite COVID-19 crisis. (2020).

  28. 28.

    Patra, P. K., Maksyutov, S. & Nakazawa, T. Analysis of atmospheric CO2 growth rates at Mauna Loa using CO2 fluxes derived from an inverse model. Tellus B Chem. Phys. Meteorol. 57, 357–365 (2005).

    ADS  Article  Google Scholar 

  29. 29.

    O’Dell, C. W. et al. Improved retrievals of carbon dioxide from orbiting carbon observatory-2 with the version 8 ACOS algorithm. Atmos. Meas. Tech. 11, 6539–6576 (2018).

    Article  Google Scholar 

  30. 30.

    Machida, T., Tohjima, Y., Katsumata, K. & Mukai, H. A new CO2 calibration scale based on gravimetric one-step dilution cylinders in National Institute for Environmental Studies- NIES09 CO2 scale. in Report of the 15th WMO Meeting of Experts on Carbon Dioxide Concentration and Related Tracer Measurement Techniques (ed. Brand, W. A.) 165–169 (WMO/GAW, 2011).

  31. 31.

    Pickers, P. A. & Manning, A. C. Investigating bias in the application of curve fitting programs to atmospheric time series. Atmos. Meas. Tech. 8, 1469–1489 (2015).

    CAS  Article  Google Scholar 

  32. 32.

    Hirsch, R. M. & Gilroy, E. J. Methods of fitting a straight line to data: Examples in water resources. J. Am. Water Resour. Assoc. 20, 705–711 (1984).

    ADS  CAS  Article  Google Scholar 

  33. 33.

    Shan, Y. et al. China CO2 emission accounts 1997–2015. Sci. Data 5, 170201 (2018).

    CAS  PubMed  PubMed Central  Article  Google Scholar 

  34. 34.

    Satoh, M. et al. the non-hydrostatic icosahedral atmospheric model: Description and development. Prog. Earth Planet. Sci. 1, 18 (2014).

    ADS  Article  Google Scholar 

  35. 35.

    Niwa, Y. et al. Imposing strong constraints on tropical terrestrial CO2 fluxes using passenger aircraft based measurements. J. Geophys. Res. Atmos. https://doi.org/10.1029/2012JD017474 (2012).

    Article  Google Scholar 

  36. 36.

    Kobayashi, S. et al. The JRA-55 reanalysis: General specifications and basic characteristics. J. Meteorol. Soc. Japan. Ser. II 93, 5–48 (2015).

    Article  Google Scholar 

  37. 37.

    Oda, T. & Maksyutov, S. A very high-resolution (1 km×1 km) global fossil fuel CO2 emission inventory derived using a point source database and satellite observations of nighttime lights. Atmos. Chem. Phys. 11, 543–556 (2011).

    ADS  CAS  Article  Google Scholar 

  38. 38.

    Oda, T., Maksyutov, S. & Andres, R. J. The open-source data inventory for anthropogenic CO2, version 2016 (ODIAC2016): A global monthly fossil fuel CO2 gridded emissions data product for tracer transport simulations and surface flux inversions. Earth Syst. Sci. Data 10, 87–107 (2018).

    ADS  PubMed  PubMed Central  Article  Google Scholar 

  39. 39.

    Takatani, Y. et al. Relationships between total alkalinity in surface water and sea surface dynamic height in the Pacific Ocean. J. Geophys. Res. 119, 2806–2814 (2014).

    ADS  Article  Google Scholar 

  40. 40.

    Iida, Y. et al. Trends in pCO2 and sea-air CO2 flux over the global open oceans for the last two decades. J. Oceanogr. 71, 637 (2015).

    CAS  Article  Google Scholar 

  41. 41.

    van der Werf, G. R. et al. Global fire emissions estimates during 1997–2016. Earth Syst. Sci. Data 9, 697–720 (2017).

    ADS  Article  Google Scholar 

  42. 42.

    Niwa, Y. et al. A 4D-Var inversion system based on the icosahedral grid model (NICAM-TM 4D-Var v1.0)—Part 1: Offline forward and adjoint transport models. Geosci. Model Dev. 10, 1157–1174 (2017).

    ADS  CAS  Article  Google Scholar 

  43. 43.

    Niwa, Y. et al. A 4D-Var inversion system based on the icosahedral grid model (NICAM-TM 4D-Var v1.0)—Part 2: Optimization scheme and identical twin experiment of atmospheric CO2 inversion. Geosci. Model Dev. 10, 2201–2219 (2017).

    ADS  CAS  Article  Google Scholar 

  44. 44.

    Saunois, M. et al. The global methane budget 2000–2017. Earth Syst. Sci. Data 12, 1561–1623 (2020).

    ADS  Article  Google Scholar 

Download references

Acknowledgements

We gratefully acknowledge the members of the Global Environment Forum, the staff of the Center for Global Environmental Research, and the local staff for their continued support in maintaining the in-situ measurements of CO2 and CH4 at HAT. The NICAM-TM simulations were performed using the NIES supercomputer system (NEC SX-Aurora). This study was supported by funds provided by the Environment Research and Technology Development Fund (JPMEERF20172010 and JPMEERF20172001) of the Environmental Restoration and Conservation Agency of Japan, and the Global Environmental Research Coordinate System from the Ministry of the Environment, Japan (FY2014, FY2019).

Author information

Affiliations

Authors

Contributions

Y.T. conducted the analysis, Y.T., T.M., H.M. and M.S. conducted the measurements, Y.N., Y.T. and P.K.P. developed the analysis strategy. All authors participated in the discussions and preparation of the manuscript.

Corresponding author

Correspondence to Yasunori Tohjima.

Ethics declarations

Competing interests

The authors declare no competing interests.

Additional information

Publisher's note

Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.

Supplementary information

Rights and permissions

Open Access This article is licensed under a Creative Commons Attribution 4.0 International License, which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence, and indicate if changes were made. The images or other third party material in this article are included in the article's Creative Commons licence, unless indicated otherwise in a credit line to the material. If material is not included in the article's Creative Commons licence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this licence, visit http://creativecommons.org/licenses/by/4.0/.

Reprints and Permissions

About this article

Verify currency and authenticity via CrossMark

Cite this article

Tohjima, Y., Patra, P.K., Niwa, Y. et al. Detection of fossil-fuel CO2 plummet in China due to COVID-19 by observation at Hateruma. Sci Rep 10, 18688 (2020). https://doi.org/10.1038/s41598-020-75763-6

Download citation

Further reading

Comments

By submitting a comment you agree to abide by our Terms and Community Guidelines. If you find something abusive or that does not comply with our terms or guidelines please flag it as inappropriate.

Search

Quick links

Nature Briefing

Sign up for the Nature Briefing newsletter — what matters in science, free to your inbox daily.

Get the most important science stories of the day, free in your inbox. Sign up for Nature Briefing