Xiaojuan Lin

Africa and the Middle East contribute significantly to global fossil fuel production, yet the associated carbon emissions from these extraction and processing activities remain insufficiently reported. The current reporting is not only delayed but also heavily reliant on energy consumption statistics that are difficult to verify. Here, we introduce an alternative approach to indirectly infer and map anthropogenic carbon dioxide (CO ₂ ) emissions using the coemitted nitrogen oxides (NO _x ) emissions derived from TROPOMI satellite observations. We find that current global inventories underestimate total CO ₂ emissions by about 30% for Africa and the Middle East. About 35% of this underestimation is related to newly identified CO ₂ emission sources, which are mainly attributable to oil and gas infrastructure. The analysis reveals that these underestimated emissions could lead to a delay of ∼5 years in achieving the 2030 Nationally Determined Contribution targets in Africa and the Middle East. This study provides an effective tool for constructing an independent carbon emission accounting framework based on satellite observations and offers a scientific database for guiding climate action.

Since anthropogenic NOx (NOx=NO+NO2) and CO2 are co-emitted species for anthropogenic sources, some studies have used the NOx emissions retrieved from satellite observations to infer the anthropogenic CO2 emissions. However, these studies did not consider the fact that satellites measure total NO2 concentrations and their inferred emissions encompass both biogenic and anthropogenic sources. In this study, we introduce a method to distinguish soil NOx emissions from satellite-based total NOx emissions. The total NOx emissions are derived by the state-of-the-art inverse algorithm DECSO (Daily Emission estimation Constrained by Satellite Observations, Mijling and van der A, 2012; Ding et al., 2017a) from TROPOMI observations. Using the characteristic seasonal cycle of soil emissions we derive these emissions for representative regions with only biogenic emissions, which are then applied to nearby regions according land-use fractions. To evaluate this approach, we compared the deviation between the tropospheric NO2 concentration observed by satellite and two atmospheric composition model simulations: one using the satellite-derived soil NOx emissions and another with the Copernicus Atmosphere Monitoring Service (CAMS) global soil emissions inventory (CAMS-GLOB-SOIL). Once the soil NOx emissions are derived, they can be subtracted from the total emissions to get anthropogenic NOx emissions. Subsequently anthropogenic CO2 emissions can be estimated using known CO2/NOx factors from bottom-up inventories. The annual CO2 emissions derived from DECSO (called DECSO-CO2) in our study area (large part of Europe) is 3.7 Gt in 2019, which is comparable with the 3.2 Gt of the CAMS CO2 inventory (called CAMS-CO2). The DECSO-CO2 and CAMS-CO2 are comparable for most large sources and cities, but the DECSO-CO2 show a larger number of low emission spots than the CAMS-CO2. The results demonstrate the potential for DECSO to expand its application to other regions in the world with less information on anthropogenic CO2 emissions.

We introduce an innovative method to distinguish soil nitrogen oxides (NO_x = NO + NO₂) emissions from satellite‐based total NO_x emissions. To evaluate the approach, we compare the deviation between the tropospheric NO₂ concentration observed by satellite and two atmospheric composition model simulations driven by the newly estimated soil NO_x emissions and the Copernicus Atmosphere Monitoring Service (CAMS) inventory. The estimated average soil NO_x emissions in Europe are 1.7 kg N ha⁻¹ yr⁻¹ in 2019, and the annual soil NO_x emissions are approximately 1.7 times larger than that of the CAMS inventory. The discrepancy originates mainly from the forests, which would mean that the soil NO_x emissions over forest areas in Europe are currently underestimated by the CAMS inventory. The model evaluation indicates that the simulations driven by DECSO‐soil emissions performed significantly better than using CAMS‐soil. Overall, the simulated RMSE% of DECSO‐soil is lower than that of CAMS‐soil, approximately 6% lower in spring and 2% lower in autumn. Our method can easily be extended to other regions in the world despite having monthly variations that are very different from those in Europe.

Quantifying atmospheric CO ₂ over long periods from space is crucial in understanding the carbon cycle's response to climate change. However, a single satellite offers limited spatiotemporal coverage, making comprehensive monitoring challenging. Moreover, biases among various satellite retrievals hinder their direct integration. This study proposed a machine learning framework for fusing the column‐averaged dry‐air mole fraction of CO ₂ (XCO ₂ ) retrievals from Greenhouse Gases Observing Satellite (GOSAT) and OCO‐2 satellites. The best model ( R ² = 0.85) presented improved consistency of GOSAT retrievals by reducing 71.5% of the average monthly bias while using OCO‐2 retrievals as a benchmark, indicating the fusion data set's potential to enhance observation coverage. Incorporating the adjusted GOSAT XCO ₂ retrievals into the OCO‐2 data set added an average of 84.7 thousand observations annually, enhancing the yearly temporal coverage by 53.6% (from 14 to 21.5 days per grid). This method can be adapted to other satellites, maximizing satellite resources for a more robust carbon flux inversion.

We introduce an innovative method to distinguish soil nitrogen oxides (NOx=NO+NO2) emissions from satellite-based total NOx emissions using its seasonal characteristics. To evaluate the approach, we compare the deviation between the tropospheric NO2 concentration observed by satellite and two atmospheric composition model simulations driven by the newly estimated soil NOx emissions and the Copernicus Atmosphere Monitoring Service (CAMS) inventory. The estimated average soil NOx emissions in Europe are 2.5 kg N ha-1 yr-1 in 2019, and the annual soil NOx emissions is approximately 2.5 times larger than that of the CAMS inventory. Our method can easily be extended to other regions at middle or high latitudes with similar seasonal characteristics of soil emissions. The soil emissions are subtracted from the total NOx emissions yielding realistic anthropogenic NOx emissions. We further show this also yields realistic anthropogenic CO2 emissions using known CO2/NOx factors from bottom-up inventories.

Top-down CO₂ emission estimates based on satellite observations are of great importance for independently verifying the accuracy of reported emissions and emission inventories. Difficulties in verifying these satellite-derived emissions arise from the fact that emission inventories often provide annual mean emissions, while estimates from satellites are available only for a limited number of overpasses. Previous studies have derived CO₂ emissions for power plants from the Orbiting Carbon Observatory-2 and 3 (OCO-2 and OCO-3) satellite observations of their exhaust plumes, but the accuracy and the factors affecting these emissions are uncertain. Here we advance monitoring and quantifying point source carbon emissions by focusing on how to improve the accuracy of carbon emission using different wind data estimates. We have selected only isolated power plants for this study, to avoid complications linked to multiple sources in close proximity. We first compared the Gaussian plume model and cross-sectional flux methods for estimating CO₂ emission of power plants. Then we examined the sensitivity of the emission estimates to possible choices for the wind field. For verification we have used power plant emissions that are reported on an hourly basis by the Environmental Protection Agency (EPA) in the US. By using the OCO-2 and OCO-3 observations over the past 4 years we identified emission signals of isolated power plants and arrived at a total of 50 collocated cases involving 22 power plants. We correct for the time difference between the moment of the emission and the satellite observation. We found the wind field halfway the height of the planetary boundary layer (PBL) yielded the best results. We also found that the instantaneous satellite estimated emissions of these 50 cases, and reported emissions display a weak correlation (R²=0.12). The correlation improves with averaging over multiple observations of the 22 power plants (R²=0.40). The method was subsequently applied to 106 power plant cases worldwide and yielded a total emission of 1522 ± 501 Mt CO₂ yr⁻¹, estimated to be about 17 % of the power sector emissions of our selected countries. The improved correlation highlights the potential for future planned satellite missions with a greatly improved coverage to monitor a significant fraction of global power plant emissions.

Top-down CO₂ emission estimates based on satellite observations are potentially of great importance for independently verifying the accuracy of reported emissions and emission inventories. Difficulties in verifying these satellite-derived emissions arise from the fact that emission inventories often provide annual mean emissions while estimates from satellites are available only for a limited number of overpasses. Previous studies have derived CO₂ emissions for power plants from OCO-2 and OCO-3 observations of their exhaust plumes, but the accuracy and the factors affecting these emissions are uncertain. We have selected only isolated power plants for this study, to avoid complications link to multiple sources in close proximity. We first compare the Gaussian plume model and cross-sectional flux methods for estimating CO₂ emission of power plants. Then we examined the sensitivity of the emission estimates to possible choices for the wind field. For verification we have used power plant emissions that are reported on an hourly basis by the Environmental Protection Agency (EPA) in the United States. By using the OCO-2 and OCO-3 observations over the past four years we identified emission signals of isolated power plants and arrived at a total of 50 collocated cases involving 22 power plants. We correct for the time difference between the moment of the emission and the satellite observation. We found the wind field halfway the height of planetary boundary layer (PBL) yielded the best results. We found that the instantaneous satellite estimated emissions of these 50 cases and reported emissions display a weak correlation (R²=0.12). The correlation improves with averaging over multiple observations of the 22 power plants (R²=0.40). The method was subsequently applied to 106 power plants cases worldwide yielded a total emission of 1522 ± 501 Mt CO₂/year, estimated to be about 17 % of the power sector emissions of our selected countries. The improved correlation highlights the potential for future planned satellite missions with a greatly improved coverage to monitor a significant fraction of global power plant emissions.

Top-down CO₂ emission estimates based on satellite observations are potentially of great importance for independently verifying the accuracy of reported emissions and emission inventories. Difficulties in verifying these satellite-derived emissions arise from the fact that emission inventories often provide annual mean emissions while estimates from satellites are available only for a limited number of overpasses. Previous studies have derived CO₂ emissions for power plants from OCO-2 and OCO-3 observations of their exhaust plumes, but the accuracy and the factors affecting these emissions are uncertain. We have selected only isolated power plants for this study, to avoid complications link to multiple sources in close proximity. We first compare the Gaussian plume model and cross-sectional flux methods for estimating CO₂ emission of power plants. Then we examined the sensitivity of the emission estimates to possible choices for the wind field. For verification we have used power plant emissions that are reported on an hourly basis by the Environmental Protection Agency (EPA) in the United States. By using the OCO-2 and OCO-3 observations over the past four years we identified emission signals of isolated power plants and arrived at a total of 50 collocated cases involving 22 power plants. We correct for the time difference between the moment of the emission and the satellite observation. We found the wind field halfway the height of planetary boundary layer (PBL) yielded the best results. We found that the instantaneous satellite estimated emissions of these 50 cases and reported emissions display a weak correlation (R²=0.12). The correlation improves with averaging over multiple observations of the 22 power plants (R²=0.40). The method was subsequently applied to 106 power plants cases worldwide yielded a total emission of 1522 ± 501 Mt CO₂/year, estimated to be about 17 % of the power sector emissions of our selected countries. The improved correlation highlights the potential for future planned satellite missions with a greatly improved coverage to monitor a significant fraction of global power plant emissions.

CO2 emissions of power plants are often self-reported and calculated based on fuel consumption. The authenticity of CO2 emission data from power plants are preferable to be verified by independent measurements. Satellite observations can provide these CO2 emission observations from isolated power plants. However, there are two difficulties in the current top-down anthropogenic CO2 emission inversions, that is, (1) the anthropogenic CO2 emission signal is difficult to distinguish from the background of other emitted CO2, and (2) the temporal and spatial resolution of CO2 satellite observation data are currently limited. In this study, we focus on how to improve the accuracy of CO2 emissions using different methods and wind data estimates. We assess these emission estimates by comparison with USA EPA emission data, and identify and explore suitable cases elsewhere in the world. We have selected only isolated power plants for this study, to avoid complications because of multiple sources in close proximity. We first compare the Gaussian plume model and cross-sectional flux methods for estimating CO2 emissions of power plants. Then we examine the sensitivity of the emission estimates to possible choices for the wind field. For verification we have used power plant emissions that are reported on an hourly basis by the Environmental Protection Agency (EPA) in the United States. By using the OCO-2 and OCO-3 observations over the past four years we identified emission signals of isolated power plants and arrived at a total of 50 collocated cases involving 22 power plants. We found the wind field halfway the height of planetary boundary layer (PBL) yielded the best results. We found that the instantaneous satellite estimated emissions of these 50 cases and reported emissions display a weak correlation (R2=0.12). The correlation improves with averaging over multiple observations of the 22 power plants (R2=0.40). The method was subsequently applied to 106 power plants cases worldwide. We demonstrate that accurate estimation of power plant emissions can be achieved by monitoring from future satellite missions with more frequent observations.