Quantifying atmospheric CO2 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 CO2 (XCO2) retrievals from Greenhouse Gases Observing Satellite (GOSAT) and OCO‐2 satellites. The best model (R2 = 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 XCO2 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.
Plain Language Summary
CO2 sources and sinks are primarily regulated by anthropogenic emissions, photosynthesis and respiration on land and in the ocean, as well as by physical dissolution and carbonate chemistry with ocean circulation. The consistent long‐term quantification of atmospheric CO2 concentrations using satellite observations plays a pivotal role in understanding the response of global and regional carbon cycles to climate change. However, satellites have a revisit period, and factors like cloud and aerosol scattering impact the quality and quantity of their observations. A single satellite currently falls short of the demand to monitor global carbon sources and sinks, necessitating the integration of observations from various satellites to conduct carbon flux inversions. Different satellites come with distinct sampling patterns, instrument parameters, and retrieval algorithms, which leads to biases in their retrieval products. Our study, focusing on OCO‐2 and GOSAT, employs machine learning models to improve consistency between retrievals derived from these two satellites, thus generating a harmonized data set. The bias‐corrected GOSAT XCO2 retrievals exhibit high spatiotemporal consistency with OCO‐2 XCO2 retrievals, immensely enhancing the observational constraints for carbon flux inversions. This method holds promise for application to recently launched and future satellites, aiming to offer carbon flux inversions with amplified spatiotemporal observational constraints.
Key Points
This study employs machine learning (ML) models to enhance XCO2 consistency between OCO‐2 and Greenhouse Gases Observing Satellite (GOSAT), reducing monthly inconsistency by 71.5%
Integrating the OCO‐2 data set with GOSAT retrievals increased yearly observations by 56.2%
Fusing satellite data through ML models can pave the way for improved carbon flux inversions in the future
Carbon dioxide (CO2) is the most important long-lived greenhouse gas and can be retrieved using solar absorption spectra recorded by a ground-based Fourier-transform infrared spectrometer (FTIR). In ...this study, we investigate the CO2 retrieval strategy using the Network for the Detection of Atmospheric Composition Change–Infrared Working Group (NDACC–IRWG) type spectra between August 2018 and April 2022 (~4 years) at Xianghe, China, aiming to find the optimal observed spectra, retrieval window, and spectroscopy. Two spectral regions, near 2600 and 4800 cm−1, are analyzed. The differences in column-averaged dry-air mole fraction of CO2 (XCO2) derived from spectroscopies (ATM18, ATM20, HITRAN2016, and HITRAN2020) can be up to 1.65 ± 0.95 ppm and 7.96 ± 2.02 ppm for NDACC-type 2600 cm−1 and 4800 cm−1 retrievals, respectively, which is mainly due to the CO2 differences in air-broadened Lorentzian HWHM coefficient (γair) and line intensity (S). HITRAN2020 provides the best fitting, and the retrieved CO2 columns and profiles from both 2600 and 4800 cm−1 are compared to the co-located Total Column Carbon Observing Network (TCCON) measurements and the greenhouse gas reanalysis dataset from the Copernicus Atmosphere Monitoring Service (CAMS). The amplitude of XCO2 seasonal variation derived from the NDACC-type (4800 cm−1) is closer to the TCCON measurements than that from the NDACC-type (2600 cm−1). Moreover, the NDACC-type (2600 cm−1) retrievals are strongly affected by the a priori profile. For tropospheric XCO2, the correlation coefficient between NDACC-type (4800 cm−1) and CAMS model is 0.73, which is higher than that between NDACC-type (2600 cm−1) and CAMS model (R = 0.56).
Methane (CH4) is an important greenhouse as well as a chemically active gas. Accurate monitoring and understanding of its spatiotemporal distribution are crucial for effective mitigation strategies. ...Nowadays, satellite measurements are widely used for CH4 studies. Here, we use the CH4 products from four commonly used satellites (GOSAT, TROPOMI, ARIS, and IASI) during the period from 2018 to 2020 to investigate the spatiotemporal variations of CH4 in China. In spite of the same target (CH4) for the four satellites, differences among them exist in terms of the instrument, spectrum, and retrieval algorithm. The GOSAT and TROPOMI CH4 retrievals use shortwave infrared spectra, with a better sensitivity near the surface, while the IASI and AIRS CH4 retrievals use thermal infrared spectra, showing a good sensitivity in the mid–upper troposphere but a weak sensitivity in the lower troposphere. The GOSAT and TROPOMI observe high CH4 concentrations in the east and south and low concentrations in the west and north, which is highly related to the CH4 emissions. The IASI and AIRS show a more uniform CH4 distribution over China, which reflects the variation of CH4 at a high altitude. However, a large discrepancy is observed between the IASI and AIRS despite using a similar retrieval band, e.g., significant differences in the seasonal variations of CH4 are observed between the IASI and AIRS across several regions in China. This study highlights the CH4 differences observed by the four satellites in China, and caution must be taken when using these satellite products.
Display omitted
•A deep-learning approach is developed for mapping satellite-observed XCO2.•State-of-the-art deep-learning techniques are used in the model training.•The approach does not rely on ...model XCO2 simulations for model training.•The mapping XCO2 data effectively capture fine-scale CO2 variabilities.
Satellite measurements of the column-averaged dry air mole fraction of atmospheric carbon dioxide (XCO2) play a crucial role in monitoring CO2 emissions and sinks. However, the current limitations of satellite observations, including sparse sampling, narrow swath coverage, and data gaps caused by factors like clouds, significantly hinder their ability to accurately capture local-scale CO2 sources and sinks. This study introduces an innovative data-driven approach based on deep learning, which takes into consideration both spatial and temporal variations, to map XCO2 using observations from multiple satellites. By leveraging advanced deep learning techniques like conventional neural network (CNN), long short-term memory network (LSTM), channel-spatial attention, and artificial neural network (ANN) in the model training, this approach not only incorporates spatiotemporal variations of XCO2 but also integrates information from related terrestrial, anthropogenic, and meteorological variables. The results demonstrate a notable improvement in the predictive capability of the approach. An important advancement over previous studies is that this approach breaks away from the conventional practice of using model-generated XCO2 simulations for both training and validation. Monthly deep-learning XCO2 (DL-XCO2) of China from 2014 to 2022 was generated with a spatial resolution of 0.25° based on satellite retrievals from GOSAT and OCO-2/3. Cross-validation results show an average prediction bias of −0.16 ppm. Additionally, DL-XCO2 exhibits high precision when compared to two TCCON stations, with errors of 0.93 and 1.29 ppm in Hefei and Xianghe, respectively. Ultimately, the DL-XCO2 data effectively capture urban CO2 variations, showcasing the potential in accurately characterizing fine-scale CO2 sources and sinks.
The seasonal evolution of O3 and its photochemical
production regime in a polluted region of eastern China between 2014 and 2017
has been investigated using observations. We used tropospheric ozone
...(O3), carbon monoxide (CO), and formaldehyde (HCHO, a marker of VOCs
(volatile organic compounds)) partial columns derived from high-resolution
Fourier transform spectrometry (FTS); tropospheric nitrogen dioxide
(NO2, a marker of NOx (nitrogen oxides)) partial
column deduced from the Ozone Monitoring Instrument (OMI); surface meteorological
data; and a back trajectory cluster analysis technique. A broad O3
maximum during both spring and summer (MAM/JJA) is observed; the day-to-day
variations in MAM/JJA are generally larger than those in autumn and winter
(SON/DJF). Tropospheric O3 columns in June are 1.55×1018 molecules cm−2 (56 DU (Dobson units)), and in December they are
1.05×1018 molecules cm−2 (39 DU). Tropospheric O3
columns in June were ∼50 % higher than those in December. Compared
with the SON/DJF season, the observed tropospheric O3 levels in MAM/JJA
are more influenced by the transport of air masses from densely populated and
industrialized areas, and the high O3 level and variability in
MAM/JJA is determined by the photochemical O3 production. The
tropospheric-column HCHO∕NO2 ratio is used as a proxy to
investigate the photochemical O3 production rate (PO3).
The results show that the PO3 is mainly nitrogen oxide (NOx) limited in MAM/JJA, while it is mainly VOC or mixed VOC–NOx limited in SON/DJF. Statistics show that
NOx-limited, mixed VOC–NOx-limited, and VOC-limited PO3 accounts for 60.1 %, 28.7 %, and 11 % of days,
respectively. Considering most of PO3 is NOx
limited or mixed VOC–NOx limited, reductions in
NOx would reduce O3 pollution in eastern China.
TanSat-2, the next-generation Chinese greenhouse gas monitoring satellite for measuring carbon dioxide (CO2), has a new city-scale observing mode. We assess the theoretical capability of TanSat-2 to ...quantify integrated urban CO2 emissions over the cities of Beijing, Jinan, Los Angeles, and Paris. A high-resolution emission inventory and a column-averaged CO2 (XCO2) transport model are used to build an urban CO2 inversion system. We design a series of numerical experiments describing this observing system to evaluate the impacts of sampling patterns and XCO2 measurement errors on inferring urban CO2 emissions. We find that the correction in systematic and random flux errors is correlated with the signal-to-noise ratio of satellite measurements. The reduction in systematic flux errors for the four cities are sizable, but are subject to unbiased satellite sampling and favorable meteorological conditions (i.e., less cloud cover and lower wind speed). The corresponding correction to the random flux error is 19–28%. Even though clear-sky satellite data from TanSat-2 have the potential to reduce flux errors for cities with high CO2 emissions, quantifying urban emissions by satellite-based measurements is subject to additional limitations and uncertainties.
The nationwide lockdown due to the COVID-19 pandemic in 2020 reduced industrial and human activities in China. In this study, we investigate atmospheric carbon monoxide (CO) concentration changes ...during the lockdown from observations at the surface and from two satellites (TROPOspheric Monitoring Instrument (TROPOMI) and Infrared Atmospheric Sounding Interferometer (IASI)). It is found that the average CO surface concentration in 2020 was close to that in 2019 before the lockdown, and became 18.7% lower as compared to 2019 during the lockdown. The spatial variation of the change in the CO surface concentration is high, with an 8–27% reduction observed for Beijing, Shanghai, Chengdu, Zhengzhou, and Guangzhou, and almost no change in Wuhan. The TROPOMI and IASI satellite observations show that the CO columns decreased by 2–13% during the lockdown in most regions in China. However in South China, there was an 8.8% increase in the CO columns observed by TROPOMI and a 36.7% increase observed by IASI, which is contrary to the 23% decrease in the surface CO concentration. The enhancement of the CO column in South China is strongly affected by the fire emissions transported from Southeast Asia. This study provides an insight into the impact of COVID-19 on CO concentrations both at the surface and in the columns in China, and it can be extended to evaluate other areas using the same approach.
Since its launch on 13 October 2017, the TROPOspheric Monitoring Instrument (TROPOMI) on board the Sentinel-5 Precursor (S5P) mission has been measuring the solar radiation backscattered by Earth’s ...atmosphere and surface. In this study, we evaluate the TROPOMI operational methane (CH4) and carbon monoxide (CO) products’ performance results covering about 3 years using the only two global Total Carbon Column Observing Network (TCCON) sites in China, i.e., the Hefei site and the Xianghe site. These two sites have recently joined the TCCON, and this study uses the both sites simultaneously to validate the TROPOMI products over China for the first time. We found that the systematic bias with rescaling between the TROPOMI CO products and the Hefei site is on average 1.78 ± 6.35 ppb or 1.18 ± 5.35%. The systematic bias with rescaling between the TROPOMI CO products and the Xianghe site is on average 5.33 ± 14.24 ppb or 3.85 ± 10.30%. Both the stations show a correlation above 0.9. The TROPOMI CO data are systematically higher than the two TCCON sites measurements in China. We found that the systematic bias with rescaling between the TROPOMI CH4 products and the Hefei site is on average −4.13 ± 11.65 ppb or −0.22 ± 0.62%. The systematic bias between the TROPOMI CH4 products and Xianghe site is on average −7.25 ± 10.72 ppb or −0.39 ± 0.57%. Both the stations show a correlation above 0.9. The TROPOMI CH4 data are systematically lower than the two TCCON sites measurements in China. We found that the bias between the TROPOMI and the two sites’ data as a function of the coincident radius around the two sites is mostly affected by localized emissions for both CO and CH4. We also observe a CO decreasing trend and a CH4 increasing trend in the year-to-year relative changes from 2019 to 2021. Validating against reference from Hefei and Xianghe TCCON site demonstrates the high quality of TROPOMI CO and CH4 data over China.
Carbon dioxide (CO2) is the most important greenhouse gas and several satellites have been launched to monitor the atmospheric CO2 at regional and global scales. Evaluation of the measurements ...obtained from these satellites against accurate and precise instruments is crucial. In this work, aircraft measurements of CO2 were carried out over Qinhuangdao, China (39.9354°N, 119.6005°E), on 14, 16, and 19 March 2019 to validate the Greenhous gases Observing SATellite (GOSAT) and the Orbiting Carbon Observatory 2 (OCO-2) CO2 retrievals. The airborne in situ instruments were mounted on a research aircraft and the measurements were carried out between the altitudes of ~0.5 and 8.0 km to obtain the vertical profiles of CO2. The profiles captured a decrease in CO2 concentration from the surface to maximum altitude. Moreover, the vertical profiles from GEOS-Chem and the National Oceanic and Atmospheric Administration (NOAA) CarbonTracker were also compared with in situ and satellite datasets. The satellite and the model datasets captured the vertical structure of CO2 when compared with in situ measurements, which showed good agreement among the datasets. The dry-air column-averaged CO2 mole fractions (XCO2) retrieved from OCO-2 and GOSAT showed biases of 1.33 ppm (0.32%) and −1.70 ppm (−0.41%), respectively, relative to the XCO2 derived from in situ measurements.
As an important greenhouse gas (GHG) in the atmosphere, carbon dioxide (CO 2 ) has a great impact on global climate change. Accurate knowledge of the spatiotemporal variations of CO 2 is of great ...significance for understanding the carbon cycle and evaluating the effectiveness of carbon emission reduction. In recent years, several satellites with CO 2 sensors have been launched and a series of atmospheric CO 2 concentration products have been developed using different retrieval algorithms. This study validated nine satellite XCO 2 products derived from Greenhouse gases Observing SATellite (GOSAT), GOSAT-2, Orbiting Carbon Observatory-2 (OCO-2), and OCO-3: including ACOS-GOSAT, NIES-GOSAT, BESD-GOSAT, OCFP-GOSAT, SRFP-GOSAT, EMMA, GOSAT-2, OCO-2, and OCO-3 XCO 2 . The remotely sensed XCO 2 products were compared with the XCO 2 observations from six Total Carbon Column Observing Network (TCCON) stations in East Asia for validation. The results showed that the OCO-2 XCO 2 product outperformed other products, with the highest R 2 of 0.94 and the lowest MAE of 1.24 ppm. The ACOS-GOSAT and EMMA-GOSAT XCO 2 products also showed favorable accuracies, both achieving R 2 of 0.93 and corresponding MAE values of 1.29 and 1.31 ppm, respectively. The GOSAT-2 XCO 2 product showed the poorest accuracy, with an R 2 of 0.77 and a mean absolute error of 3.28 ppm. There was a significant overestimation of the bias-uncorrected GOSAT-2 XCO 2 product in East Asia, and it indicated that bias correction must be performed for this XCO 2 product. The accuracy of TCCON XCO 2 was not consistent with remotely sensed XCO 2 at different stations. The RJ, JS, AN, and TK TCCON stations generally showed better agreements between satellite estimates and TCCON observations, except for the GOSAT-2 XCO 2 product.