Multiannual changes of CO 2 emissions in China: indirect estimates derived from satellite measurements of tropospheric NO 2 columns

E. V. Berezin, I. B. Konovalov, Philippe Ciais, Andreas Richter, Shu Tao, Greet Janssens‐Maenhout, Matthias Beekmann, Ernst‐Detlef Schulze

TL;DR

This study uses satellite measurements of nitrogen dioxide (NO2) as a proxy to estimate changes in China's carbon dioxide (CO2) emissions from 1996-2008. The findings suggest that global emission inventories might overestimate the non-linearity of China's CO2 emission growth, particularly around the year 2000, and highlight significant discrepancies in emission trends for the late 1990s compared to existing bottom-up estimates.

What Problem Does This Paper Address?

Accurate tracking of CO2 emissions is crucial for understanding and mitigating climate change. China, as a major global emitter, presents a particular challenge due to the scale and rapid evolution of its economy. Traditional "bottom-up" emission inventories rely on activity data (like fuel consumption) and emission factors, but these can be prone to significant uncertainties, especially in rapidly developing economies. This paper addresses the need for independent verification of these inventories by leveraging satellite observations of co-emitted gases. Specifically, it investigates whether tropospheric NO2, a gas strongly correlated with fossil fuel combustion (a major source of CO2), can serve as a reliable indicator of CO2 emission trends in China, particularly in identifying potential biases or exaggerations in existing inventory data.

Methods

The core of the methodology involves an "indirect top-down" approach. This means using satellite measurements of NO2 to infer CO2 emissions, rather than directly measuring CO2. The authors employ a simple inverse modeling procedure:

  1. NOx Emission Estimation: They first estimate interannual changes in anthropogenic nitrogen oxide (NOx) emissions. This is done by relating observed tropospheric NO2 columns to simulated NOx emissions and a "background" NO2 level (representing non-anthropogenic sources) using a mesoscale chemistry-transport model (CHIMERE).
  2. CO2 Emission Derivation: The estimated NOx emission changes are then converted into CO2 emission changes using CO2-to-NOx emission ratios derived from established emission inventories (EDGAR and REAS). This conversion factor accounts for the fact that different combustion processes and fuels produce different ratios of CO2 to NOx.
  3. Trend Analysis: Exponential trends are calculated for both the top-down CO2 emission estimates and the bottom-up inventory data for two distinct periods: 1996-2001 and 2001-2008. This split was chosen to investigate a potential "sharp bend" in emission trends observed in some inventories around 2000-2002.

Several "cases" were analyzed to assess the sensitivity of the results to assumptions about background NO2 levels and the NOx-to-CO2 conversion factors.

Datasets

  • Satellite Data:
    • GOME (Global Ozone Monitoring Experiment) tropospheric excess NO2 columns (1996-2002).
  • SCIAMACHY (Scanning Imaging Absorption Spectrometer for Atmospheric Chemistry) tropospheric excess NO2 columns (2003-2008).
  • Emission Inventories (for NOx and CO2):
    • EDGAR v4.2: Emission Database for Global Atmospheric Research (global, gridded and national data).
  • GCP: Global Carbon Project (national totals for CO2).
  • REAS (v1.11 and v2.1): Regional Emission Inventory in Asia (gridded and national data).
  • PKU-CO2: Peking University CO2 emission inventory (gridded data, used for provincial-level analysis).
  • Model Data:
    • CHIMERE: Mesoscale chemistry-transport model used for simulating NO2 columns and background levels.
  • WRF-ARW: Weather Research and Forecasting model for meteorological data.
  • NCEP Reanalysis-2: Meteorological reanalysis data.

Key Insights and Ideas

  • NO2 as a CO2 Proxy: The study demonstrates that satellite NO2 measurements can provide valuable, independent information on CO2 emission trends, especially for regions with strong co-location of CO2 and NOx sources, like China's industrial areas.
  • Challenging Inventory Linearity: The most striking insight is the challenge to the sharp, non-linear "bend" in CO2 emission trends around 2000-2002 often seen in bottom-up inventories. Satellite data suggests a more gradual acceleration.
  • Underestimation of Early Trends: The top-down estimates indicate a significant positive trend in China's CO2 emissions during the 1996-2001 period, which is not supported by global emission inventories (EDGAR, GCP) that suggest near-zero or slightly negative trends. This points to potential underestimation or time-lag issues in early inventory data.
  • Regional Differences Matter: The analysis of provincial-level data suggests that regional emission inventories (like REAS and PKU-CO2) might better capture the spatial distribution of emission changes compared to global inventories (EDGAR), which rely on more generalized proxy data.

Key Results

  • Emission Trends:
    • 1996-2001: Top-down estimates show a strong positive trend (3.7-8.3% per year), while EDGAR/GCP show near-zero trends. REAS shows a smaller positive trend.
  • 2001-2008: All estimates, both top-down and bottom-up, show large positive trends, with top-down estimates around 11.0-13.2% per year.
  • Non-linearity: The acceleration of CO2 emission growth is confirmed, but the "sharp bend" around 2000-2002 seen in EDGAR/GCP is not supported by the satellite-derived NO2 data.
  • Spatial Analysis: Top-down estimates showed better agreement with the PKU-CO2 inventory than with EDGAR at the provincial level, suggesting that regional inventories might be more accurate in capturing spatial emission dynamics.

Limitations and Future Work

  • Systematic Uncertainties: The authors acknowledge that systematic uncertainties in satellite retrieval, model parameterizations (especially for background NO2), and the NOx-to-CO2 conversion factors exist and are difficult to fully quantify.
  • Conversion Factor Sensitivity: While the choice of conversion factor (EDGAR vs. REAS) influenced the absolute trend values, it did not resolve the fundamental discrepancies between top-down and bottom-up estimates for the 1996-2001 period.
  • Data Quality: The study reinforces findings that activity data for China in international energy statistics may have significant uncertainties and time lags, particularly concerning coal production and consumption.
  • Future Work: The authors emphasize the need for further research to verify emission inventories using various satellite and ground-based data, and to improve GHG emission estimates, especially given China's role as a major global emitter. They suggest that satellite NO2 measurements are a valuable tool for this purpose.

Summary updated

Abstract

Multiannual satellite measurements of tropospheric NO2 columns are used for evaluation of CO2 emission changes in China in the period from 1996 to 2008. Indirect top-down annual estimates of CO2 emissions are derived from the satellite NO2 column measurements by means of a simple inverse modeling procedure involving simulations performed with the CHIMERE mesoscale chemistry–transport model and the CO2-to-NOx emission ratios from the Emission Database for Global Atmospheric Research (EDGAR) global anthropogenic emission inventory and Regional Emission Inventory in Asia (REAS). Exponential trends in the normalized time series of annual emissions are evaluated separately for the periods from 1996 to 2001 and from 2001 to 2008. The results indicate that the both periods manifest strong positive trends in the CO2 emissions, and that the trend in the second period was significantly larger than the trend in the first period. Specifically, the trends in the first and second periods are best estimated to be in the range from 3.7 to 8.3 and from 11.0 to 13.2% per year, respectively, taking into account statistical uncertainties and differences between the CO2-to-NOx emission ratios from the EDGAR and REAS inventories. Comparison of our indirect top-down estimates of the CO2 emission changes with the corresponding bottom-up estimates provided by the EDGAR (version 4.2) and Global Carbon Project (GCP) glomal emission inventories reveals that while acceleration of the CO2 emission growth in the considered period is a common feature of both kinds of estimates, nonlinearity in the CO2 emission changes may be strongly exaggerated in the global emission inventories. Specifically, the atmospheric NO2 observations do not confirm the existence of a sharp bend in the emission inventory data time series in the period from 2000 to 2002. A significant quantitative difference is revealed between the bottom-up and indirect top-down estimates of the CO2 emission trend in the period from 1996 to 2001 (specifically, the trend was not positive according to the global emission inventories, but is strongly positive in our estimates). These results confirm the findings of earlier studies that indicated probable large uncertainties in the energy production and other activity data for China from international energy statistics used as the input information in the global emission inventories. For the period from 2001 to 2008, some quantitative differences between the different kinds of estimates are found to be in the range of possible systematic uncertainties associated with our estimation method. In general, satellite measurements of tropospheric NO2 are shown to be a useful source of information on CO2 sources collocated with sources of nitrogen oxides; the corresponding potential of these measurements should be exploited further in future studies.
BibTeX
@inproceedings{berezin2013multiannual,
  author = {E. V. Berezin and I. B. Konovalov and Philippe Ciais and Andreas Richter and Shu Tao and Greet Janssens‐Maenhout and Matthias Beekmann and Ernst‐Detlef Schulze},
  title = {{Multiannual changes of CO <sub>2</sub> emissions in China: indirect estimates derived from satellite measurements of tropospheric NO <sub>2</sub> columns}},
  booktitle = {Atmospheric},
  year = {2013},
  doi = {10.5194/acp-13-9415-2013},
}
Figure