It seemed like a straightforward task: Washington Post reporters wanted to measure the gap between what countries acknowledge releasing into the air and the total emissions actually found in the atmosphere.

On the eve of the U.N. Climate Change Conference in Glasgow, those unaccounted-for gases seemed highly important: If world leaders were underreporting emissions or not taking full responsibility for their effects on the planet, how could any global agreement to reduce greenhouse gases aim at the right target?

But when The Post set out to determine the size of the gap, it found that little about emissions data is straightforward or easy.

To start, many of the country reports to the United Nations were out-of-date, in different measurement units and formats or had numbers buried in hundreds of pages of text. Some types of emissions — notably man-made fluorinated gases and land-use sector gases such as carbon dioxide emissions contained in forest fire smoke — varied wildly from year to year, and countries’ spotty reporting made it difficult to estimate 2019 figures.

Finally, even independent measurements of the total level of emissions vary greatly. Some include certain emission sources that others omit. Different measurement techniques find a wide range of results. Some measurements of certain gases, such as methane and nitrous oxide, are based on a “top-down” approach using atmospheric and satellite-based methods that most countries don’t use in their reporting. Other approaches use a “bottom-up” method, more similar to country techniques, that tallies gases based on power generation, industrial output, agricultural production, transportation and other emissions-producing activities. The Post relied on comprehensive emissions measurements from several different scientific groups.

The result found countries claimed roughly 44.2 billion metric tons of emissions from all sources in 2019, while independent measurements discovered anywhere from 52.7 to 57.4 billion tons of gases in the atmosphere — a gap of at least 16 and up to 23 percent.

To complete their calculations, reporters for this article broke their work into eight steps:

1. Hand-building a data set to collect 30 years of emissions figures reported to the United Nations in a variety of ways

  • The Post pulled each year of data from 45 countries directly from the latest 2019 UNFCCC reports. For 148 other nations, it pulled in all years available and supplemented that with hand-entered data gathered by a team of reporters drawing on other documents submitted to the United Nations. A handful of countries either had never reported or had errors in their data that made it unusable.

2. Standardizing the measurements of emissions into comparable figures

  • Depending on the year and country reporting, nations used differently weighted formulas to convert other gases to standardized carbon dioxide equivalents, based on the most current scientific understanding of the global warming potential of each gas. The Post converted these emissions back to original units so they could be compared over time and between countries.

3. Creating a model to estimate what emissions each country would have reported in 2019, if they only reported in an earlier year

  • To overcome missing data, The Post used a linear regression technique to model what countries would have reported in 2019, measuring past years of reports against independent estimates from Minx et al., a research effort that has totaled each country’s greenhouse gases.

4. Using other analysis techniques to gauge the accuracy of our model and conclusions

  • We tested the regression results by doing a percent change adjustment on the most recent year of data reported and found that results tracked the model approach.

5. Creating a similar model to adjust total global reported fluorinated gas emissions to 2019

  • As in Step 4, The Post did a percent change adjustment on the most recent year of fluorinated gases, or F-gases, reported to arrive at a total for fluorinated gases released in 2019.

6. Making accommodations for land-use sector gases and using a simulation to check our work

  • To handle land-use sector gases, which present significantly more scientific uncertainty in their measurements, The Post used an average (mean) of each country’s reported land-use gases and added those together for a global land-use total. We then ran a Monte Carlo simulation of 10,000 possible global land-use sector readings to test the range of possible values for reporting from countries that didn’t provide 2019 figures.

7. Finding independent estimates of emissions in the atmosphere to compare the U.N. reports against

  • The Post identified several other independent data sets of global greenhouse gases to compare its total against. These included data from the Global Carbon Project, Minx et al. and the U.N. Food and Agriculture Organization.

8. Calculating the gap between the independent estimates and figures reported to the United Nations

  • We took the difference between the independent reports and our calculated 2019 total to estimate the gap in emissions reporting.
Billions of metric tons of carbon dioxide-equivalent emissions (AR5), subtracting international shipping and aviation
gas FAO WMO GCP Minx Post
CO₂ 37.842 NA 39.270 43.127 31.837
CH₄ 10.267 NA 10.052 10.603 8.469
N₂O 2.878 NA 2.998 2.546 2.878
F-gases 1.666 1.17 NA 1.167 0.997
Total 52.654 NA 53.490 57.443 44.181
Note: Independent global emissions measurements come from the Global Carbon Project, Minx et al., the U.N.'s Food and Agriculture Organization and the World Meteorological Organization. Global Carbon Project does not have an estimate for f-gases, so the WMO’s number was added to their total.

The Post’s methods were guided by conversations with several climate and emissions experts, and then reviewed by still others. Experts The Post spoke with about the findings and methods used to derive them included:

  • Anita Ganesan, associate professor, University of Bristol
  • Rob Jackson, professor, Stanford University
  • Julia Pongratz and Clemens Schwingshackl, University of Munich
  • Leehi Yona, Stanford University
  • Steven Smith, Pacific Northwest National Laboratory
  • Philippe Ciais, Laboratory of Climate and Environmental Sciences, France
  • Glen Peters and Robbie Andrew, Center for International Climate Research, Oslo
  • Francesco Tubiello, U.N. Food and Agriculture Organization
  • Giacomo Grassi, European Commission, Joint Research Centre

For a more detailed methodology of how The Post handled the data and conducted its analysis, see here.

Taylor Telford and Juliet Eilperin contributed to this report. Nick Trombola and Caroline Cliona Boyle with the American University-Washington Post practicum program also contributed reporting.