Historical fossil fuel CO2 emissions can be reconstructed back to 1751 based on energy statistics. These reconstructions detail the production quantities of various forms of fossil fuels (coal, brown coal, peat and crude oil), which when combined with trade data on imports and exports, allow for national-level reconstructions of fossil fuel production and resultant CO2 emissions. More recent energy statistics are sourced from the UN Statistical Office, which compiles data from official national statistical publications and annual questionnaires. Data on cement production and gas flaring can also be sourced from UN data, supplemented by data from the US Department of Interior Geological Survey (USGS) and US Department of Energy Information Administration. A full description of data acquisition and original sources can be found at the Carbon Dioxide Information Analysis Center (CDIAC).

As an example: how do we estimate Canada’s CO2 emissions in 1900? Let’s look at the steps involved in this estimation.

• Step 1: we gather industrial data on how much coal, brown coal, peat and crude oil Canada extracted in 1900. This tells us how much energy it could produce if it used all of this domestically.

• Step 2: we cannot assume that Canada only used fuels produced domestically—it might have imported some fuel, or exported it elsewhere. To find out how much Canada actually burned domestically, we therefore have to correct for this trade. If we take its domestic production (account for any fuel it stores as stocks), add any fuel it imported, and subtract any fuel it exported, we have an estimate of its net consumption in 1900. In other words, if we calculate: Coal extraction − Coal exported + Coal imported − Coal stored as stocks, we can estimate the amount of coal Canada burned in 1900.

• Step 3: converting energy produced to CO2 emissions. we know, based on the quality of coal, its carbon content and how much CO2 would be emitted for every kilogram burned (i.e. its emission factor). Multiplying the quantity of coal burned by its emission factor, we can estimate Canada’s CO2 emissions from coal in 1900.

• Step 4: doing this calculation for all fuel types, we can calculate Canada’s total emissions in 1900.

Providing good estimates of  CO2 emissions requires reliable and extensive coverage on domestic and traded energy—the international framework and monitoring of this reporting has significantly improved through time. For this reason, our understanding of emissions in the late 20th and 21st centuries is more reliable than our long-term reconstructions. The Intergovernmental Panel on Climate Change (IPCC) provide clear guidelines on methodologies and best practice for measuring and monitoring CO2 estimates at the national level.16

There are two key ways uncertainties can be introduced: the reporting of energy consumption, and the assumption of emissions factors (i.e. the carbon content) used for fuel burning. Since energy consumption is strongly related to economic and trade figures (which are typically monitored closely), uncertainties are typically low for energy reporting. Uncertainty can be introduced in the assumptions nations make on the correct CO2 emission factor for certain fuel types.

Country size and the level of uncertainty in these calculations have a significant influence on the inaccuracy of our global emissions figures. In the most extreme example to date, Lui et al. (2015) revealed that China overestimated its annual emissions in 2013 by using global average emission factors, rather than specific figures for the carbon content of its domestic coal supply.17

As the world’s largest CO2 emitter, this inaccuracy had a significant impact on global emissions estimates, resulting in a 10% overestimation. More typically, uncertainty in global CO2 emissions ranges between 2-5%.