More than 60 years ago, atmospheric scientist Charles David Keeling began regular measurements of carbon dioxide concentrations in the atmosphere. In the heart of the Pacific and far from the largest human sources of the gas, Hawaii’s Mauna Loa Observatory was an ideal location for these measurements. Within just two years, Keeling had detected two patterns in the data. The first was an annual rise and fall as the seasons came and went. But the second — a year-by-year increase — suggested something alarming: a rise in carbon dioxide produced by the widespread burning of fossil fuels. In 1965, Keeling’s measurements were incorporated into a report for US President Lyndon B. Johnson that described carbon dioxide from fossil fuels as “the invisible pollutant” and warned of its dangers.

Since then, global emissions of carbon dioxide and other greenhouse gases have continued to rise, as have the concerns over the changes that such an atmospheric shift brings. Both of these trends took center stage this week in Madrid, where the United Nations is holding its annual climate change summit, the 25th Conference of the Parties (COP25).

Observations are still taken at Mauna Loa today, and the resulting “Keeling Curve” reveals that atmospheric carbon dioxide levels have increased by almost a third since the first measurements were taken. The world's average temperature has already warmed by around 1 degree Celsius (1.8 degrees Fahrenheit) since preindustrial times, driving increases in everything from sea levels to the frequency of extreme weather events.

Graph shows the patterns of carbon dioxide measured over time from Mauna Loa Observatory, showing both a seasonal rise and fall and an overall rise year over year.

Continuous measurements of carbon dioxide levels at the Mauna Loa Observatory have revealed that humans are contributing to a relentless increase. But these measurements only reveal the portion of emissions that remain in the atmosphere, and not the total emitted by human activities.

CREDIT: SCRIPPS INSTITUTION OF OCEANOGRAPHY

For those groups and nations striving to limit global warming, accurately tracking carbon emissions will be key to assessing progress and validating international agreements. But how do scientists do that? And how does the amount released into the air relate to what scientists end up measuring at outposts such as Mauna Loa?

Here’s the current state of counting carbon, explained.

Why is monitoring global carbon emissions important?

A comprehensive tally of carbon released is essential not just for assessing which countries are pulling their weight and meeting agreed targets. It’s also key to improving understanding of carbon’s natural cycle and to more precisely quantifying the link between humankind’s emissions and the planet’s temperature. But calculating, much less measuring, global carbon dioxide emissions remains an immense technical challenge, since almost every human activity is implicated in the molecule’s release.

In Paris in 2015, most of the world reached an agreement on climate change. The deal was to limit the world’s warming to well below 2 degrees Celsius (3.6 degrees Fahrenheit), with a target of just 1.5 degrees Celsius (2.7 degrees Fahrenheit) above preindustrial levels. Nations pledged to cut their emissions, and the Paris Agreement aims to periodically review progress. While these pledges are insufficient to achieve the deal’s targets, the hope is that countries will gradually ramp up their ambitions, and further ramp down emissions.

The emissions pledges are exactly that — pledges. They are not legally binding, and if a country misses its intended targets, the only diplomatic consequence would be the judgment of the international community.

But all of this relies on a clear picture of the country’s emissions in the first place. It’s a crucial undertaking, because “monitoring emissions is directly at the heart of the pledge-and-review concept,” says Gabriel Chan of the Humphrey School of Public Affairs at the University of Minnesota, who reviewed the state of international climate policy in the Annual Review of Resource Economics.

Atmospheric observations — like those carried out at Mauna Loa — provide a global, cumulative picture, but cannot be decomposed into year-by-year national contributions. Global measures also fail to account for the natural carbon “sinks” — the portion of carbon dioxide emissions that are taken up by the oceans and land. For a clear picture of national emissions, researchers have to start from the bottom up.

How do you calculate a country’s carbon emissions?

In theory this is just a matter of math, but in practice it’s a question of huge-scale accounting. To get a picture of the carbon dioxide a country emits by burning fossil fuels, all energy use must first be counted. These assessments are already carried out for economic reasons and include tabulating the quantities of different fuels — such as coal, gas or kerosene — that are produced, traded, converted or used by a country across all sectors. While the contribution of large sources such as power plants can be relatively straightforward to assess, other ledger entries — such as household activities — “are very hard to account for,” says Chan. Accurately estimating these sources requires surveys to assess what goes on within a typical home and extrapolating from those.

A map has been distorted to reflect each nation’s per capita carbon dioxide emission.

Countries are resized to reflect their 2016 carbon dioxide emissions, with shading indicating the level of their per capita emissions.

CREDIT: WORLDMAPPER.ORG / DATA BY EMISSION DATABASE FOR GLOBAL ATMOSPHERIC RESEARCH (EDGAR)

Figures from these energy assessments can be used to estimate national carbon dioxide emissions. Inventories provided by the Intergovernmental Panel on Climate Change list the amount of carbon dioxide released when an amount of a particular fuel is burned. These “emissions factors” can be combined with energy data to calculate the amount of carbon dioxide that will be released from all of a nation’s fossil fuel combustion.

The International Energy Agency, which has been collecting energy data for over 40 years and calculates its own statistics on emissions, recognizes the difficulty in getting it right. “We do really spend a substantial part of our time validating the data,” says Roberta Quadrelli of the IEA’s Energy Data Centre. For example, if a refinery disappears from the data, it’s essential to find out whether its absence was caused by the refinery closing or by being missed in the reporting.

Issues can also crop up when converting energy use to emissions. A 2015 study found that in one year China’s emissions had been overestimated by some 14 percent. “The error bar was like an entire Germany,” says Chan. This huge miscalculation was primarily caused by a misassessment of the quality of the coal burned in Chinese power plants. Given the scale of China’s emissions (currently higher than those of any other nation, although not on a per person basis), some errors are not a surprise, says climate scientist Corinne Le Quéré, who leads the annual Global Carbon Budget report. “I don’t want to give them excuses, but it’s a big challenge.”

What about tracking smaller, less obvious sources of carbon?

In fact, one of the biggest challenges in tracking carbon dioxide emissions isn’t related to burning fossil fuels at all. Certain changes in land use — such as deforestation or urbanization — can lead to an uptick in carbon dioxide entering the atmosphere through a number of complex processes. An area of much current research, these factors are far harder to assess than emissions from transportation or power plants. And while land use changes were estimated to be responsible for only around 12 percent of global emissions in 2016, they remain a major source of uncertainty about how much carbon is entering the atmosphere.

Photo shows workers in Cameroon involved in logging a forest.

Land use changes — such as deforestation — can lead to a range of processes that release greenhouse gases. Because of the complexity of these processes, the emissions from changing land use remain difficult to quantify.

CREDIT: PHOTO BY OLLIVIER GIRARD / CIFOR

For all these reasons and more, overall uncertainty in total carbon emissions remains high, equivalent to nearly 10 percent of the calculated annual emissions and more than the European Union’s entire fossil fuel emission tally for 2017.

How current are carbon emissions numbers?

Timing poses another challenge. The complexity of tabulating national emission totals also causes delays in reporting. These delays can make a big difference for policy. Official statistics may take many months to appear, meaning negotiators are often working with outdated information, says Niklas Höhne, who founded Climate Action Tracker, which monitors nations’ climate commitments and actions.

At the extreme end, during the Copenhagen climate negotiations in 2009, negotiators were working with an IPCC report published in 2007. The report included emissions only up to 2004, and this chain of delays meant that there was a half-decade gap between policy and reality. These five years — it was later shown — had seen a significant increase in emissions, and the scenarios in terms of emissions and temperature targets sketched out in the negotiations were misaligned with the real world. Even as they were unveiled, they were out of date.

Is there any way to directly monitor carbon emissions?

Given the difficulties, climate researchers are exploring other ways to count carbon that could allow for more up-to-date data. Rather than estimating the amount of fuel burned, and land altered, researchers are keen to start directly measuring emissions themselves. This could be achieved remotely, using satellites that measure how carbon dioxide absorbs sunlight as it passes through the atmosphere. “That, I think, is a significant advantage, because then you can better adjust climate policies to what’s really happening,” says Höhne.

Photo shows workers loading bags of cement mix in the Philippines.

Cement production contributes some 8 percent of global carbon dioxide emissions, but remains relatively weakly regulated.

CREDIT: ADAM COHN / FLICKR

Satellites using remote sensing to monitor carbon dioxide levels are already in operation but provide too sparse a picture to regularly track emissions across the globe. The European Space Agency is planning a new fleet for launch starting in 2025 that it expects will watch emissions unfold in unprecedented detail — resolving plumes of carbon dioxide just 2 kilometers across, while aiming to measure each location on Earth every three days. Together with ground-based observations and information from other agencies, these space missions will provide a far more current picture of emissions patterns.

Watching emissions from space will also allow the deeply uncertain impacts of land use changes to be directly observed. The interplay between land and climate change is so complex that a special IPCC report was recently published on the topic. “We really hope we can make a significant contribution there,” says Richard Engelen, deputy head of the Copernicus Atmosphere Monitoring Service at the European Centre for Medium-Range Weather Forecasts. “The challenge is there for us to tackle.”

How will better emissions data help improve climate models?

Gathering a more fine-grained picture of human-made emissions also has substantial consequences for understanding the mechanisms of climate change itself. A core question of climate science is how much the world warms for a given amount of carbon dioxide emitted. (And that’s notably ignoring other potent greenhouse gases such as methane, ozone and nitrous oxide, among others.) Because carbon dioxide can stick around in the atmosphere for hundreds — even thousands — of years, this means counting emissions not just from the last few years but for more than a century, as industrialization has grown.

While measurements like those at Mauna Loa can reveal how much carbon dioxide has ended up in the atmosphere, it doesn’t tell you what has been put in. Less than half of emissions actually remain in the air. The rest are absorbed by land and oceans, where the total stored carbon cannot be directly measured. Instead, humanity’s cumulative carbon dioxide contribution must be estimated in the same way as contemporary emissions: by accounting for energy use and changes in land use, and converting these figures into emissions stats.

Getting such a long view helps further refine predictions of how much more carbon dioxide can be released before breaching the temperature limits agreed on in the Paris Agreement — limits that are fast approaching.

What role does monitoring play in meeting climate targets?

Monitoring emissions can identify positive signs, too. Le Quéré and colleagues tracked carbon dioxide emissions of the 18 countries with the most dramatic reductions, searching for patterns driving the drops. They saw renewable energy replacing fossil fuels, and that the countries with the biggest decreases had the most ambitious policies for reducing emissions. Seeing that such changes really can drive down emissions “was very encouraging,” says Le Quéré.

These countries remain exceptions, rather than the rule. To limit global warming to the Paris Agreement’s more ambitious target of 1.5 degrees C, global carbon dioxide emissions must reach net-zero — meaning any carbon released would need to be balanced by active removal of carbon dioxide from the atmosphere — by around the middle of the century. And yet in 2019, estimates indicate that emissions didn’t drop — they reached a new high.

Counting carbon may be imperfect, but one thing remains obvious: Global emissions are rising even as international agreements call for them to be falling.