How many tons of coal did China consume last year? It's a simple question with no simple answer, despite its critical importance to climate change.
So far, China's National Bureau of Statistics (NBS) has said only that coal consumption rose 0.4 percent in 2017 from a year earlier, marking the first annual increase since 2013.
But since the NBS has not released tonnage figures for 2016, the calculation for last year remains a black box.
The unknown number of physical tons is important because China produces and consumes about half the world's coal, making it the single largest source of man-made carbon emissions.
Climate scientists generally agree that carbon dioxide (CO2) emissions rose last year after a three-year period of little or no growth. But estimates vary, and experts warn that preliminary figures may be revised.
Trends in China have been given much of the credit, and alternately, the blame, as the government pushes cleaner energy sources like renewables and natural gas while meeting demands for economic growth and power supplies.
Last year, official economic and energy data charted the good news-bad news dichotomy as gross domestic product overshot the government's target with a 6.8-percent growth rate.
Electricity use jumped 6.6 percent, continuing to rebound from a slump in 2015. Coal consumption rose in volume, but it fell as a percentage of total energy by 1.6 points to 60.4 percent, the National Development and Reform Commission (NDRC) and the National Energy Administration (NEA) said.
Although coal's share of the energy mix declined, the bullishness of consumption throughout the year drew the attention of environmentalists.
In November, a group of prominent climate scientists from the Global Carbon Project projected that energy-related CO2 would grow by two percent in 2017, thanks largely to a 3.5-percent increase in China's emissions, driven in turn by a three-percent rise in coal use.
The forecast announced during the United Nations Climate Change Conference in Bonn, Germany drew fire from experts at the Brookings-Tsinghua Center for Public Policy, writing in the official English-language China Daily.
The consumption estimate was "likely to be an overestimate," based on half-year data from the China Coal Industry Association (CCIA) and the NEA, said center director Ye Qi and research associate Jiaqi Lu in January.
The Brookings experts charged the Global Carbon Project scientists with "inappropriate use of the data sources," concluding that coal consumption had probably increased by "around 1 percent" to 3.82 billion metric tons instead of the 3.90 billion tons implied by the scientists' group.
The emissions rise was also likely to be "closer to 1 percent," considerably less than the scientists' 3.5-percent projection, the Brookings experts said, citing other NEA and NDRC data. In any case, coal consumption fell far short of the record 4.24 billion tons set in 2013, they said.
Comparisons are complicated
Calculations based on the announced NBS growth rates and the last annual tonnage data from 2015 suggest that 2017 usage would have been lower than both estimates at 3.79 billion metric tons.
Comparisons are complicated by differing measures of consumption that rely on obscure conversion factors.
The IEA, for example, reports data in millions of tons of oil equivalent (Mtoe), allowing calculations across various forms of energy. The NBS uses millions of tons of coal equivalent (Mtce), a standard that reflects the relatively low energy content of Chinese coal.
The China Energy Group at the Lawrence Berkeley National Laboratory in California has listed conversion factors for the various forms of energy in a 2016 statistical report.
Philip Andrews-Speed, a China energy expert at National University of Singapore, noted that last year's physical coal consumption can be calculated in at least two ways from the NBS data published so far.
The first is to multiply the coal equivalent data for 2016 by the reported 0.4-percent growth rate. The second is to take the coal share of total energy for 2017. Both can be converted using the factor of 0.714 to produce 3.79-3.80 billion physical tons of coal.
This is about the same result derived from applying the NBS reported growth rates for 2016 and 2017 to the last official tonnage figure for 2015. Consumption fell 4.7 percent in 2016, according to the NBS.
The variations seem minor, until one considers that the difference between the high and low estimates is about 110 million tons, because China's consumption is so huge.
The statistical range is roughly equal to the annual consumption of major coal users like Indonesia and Australia, or about half that of Germany. Other estimates and past revisions suggest the range could be greater.
In March, the Paris-based International Energy Agency (IEA) said that China's coal demand rose 0.3 percent last year, slightly lower than the NBS consumption rate and considerably less than both the Brookings and the Global Carbon Project estimates.
The IEA found that global energy-related CO2 emissions increased 1.4 percent, reaching a record high of 32.5 billion tons in 2017. The figure did not appear to be directly comparable with totals from the Global Carbon Project, which had already reported higher CO2 of over 36 billion tons for 2016.
Jan Ivar Korsbakken, a climate scientist and member of the Global Carbon Project, said the variations in last year's coal estimates are slight compared with some of China's major statistical revisions in the past.
"The uncertainty in China's coal data certainly creates some challenges for assessing global emissions with pinpoint accuracy," said Korsbakken, a senior researcher at Norway's Center for International Climate Research (CICERO). "But it's a minor nuisance in the grand scale of things."
The differences in the 2017 estimates are equal to "less than 1 percent of global emissions." Korsbakken said by email. "It doesn't flip us from one type of climate scenario to another."
"The real challenge looking ahead is that nobody, perhaps not even the Chinese authorities, knows for sure exactly what China will do policy-wise with economic stimulus, transition from heavy industry to services and domestic consumption, or environmental regulations, or how fast they will do it," he said.
Why the mystery?
But the question remains why China makes such a mystery of its coal data, and why it just doesn't report estimates in physical tons.
David Fridley, staff scientist at the Berkeley Lab's China Energy Group, said that the NBS energy data reported in February are only preliminary numbers that go through a series of revisions.
The final physical tonnage numbers are reported annually in the China Energy Statistical Yearbook with a lag time of 18 to 20 months after the end of each year. Numbers are also published in the annual statistical section of the NBS website, but these may be at odds with those in the yearbook.
Unannounced annual revisions may also upset calculations based on prior years.
Fridley noted that the big 4.7 percent drop in consumption reported for 2016 may be only about 1.3 percent due to revisions.
While significant for emissions calculations, the unannounced adjustments pale in comparison to major corrections, like the double-digit retroactive revision that NBS made in 2015 to compensate for undercounting.
That adjustment, made after a five-year economic census, caused scientists to raise emissions estimates by over 1 billion tons of carbon per year.
"In some of the years in question, it was like suddenly having an extra Germany added to global emission numbers," Korsbakken said.
The five-year revisions may do little to resolve the year- to-year uncertainties over consumption estimates.
The heat content of China's coal remains a major wild card because the energy estimate can change annually. Calculations based on current Mtce data and past heat content assumptions can produce variations of well over 100 million tons.
"Of course, this makes problems for people trying to estimate CO2 emissions, but no country in the world can provide an accurate reporting on their energy consumption and CO2 emissions just months after the year ended," Fridley said.
Long delays may be especially prevalent in the case of resource industries as sprawling as China's. The government counted as many as 500,000 small coal mines when reforms started in the 1990s and now officially recognizes fewer than 5,000 operations, either producing or under construction.
"I wouldn't ascribe anything nefarious to this. The data system is actually better than it used to be," Fridley said.
Playing the game
Still, it may be tempting to see other reasons for not reporting physical tonnage estimates, even if they are only preliminary. The "Mtce" coal equivalent numbers look a lot lower than the physical tonnage data, drawing less attention to China's role in global emissions.
Consumption of 2.7 billion tons of coal equivalent certainly sounds less than 3.8 billion tons of physical coal.
While the statisticians may not be playing this game, there are signs that China's official press avoids reporting the tonnage figures, even when government agencies make them available.
On April 24, NEA officials held a press conference in Beijing to report energy results for the first quarter. Although the coal data remained incomplete, the figures for January and February showed a marked increase in consumption, far higher than the 0.4-percent growth rate for last year.
"In the first two months, the national coal consumption was about 600 million tons, an increase of 4.2 percent year- on-year," said Li Fulong, director of the NEA Department of Planning and Development.
The official Xinhua news agency report from the conference on the same day included consumption data for electricity, oil and gas, but no mention of coal, China's largest energy source.
The coverage seems to be a sign of sensitivity on the subject of how much coal China is burning, but whether the reticence is reflected in NBS reporting of physical tonnage remains to be seen.
"The benign explanation is that there are so many different types of coal in China that it is better for statistical and policy purposes to give a single number in terms for equivalence," Andrews-Speed said.