Industrial production of carbon dioxide from fossil
fuels and limestone
By CHARLES D. KEELING, Scripps Institution of Oceanography, University of California,
P.O. Box 109, Sun Diego 92037, U S A
(Manuscript received March 11; revised version November 9, 1972)
ABSTRACT
The release of carbon dioxide into the atmosphere by the burning of fossil fuels is
significantly altering the carbon cycle by adding to the amount of carbon in the
atmosphere and in the more rapidly interacting portions of the biosphere and oceans.
In order better t o assess these changes, the basis for calculating global CO, emissions
is reviewed and new annual values are computed for the period 1800 through 1969.
The world average fractions of carbon in coal and lignite, estimated from calorific
data, are found t o be lower than previously assumed. When account is taken of handling
losses and partial diversion t o produce petrochemicals, road asphalt, and other nonfuels, the calculated CO, emissions are further reduced by several percent even after
allowing that most unburned materials eventually oxidize t o CO, in the environment.
On the other hand, the production of C O , by kilning of limestone adds 1 t o 2 % t o
the annual totals. The cumulative increase in carbon in the short term carbon cycle,
owing to man’s industrial and domestic activities up to 1970, is estimated to be 1.12+
0.14 A 10’’ g (4.1 k0.5 x 10’7 g CO,), or about 18 % of the amount of CO, in the atmosphere during the late nineteenth century.
1. Introduction
The combustion of fossilfuels by worldindustry
is causing an increase in carbon dioxide in the
atmosphere and in reservoirs which exchange
this gas rapidly with the air, including all
surface and some deep waters of the oceans
and probably land plants and soil as well.
Of special interest is to know what fraction of
the industrial input is remaining air-borne
(Callendar, 1938, 1957, 1958; Fonselius et al.,
1956; Bolin & Eriksson, 1959; Bray, 1959;
Keeling, 1960; Bolin & Bischof, 1970), and an
essential ingrcdient in determining this fraction
is to know accurately the amount of CO,
produced from fossil fuel. This in turn depends
on estimates of the amounts of fossil fuel
consumed and of the composition of the byproducts.
The quality of source data on fossil fuel has
improved in recent years because the Statistical
Office of the United Nations now compiles
detailed records of world-wide fossil fuel production and consumption. Revelle & Suess
(1957), Revelle (1965), Baxter & Walton (1970),
and Broecker e t al. (1971) already have used
these data to calculate CO, emissions but
without sufficient documentation to allow their
calculations to be fully verified or consistently
updated. Aneed for accurate predictions and the
recent attainment of more than ten years of
precise atmospheric observations (Bolin &
Bischof, 1970; Keeling e t al., 1973a and b )
now justify a new look a t fossil fuel usage.
The present paper is intended as a step
towards obtaining better estimates of industrial
CO, production. First, the total annual release
of fossil carbon into the global environment is
computed from data on production of coal,
lignite, crude oil, and natural gas. Then the
proportions of hydrocarbons and carbon-bearing
oxides are estimated as they exist directly after
product utilization or as losses through handling
and processing. Next, for those substances
which are initially released to the environment
in an under-oxidized state, the probable time is
estimated between release to the environment
and final oxidation to carbon dioxide. Finally
the total annual production of carbon dioxide
is calculated after account is taken of the burning of limestone to produce cement and of the
slow oxidation of such products as leaked natural gas, petrochemicals and road asphalt.
Tellus XXV (1973), 2
INDUSTRIAL PRODUCTION OF CARBON DIOXIDE FROM FOSSIL FUELS AND LIMESTONE
From the first to the last stage in computing
CO, emissions, the source information is increasingly uncertain. Also, I cannot be sure
that I have found all, or even the best, data on
which to base the calculations. From an analysis
of probable errors entering each stage of the
computations, I estimate below that the new
C O , production figures for the last decade are
likely to be within 13 % of the true values. As a
result of discounting uncombusted hydrocarbons
and of relating the carbon content of fuels to
the average calorific content, these new figures
are about 14% lower than those of Revelle
(1965).
2. Fossil fuel production
2.1. United Nations reference data
Basic source data since 1949 are found in an
annual series of documents on world energy
supplies published by the United Nations (19511969). Fuel production and consumption are
given by fuel type for each country. The data,
in general, are taken from official sources (listed
in Reports 1 and 2), but some estimates were
made by the U.N. Statistical Office owing to
missing or inconsistent information. The tables
are not entirely free of typographical errors.
After Report 1, only “commercial” fuels are
tabulated. Each annual report covers four
past years for individual fuels and ten years
in a special historical series. Inspection of the
overlapping data indicates that the figures are
repeatedly updated, but by only relatively
small amounts after the second revision.
2.2. Solid fuels
For coal, soft coal, and lignite the statisticians
have thrice revised their methods of reporting.
I n Report 1 (1929, 1939, 1949-50) and in
Report 2 (1951-54) all solid fuels are combined
under the heading of coal. I n Reports 3 (195558) and 4 (1956-59) coal and lignite are reported
separately, and after Report 4 soft coal is
removed from the coal category and included
with the lignite. Beginning with Report 8
(1960-63) lignite (including soft coal) is reported
in tons of coal equivalent. No precise definitions
of the various classes of coals are given in any
report.
To establish from these tabulations a consistent series of annual data, it is necessary to
Tellus XXV (1973), 2
175
know the coal energy equivalent of lignite. This
is reported to vary from 0.30 to 0.51 tons per
ton. The equivalency factor is listed for some
countries in the introduction of all reports and
for all countries in Report 2. By considering
data for the individual countries, and with the
aid of overlapping data for 1960-62 to establish
a factor for the USSR (for which the inclusion
of soft coal leads to a value different from that
cited in the introductions) the weighted world
average equivalency factor for 1958-67 is found
to lie between 0.437 and 0.443 with a time
average of 0.440.
Revelle (1965) quotes separate figures for
metric tons of lignite plus soft coal for all
years between 1950 and 1962, inclusive. SCEP
(1970) repeats his numbers and updates them
for 1963-67 by dividing the reported U.N.
values for these years by 0.440. Revelle does
not explain how he obtained lignite tonnages
for the years 1950-56 when the U.N. reports
gave no direct values, but for 1950-54 it seems
likely that he attributed exactly 1/6 of the
tonnage of coal to lignite (including soft coal),
and then used an equivalency factor of 0.50.
For 1955 and 1956 he appears to have increased
the published figures for lignite, excluding soft
coal, by a factor (1/0.85).
A comparison of the U.N. tables for the
individual solid fuels (their Tables 5 and 6)
with the historical series (their Table l ) , where
all solid fuels are combined in tons of coal
equivalent, demonstrates that the statisticians
have already performed the required conversions to yield consistent, values for lignite and
soft coal within the precision of the original
data. For 1951-54, where lignite was not
separately reported, the updated values in
Reports 3 through 9 make small changes
(increases up to 2%) in the original values in
Report 2. From 1960 onward the totals for
lignite, in coal equivalents, from Table 6, when
added to the figures for coal in Table 5, agree
within 1 % with updated figures in Table 1.
Before 1960 no direct verification is possible;
but the data for these years are reported in
eight reports with minor adjustments which
suggest careful revisions from the original
sources. I conclude that no advantage is gained
by calculating the carbon produced separately
from each solid fuel, since, as shown below,
the carbon fraction in the combined fuels can
be established as accurately as for the separate
176
C. D . KEELING
Table 1. World production of fossil fuels
UN Report Number
Coal and lignite
~
Year
1929
1937
1949
1950
1951
1952
1953
1954
1955
1956
1957
1958
1959
1960
1961
1962
1963
1964
1965
1966
1967
1968
1969
for
for
refined
all other
fuel oils
fuels
4
4
4
4
6
7
8
9
10
11
12
13
14
14
14
14
14
14
14
14
14
14
14
1C
1C
1C
1C
2
2
2
2
3
4
5
6
7
8
9
10
11
12
13
14
14
14
14
lo6 tons carbon
petroleuma
lo6 tons
Refined
fuel oils
lo6 tons
Non-fuel
petroleum
106 tons
Natural
gasb
100 ma
1412
1404
1476
1 605
1665
1661
1669
1666
1807
1887
1954
2 026
2 101
2 191
2 022
2 073
2 159
2 238
2 268
2 309
2 206
2 270
2 330
212
293
482
539
620
652
688
720
789
858
902
925
997
1074
1143
1238
1329
1436
1539
1671
1792
1956
2 108
173
234
399
447
530
564d
603
635
704
768
797
829
899
964
1030
1117
1211
1313
1414
1524
1631
1777
1911
39
59
83
92
90
88
85
85
85
100
105
96
98
110
113
121
118
123
125
147
161
179
197
57
78
169
196
228
247
261
274
299
324
354
382
427
467
504
551
602
658
704
766
824
896
981
Crude
%
69.2
69.1
69.0
69.0
69.0
68.9
68.9
68.9
68.8
68.8
68.8
68.8
68.8
68.8
68.6
68.6
68.6
68.6
68.6
68.6
68.5
68.6
68.6
Corrected t o “Trend A” (see text):
1958
1959
1960
1971
1988
2 005
U.N. data quoted in tons of coal equivalent were converted to tons of petroleum using the factor 1.3.
U.N. data quoted in tons of coal equivalent were converted to m3 of gas using the factor 1332 x 10-8.
Data for refined fuel oils were copied from United Nations Report 1, p. 29 (p. 14 for the year 1949).
A figure of 584 from United Nations Report 2, p. 29, used by SCEP [1970], is in error (cf. p. 79 of same
report).
a
fuels. Therefore, I will accept the combined
solid fuel values from Table 1of the U.N. reports
as the most reliable source data to calculate C 0 2
emissions.
2.3. Liquid and gaseous fuels
The tabulations for refined oil fuels, crude
petroleum, and natural gas present no reporting
problem such as just discussed for solid fuels.
The updated figures show little revision. This,
very likely, reflects the use of these fuels
principally by industrially advanced nations
that tend promptly to publish precise (or a t
least unalterable) records of production.
2.4. S u m m a r y of world production
World totals of production of all fossil fuels
for 1929,1937, and 1949-69 are given in Table 1.
All years are included for which any U.N.
report furnishes data. The values for total
solid fuels are copied directly from the ten
year series of the latest possible report. For
crude petroleum and natural gas, the conversion
factors quoted in the introduction of each
report are used to convert the U.N. values of
their Table 1 back into metric tons of petroleum
and cubic meters of natural gas. These reconverted values agree with direct figures reported
in earlier reports except for minor updating.
Tellus XXV (1973), 2
INDUSTRIAL PRODUCTION OF CARBON DIOXIDE FROM FOSSIL FUELS AND LIMESTONE
177
trend which implies considerable mining success
by the Great Leap Forward, is perhaps more
reasonable than the latter. The figures based
on trend A are listed a t the end of Table 1 .
On the assumption that the difference between
the reported trend and trend B is the worst
discrepancy to be found in the U.N. data, I
estimate that the relative uncertainty in solid
fuel tonnage is about & 8 %.
500
200
3. Carbon content of fossil fuels
100
3.1. Coal
I06 TONS
0
1950
1960
1970
Fig. 1 . Annual production, in metric tons, of solid
fuels for China, Mongolia, North Korea, and North
Vietnam. -, As reported; Trend A , t o produce a
linear increase in world totals from 1957 t o 1961;
Trend B , t o produce a linear increase for China
and her neighbors from 1957 t o 1962.
Refined oil fuels, not quoted in the ten year
series, were copied (after rounding to the nearest
ton) from the most recent compilation in Table
9 of the U.N. reports.
From the minor degree of updating indicated
in succeeding reports and from studies of source
material for the United States, I estimate that
the figures for liquid and gaseous fuels are
likely to be correct within & 3 %, to a confidence
of about 90% (see section 6.2). The tonnages
for solid fuels, on the other hand, are more
open to question. For example, the coal production of mainland China, shown in Fig. 1,
rises remarkably from 1957 to 1960 and then
declines so suddenly that the computed total
world production of fuel energy falls off in 1961
(Fig. 2).
It is questionable whether any already industrialized country could triple its coal production and consumption within three years,
even under the stimulus of a “Great Leap
Forward” as China’s economic exertion from
1956 to 1960 has been called. Two proposals
are shown in Figs. 1 and 2 for reducing the
United Nations figures for coal to agree with
more normal growth rates for China during the
period of unusually high reported production.
Trend A assumes a linear increase in world
production of coal, trend B a linear increase
just for China and her neighbors. The former
Tellus XXV (1973),2
I n the United Nations reports, fuels are
compared on the basis of calorific value: i.e.,
the heat liberated by complete combustion
with oxygen and the condensation and cooling
of the products to the temperature of determination. For average grade coal the reports
state that 0.125 tons have a calorific value
of 3.412 x lo6 B.Th.U. a t 60°F. This leads to a
unit calorific value of 6 8 9 0 cal g-’ (at 15”C),
or 12 400 B.Th.U. lb-l, almost identical with
the unit value quoted by Brame & King (1967,
2300
2200
2100
2200
1900
IS00
I700
1600
lo6TONS
1950
1960
1970
Fig. 2. Annual world production of solid fuels in
metric tons. Trend A and B have same meaning
as in Fig. 1.
178
C. D. KEELING
Table 2. Composition of air-dried solid fuelsaeb
%C
ash-free
dry basis
(1)
Name
Anthracite coal
Carbonaceous coal
Bituminous coal, Grade
Bituminous coal, Grade
Bituminous coal, Grade
Bituminous coal, Grade
Brown lignite
Peat
Wood
“Average grade coal”
Brown lignite in coal
equivalents
4
3
2
1
94
92
87
85.6
84
77
67
60
50
% rtyh
(2)
3
4
5
5
5
10
7
3
0.5
% hygroscopic
moisture
(3)
96 C
air-dried
basis
(4)
Calorific
value
B.Th.U.
(B.Th.U. lb-l) per % C
(5)
(6)
1
1
1
1
3
10
18‘
20
20
90.2
87.4
81.8
80.5
77.3
61.6
50.2
46.2
39.8
70.1
15 000
15 000
14 400
14 300
13 900
10 700
9 900
7 700
0 400
12 400d
166
172
176
178
180
174
197
167
161
177e
63.0
12 400d
197
a Data in columns 1, 2 , 3 and 5 copied from Brame & King (1967, p. 86). (English units in column 6 retained
t o avoid rounding errors. To convert to cal g-’ multiply by 0.556.)
in column 4 calculated by the formula: % C (air-dried basis) = % C (ash-free basis) x (100 % ash - % hygroscopic moisture)/100.
This is 10 % by weight of the original fuel if the latter contains 45 % free moisture.
Calorific value as reported by United Nations Report 14, p. 2.
Assumed to be the average of the 4 grades of Bituminous coal tabulated above.
’ Values
‘
p. 235). Average coal contains several percent
free moisture [cf. Brame & King. 1967, p. 761
so that when air-dried t o zero free moisture i t
would yield 12 600 t o 12 800 B.Th.U. Ib-l.
T h e calorific value and carbon content of bitu-
minous coals, as shown i n Table 2 and plotted
in Fig. 3, a r e essentially proportional, so that
an uncertain value for the free moisture will
n o t alter the reliability of the relationship.
T h u s a calorific value of 12 400 B.Th.U. lb-l
too
80
60
O/O
CARBON
40
20
0
BRITISH T H E R M A L UNITS PER POUND
Fig. 3. Calorific values of various solid fuels as a function of their carbon content. -, Average relation
for bituminous coals. The classification of coals is as defined in British commerce.
Tellus XXV (1973), 2
INDUSTRIAL PRODUCTION OF CARBON DIOXIDE FROM FOSSIL FUELS AND LIMESTONE
179
Table 3. Source data to compute % carbon in 801dfuels
~
Coal
lo6 ton
Lignite
106 ton
Lignitea
loe ton
coal
equiv.
(3)
(4)
(5)
d
1 325e
1 302e
1 342e
1 454e
1 524e
1499e
1 500e
232
252
332
361
397
423
450
69.5e
69.4e
69.3e
69.3e
69.3e
69.2e
69.2e
3
4
4
5
4
6
4
7
8
9
10
11
12
13
14
14
14
14
1 596e
1 68Se
1 73Se
1626
1 822e
1664
1 896e
1730
1808
1627
1676
1738
1798
1816
1848
1749
1810
1859
535
566
594
763
615
826
618
847
874
902
906
69.1e
69.1e
69.1e
68.8
69.1e
68.8
69.1e
68.8
68.8
68.6
68.6
68.6
68.6
68.6
68.6
68.5
68.6
68.6
U.N.
report
no.
(2)
Year
1929
1937
1949
1950
1951
1952
1953
1964
1955
1966
1957
d
d
d
d
d
d
1958
1959
1960
1961
1962
1963
1964
1965
1966
1967
1968
1969
~~
% Cb
%C
adjusted'
(6)
(7)
69.2
69.1
69.0
69.0
69.0
68.9
68.9
68.9
68.8
68.8
68.8
68.8
68.8
68.8
68.6
68.6
68.6
68.6
68.6
68.6
68.5
68.6
68.6
~
Figures in parentheses: ( ) calculated from previous column by multiplying by factor 0.44.
Assuming coal t o be 70 % carbon; lignite 63 % carbon.
Values from col. 6 reduced by 0.3 yo carbon for years in which soft coal is included with coal. (Adjustment
established by overlapping data for 1957-9.)
From United Nations (1955, p. 29).
Soft coal included with coal rather than lignite.
a
corresponds to a carbon content of 70 f4%
by weight, where half the uncertainty is
estimated from the maximum scatter of the
individual values of calorific value against
percent carbon content for various grades of
bituminous coal and the other half is to allow
for uncertainty in the estimate of average calorific value by the United Nations.
3.2. Lignite
If we accept the coal equivalent of average
lignite (including soft coal) as established above,
the latter fuel has a calorific value of 0.440 x
12 400 = 5 460 B.Th.U. lb-* (3 030 cal g-l).
According to Brame & King (1967, p. 112) black
lignites are not important commercially. Thus
the lignite consumed is predominately brown
Tellus XXV (1973),2
lignite with a typical calorific value of 9 900
B.Th.U. lb-l, when air-dried (Table 2). Average
lignites, with a much lower calorific value therefore appear to contain about 45 % free moisture.
Since 45-55 % total moisture is typical of
brown lignite, and only about 10% is hygroscopic and retained on drying (Brame & King,
1967, pp. 86 and 110), the moisture figure
deduced from the U.N. data seems not unreasonably high. Air-dried lignite contains about
50% carbon by weight (Table 2). Therefore, I
deduce that the United Nations' reported
average (moisture included) contains 28 yo
carbon. If the weight of lignite is reduced t o
coal equivalent tonnage, the carbon fraction
is 63%. This fraction does not depend on
knowing the conversion factor to coal equiva-
180
C. D. KEELING
lents, but only the relative calorific values of
coal and lignite, i.e.:
50%x
12 400 (B.Th.U. lb-' for coals)
9 900 (B.Th.T. lb-' for lignite)
=
63%
This value lies so close to that for average coal
that the percent carbon in combined coal and
lignite can be established for coal equivalent
tonnage with very little loss of accuracy owing
to uncertainty in relative tonnages. I n Table 3
the available separate tonnages are listed and
the percent carbon computed for all years
reported. The figures are always so close t o
6 9 % that this value (with an assigned uncertainty of & 4 % similar to coal) is applicable
to the entire period from 1929 through 1969.
tonnages for 1968 is 19 000 B.Th.U. lb-l.)
I do not know how to account for the small
remaining calorific loss of about 2%. The
statistician's use of the same factors, 1.3 and
1.5, for all recent years is in any case questionaable since the fractional tonnage used as fuel
rises from 8 7 % to 9 1 % between their base
time period of 1951-5 and the most recent
compilation for 1969.
Owing to tho uncertain significance of the
U.N. calorific values for liquid fuels, it seems
preferable t o establish carbon content for these
fuels on a different basis. Fortunately, the range
in composition (79-87%) is less than for solid
fuels and an average of 8 4 % cited by Brame
& King (1967, p. 238) is probably correct t o
+ 3 yo.
3.3. Crude and refined liquid fuels
For crude petroleum, the U.N. statisticians
report a calorific value 1.3 times that of average
coal, corresponding to 16 100 B.Th.U. lb-1
(8 960 cal g-l). This value is significantly lower
than typical measured values for crude petroleum (Brame & King, 1967, p. 235). Apparently
the statisticians, in determining a coal equivalency factor for crude petroleum, have subtracted losses and non-energy uses of crude
petroleum. This view seems reasonable because
the statisticians compute the total energy from
all liquid fuels using figures for crude petroleum
converted to coal equivalents. Furthermore,
they use an equivalency factor of 1.5 for refined
fuel oil as though there were a tonnage loss in
handling, transporting, refining, and diversion
t o non-energy uses of (1.5-1.3)/1.5 = 13%. Both
calorific factors have been used ever since
Report 2, which presents data for 1951-54.
For these four years the combined loss and
diversion, as measured by the tonnage difference
between crude petroleum and refined oil fuels,
is indeed 13% of the crude tonnage (348/2 680
tons, see Table 1). Since even a factor of 1.5
corresponds to only 18 600 B.Th.U. lb-l, an
additional calorific loss of 5 % ((19 500-18 600)/
19 500) is implied in converting refined oil
fuels to energy. About half of this loss is
accounted for by a decrease in calorific value
brought about by the refining process itself.
(Brame & King (1967) quote calorific values
(in B.Th.U lb-I) for gasoline 20 200 (p. 305),
kerosine 20 200 (p. 317), fuel oil 18 100-18 500
(p. 274). The weighted average based on world
Table 4. Carbon content o/ natural gcasa,b*c
Formula
weight
Vol. yo (g mol-I)
Carbon
content
(g of c
per 1 of
total gas
mixture)
Gas
(1)
(2)
Density
at O°C
(g 1-l)
(3)
CH,
CZH,
CP,
89.6
3.1
1.7
0.7
16.04
30.05
44.06
58.07
44.00
28.02
0.7168
1.341
1.966
2.591
1.963
1.250
0.4808
0.0333
0.0273
0.0150
0.0086
0.0000
0.802e
0.565
0.761
0.536
C,*IO
CO,
0.8
4.1
N?
Mixture at
0°C
Mixture at
60"Fd
(4)
a Data in col. 1 from Brame & King (1967, p. 360)
except that 1.4 % of CH, has been replmed by N,
t o agree with calorific content of 1030 instead of
1045 B.Th.U. ft-3.
Data in cols. 2 and 3 from the International
Critical Tables, 3, 3 (1929). The density (expressed
at one atmosphere partial pressure) for all gases
except CH, w w computed by the perfect gas law
(formula weight/22.414 1 mol-I).
Values in column 4 are calculated by the formula:
carbon content =(12.01 g C per atom x no. of C
atoms in formula/formula weight) x vol. % x
density. The yo carbon in the gas mixture is 0.5651
0.802 = 70.4 %,
Conversion to 60°F (15.6%) assumes an expansion
factor of 1.054 based on comparing density data, for
CH,, C,H,, CO,, and N, at 0°C and 60'F (Brame&
King, 1967, p. 508).
Weighted average.
'
Tellus XXV (1973), 2
INDUSTRIAL PRODUCTION OF CARBON DIOXIDE FROM FOSSIL FUEL3 AND LIMESTONE
3.4. Natural gas
The U.N. reports list a calorific factor of
1.332 (1.33 before Report 8) to convert 1 0 0 0
cubic meters of natural gas to 1 ton coal
equivalent, hence a calorific value for natural
).
gas of 1 0 3 0 B.Th.U. f t - 3 (9.17 cal ~ m - ~This
value compares closely with 1 045 B.Th.U. ft-’
cited by Brame & King (1967, p. 360). I n Table
4 the carbon content is deduced from the
composition quoted by Brame & King, adjusting the amount of CH, downward to yield the
slightly lower calorific value implied by the
U.N. reports. These reports do not state what
temperatures were used in determining natural
gas volumes by the various producing countries.
For the United States during recent years,
the U.N. statisticians directly convert figures
in cubic feet a t 60°F to cubic meters. (For 1968,
for example 19 322 x loo f t 3 (Minerals Year
Book, 1968, vols. 1-11, p. 723) x 35.31 ft3/ms
equals 547.2 x loo m3, which is the figure in
U.N. Report 14, p. 86.) Since the United States
accounts for over half of the total production
and since other major producing countries also
report volumes a t temperatures near the U.S.
value, this value will be used in calculating
emissions. The calculation of the carbon content
of natural gas, based on U.N. data, I would
judge to be within + 5 % of the correct value,
taking into account the uncertainty in both
density and composition.
4. Incompletely burned fossil fuel
4.1. Losses before use
Inevitably, unburned material is lost to the
environment when fossil fuels are handled,
transported, reprocessed, and ultimately consumed. Whether it is worthwhile attempting
here to calculate a given type of loss accurately
depends, however, on whether the total worldwide amount is a significant fraction of the total
production of fossil fuel and whether this
material remains uncombusted for a long enough
time to justify subtracting it from estimates
of CO, emission.
The latter point will be discussed in detail
in section 5 , below. The first point is illustrated
by the spillage of crude oil from ocean tankers
and other ships which together with all other
discharges into the world’s waters amounts t o
about 2 million metric tons (mmt) per year in
Tellus XXV (1973), 2
181
1969 (SCEP, 1970, p. 267). I n spite of widespread conspicuous changes created by this
spillage, the total reported loss was a mere 0.1 %
of the petroleum produced and consumed in
1969. Since solid fuels also produce conspicuous
spillage a t low fractions of total production, it
seems reasonable to assume that solid fuel loss
is also insignificant from the point of computing
CO, emissions.
Natural gas losses, principally as methane,
CH,, are not so easily noticed since the discharged gas is dispersed into the earth’s entire
atmosphere. No close check is yet being kept
of changes in the amount of CH, in non-urban
air, and large discharges of natural gas may go
undetected. NAPCA [ 19701 reports as “negligible” the CH, contribution to hydrocarbon
emissions in the United States, but this conclusion does not agree with data for 1968 from
the Minerals Year Book (1968, vol. 1-11,
p. 734) where 325 million cubic feet (mcf) of
produced natural gas in the United States
were reported lost or unaccounted for, and 828
mcf more were lost in processing. Additional
smaller losses of processed gas (ibid., p. 744)
comprise 36 mcf vented or flared and 37 mcf
unaccounted for. These losses do not include
1828 mcf used as plant and pipe line fuel, and
since only a small amount is reported as
“flared,” most of the losses appear to represent
emissions of unburned hydrocarbon. Compared
with a total production of 19 332 mcf (547 m3),
the reported loss is about 6%. Brame & King
(1967, p. 359) imply a n even higher figure for
the United States stating that “it is doubtful
if the wastage is now over 10 percent”. (In
earlier years much larger losses evidently
occurred.) From these figures an average loss
of 8 + 2 % for 1950-68 seems a reasonable
estimate. This figure corresponds to a worldwide emission of 40 to 70 mmt year-’ in 1968.
4.2. Further lOsSe8 and diversions
For coal and petroleum even if losses to the
environment during handling and transport are
insignificant, discharges of byproducts during
reproeessing and ultimate use are too large t o
ignore. For the United States, NAPCA (1970)
has summarized comprehensive data on nationwide emissions of particulates, carbon monoxide,
and hydrocarbons. The data were obtained by
first determining characteristic discharges of
individual sources and then summing on the
182
C. D. KEELING
basis of some factor which defined the size of
the operation. For example, automobile emissions of hydrocarbons and carbon monoxide
were determined from data on distances traveled,
using emission factors scaled according to size
and age of the engine, weight of vehicle, and
speed. Particulate emissions, on the other hand,
were estimated from the amount of fuel consumed. The data on road vehicles are comprehensive because of wide interest in automobile pollution. The NAPCA procedure of
summing over located sources of specific type
may, however, have led to underestimation for
other sources. Aircraft emissions, for example,
were summed only for operations below 3 000
feet (914 m) presumably because the study was
aimed a t assessing pollution only when it
occurred close to human victims. Additional
underestimation is likely if the NAPCA data
are used to define world emissions because the
United States, with the most advanced antipollution devices in use anywhere, conceivably
may produce less pollution per unit of fuel
than the world average.
4.3. Particulate emissions
NAPCA (1970, p. 11) estimated that 25.5
mmt of solids were added to the air by the
United States in 1968 from all sources. Of this
total 1.5 mmt are clearly identified with combustion of liquid and gaseous fossil fuels (mostly
the former) and 7.5 mmt to coal burning. An
unstated, but probably large, fraction of the
coal particles were inorganic ash. Compared
with a U.S. consumption in 1968 of 458 mmt
of coal and lignite and 564 mmt of refined
oil fuels (U.N. Report 14, pp. 32 and 56
respectively) these emission figures are not
impressivc; particles probably account for less
than 1 % of the solid and liquid fuel production
For the world as a whole the fraction may be
larger, but an over-all figure of 1 1yo (0 to 60
mmt of air borne particulate carbon in 1968)
is probably a reasonable guess.
4.4. Carbon monoxide emissiona
I f the reported total for aircraft (2.2 mmt) is
increased by a factor of 5 to allow for neglected
flight time above 3 000 feet, the total United
States emission of carbon monoxide, CO, for
1968, according to NAPCA (1970, p. 6), was
98.7 mmt of which 69.1 mmt was from gasoline,
0.6 mmt from other petroleum products, 1.8
mmt from coal and 15.7 mmt from industrial
processes and solid waste disposal. Motor
vehicles were the major source, accounting for
54% of the total discharge. This discharge
contained 13% of the carbon in the gasoline
burned on roadways that year, a stark reminder
of the inefficiency of the combustion process of
the internal combustion engine. I n contrast,
only 0.03% of the carbon in coal was known
to be converted to CO by combustion in 1968.
Because the United States in 1968 consumed
55% of the gasoline but only 35% of the world
total of liquid fossil fuel, the gasoline derived
emissions are computed separately to arrive a t
world figures for CO emissions. From the
NAPCA report it is difficult to decide what
part of the emission from industrial processes
and solid waste disposal is derived from coal
and petroleum and what part from non-fossil
fuel. Since the United States consumes fairly
similar fractions of the world’s solid and liquid
fossil fuel excluding gasoline (20 and 27%
respectively in 1968), extrapolation to produce
world figures can be made without deciding this
partitioning. For convenience, the data in Table
5 are presented in terms of liquid hydrocarbons
alone, but this does not mean to imply a complete absence of CO from coal. Depending on
whether the industrial and waste disposal
emissions are excluded, tha total is 4.3 or 6.1 %
of the refined fuel consumption. If we presume
that these figures can be used as lower and
upper limits, the CO emission is 4.2 i:0.8 % of
the crude petroleum production. The total
emission, expressed as CO, is 140 to 200 mmt
per year.
To compare this result with previous estimates of tho total man-produced emission, I
make the assumption that the fraction of the
world production of CO attributed to the United
States for wood burning and agricultural fires
is the same as for non-gasoline fossil fuel. The
total anthropogenic production (see Table 5)
is then about 250 mmt per year, a figure not
very different from the estimate of 200 mmt
per year of Robbins et al. (1968), and 270
mmt per year of Maugh (1972).
4.5. Hydrocarbon emissions
Excluding methane, NAPCA (1970, p. 21)
reports a total emission of 28.8 mmt of hydrocarbon per year in 1968. Of this, 18.9 mmt
are clearly identified with processing and comTellus X X V (1973), 2
INDUSTRIAL PRODUCTION OF CARBON DIOXIDE FROM FOSSIL FUELS AND LIMESTONE
183
Table 5 . Emhaion of carbon monoxide gas i n 1968 (in 10' tons y r - l )
Consumption of fuel in United Statesa
Air emissions calculated as CO
Air emissions calculated as hydrocarbonb
Fraction emitted
World consumption of fuela
Calculated air emissions as CO
Calculated air emissions as hydrocarbon
Fraction of total fuel consumption emitted
Fraction of total crude petroleum production
emitted
Gasoline
(1)
Other
liquid
hvdrocarbon
f;els
(2)
256
69.1
35.2
13.8 %
466
125.5
64.0
308
2.P18.1
1.2-9.2
0.4 %-3.0 %
1144
8.8-67.1
4.5-34.2
C
C
Total
liquid fuel
(3)
Non-fossil
fuel sources
(4)
564
22.1
1610
134.3-192.5
68.6-98.2
4.3 %-6.1 %
63.0d
3.6-5.0
Calculated emissions for world, all sources: 192.5 + 63.0 = 255.5 ( x lo6 tons CO yr-l)
a
Data for inland consumption from United Nations Report 14, p. 56.
Assuming one ton of CO is produced by 0.510 tons of hydrocarbon.
Assumed equal t o value for the United States.
Assuming ratio of United States emissions to world emissions same as for other liquid hydrocarbon fuels.
bustion of petroleum, 0 . 4 mmt with coal. I n
addition, 5.6 mmt from industrial processes
and solid waste disposal are probably derived
predominately from petroleum. Motor vehicles
alone discharged 14.0 mmt, or 5.5% of the
total gasoline consumption. To estimate a
world figure for hydrocarbon emissions, emissions from gasoline have been calculated separately (Table 6),as was done in the case of carbon
monoxide. If it is assumed that the fraction
of uncombusted hydrocarbon for each class
of fuel was the same world-wide as for the
United States, the emission is 2.7 or 4.0 % of the
refined oil consumption depending on whether
the emissions from industrial processes and
waste disposal are excluded or included. The
fraction emitted probably diminished gradually
from 1949 to 1969. All evidence considered, a
figure of 4 + 2 % seems a reasonable estimate
of the average fraction of crude petroleum
discharged into the air a8 hydrocarbons since
1949. The fraction for coal I judge to be negligible for purposes of computing carbon dioxide
emission.
Table 6. Emiaaion of gaseous hydrocarbons in 1968 ( i n 10' tons yr-l)
Consumption in the United States'
Air emissions
Fraction emitted
World consumptiona
Calculated air emissions
Fraction emitted
Gasoline
(1)
Other
liquid
hydrocarbon
fuels
(2)
256
14.0
5.5 %
466
25.5
b
308
4.9-10.5
1.5 %-3.4 %
1144
18.3-38.8
b
Calculated emissions for world, all sources: 64.3 + 15.1 = 79.4 ( x
' Data for inland consumption from United Nations Report
* Assumed equal to value for the United States.
Total
liquid fuel
(3)
Non-fossil
fuel sources
(4)
564
5.3
1610
43.8-64.3
2.7 %-4.0 %
loe tons hydrocarbons
15.1'
yr-l)
14, p. 56.
Assuming ratio of United States emissions to world emissions same as for other liquid hydrocarbon fuels.
Tellus XXV (1973), 2
184
C. D. KEELING
Table 7. Consumption of refined liquid hydrocarbons i n 1968
United States
Fuel uses
Liquified gasesd
Gasoline
Still gas
Kerosine
J e t fuel
Distillate fuel oil
Residual fuel oil
Not specified
Volumetric
basis'
(10" barrels)
(1)
Sp. gr.b
(2)
Conversion
factor
barrels ton-'
(3)
339
1956
150
103
348
863
680
0.54
0.74
0.74
0.81
0.81
0.87
0.95
11.64
8.50
8.50
7.76
7.76
7.23
6.62
Total
4 439
Non-energy uses
Asphalt
Road oil
1411
7
Lubricants
48'
wax
4
Miscellaneoua finished
18
products'
Cokes
76
Special naphthas
27
Petrochemical feed stock
93
Liquified refinery gases
47
for chemical use
Total
Combined uses
Total
Weight basis (10" tons)
World
Computed
from col. 1
(4)
(6)
29.1
230.1
17.6
:::;>
119.4)
102.7
557.0
From U.N. Report 14c
(5)
256.0
467.9
58.1
156.5
221.6
1040.9
44.9
74.0
580.6
1739.3
25.7=
0.91
0.91
6.91
6.91
6.9
0.6
0.91
0.91
0.81
0.81
6.91
6.91
7.76
7.76
2.6
11.0
3.5
12.0
0.54
11.64
4.0
461
66.3
4 900
' Data from Minerals Year Book (1968, vols. 1-11, pp. 827-8). One barrel is 42 gallons (159 liters).
* Specific gravity from United Nations Report 14, p. 3. Still gas is assumed t o have same sp. gr. as gasoline;
naphthas and petrochemical feed stock as kerosine; other non-fuel products as undifferentiated fuel oil.
Table 9, inland consumption plus bunker fuel.
71 x 10" barrels are liquified refinery gas (LRG). The remaining 268 x 10" barrels are liquified petroleum
gas (LPG).
Data from Minerals Yea7 Book (1968, vols. 1-11, p. 820). Conversion from short tons using factor 907.2
kg (s. ton)-'.
Absorbing, insulating, and medicinal oils; insecticides; petrochemicals and solvents.
'
4.6. Non-energy uses of hydrocarbons
A sizable fraction of petroleum products are
n o t combusted for energy, b u t rather a r e employed as base stock for road and building
construction, in chemical industry, or as lubricants. Tho total a m o u n t of non-fuel hydrocarbons was estimated b y SCEP (1970) as the
difference between crude petroleum and refined
oil fuel production. The difference (see Table 1,
column 8) diminished from 1 7 % of t h e crude
petroleum production i n 1950 to 9 % i n 1969,
a 19-year period during which fuel uses of
hydrocarbons increased considerably faster than
non-energy uses. Figures for 1929 and 1937
suggest that t h e fraction remained near 1 7 %
before 1950 during 20 years of m u c h less rapid
industrial growth.
The large variety of non-energy uses of
hydrocarbons is illustrated i n Table 7, based o n
volumetric data supplied b y the U.S. Bureau
of Mines. As a check, figures for refined fuels
a r e also calculated, and yield a t o t a l tonnage
4 % lower than given in U.N. Report 14.
This discrepancy m a y be because t h e U.N.
Tellus XXV (1973). 2
INDUSTRIAL PRODUCTION OF CARBON DIOXIDE FROM FOSSIL FUELS AND LIMESTONE
185
Table 8. Balance sheet for liquid hydrocarbons during 1968
United States
10” tons
(2)
Crude petroleum produced
Net import of crude petroleum
Liquified natural gas produced
Loss of crude
Net available for processing
Net import of refined fuels
Processing gain in volume
Increase in stocks
Unaccounted for
Total supply
3 332”
470’
550’
- 4a
4 348
483a
117’
- 55a
7a
4 900
473b
65b
47=
1956*
3b
75d
585
72e
2 034
-
-
650
2 027
Fuel uses
Non-energy uses
Adjustment to balance supply
Total demand
4 439a
461‘
583h
66‘
1’c
650
1777h
250j
Line
8UPP2Y
1
2
3
4
5
6
7
8
9
10
World
10” tons
10“ barrels
(1)
- 7f
(3)
- 70
Demand
11
12
13
14
4 900
2 027
Data from Minerals Year Book (1968, vols. 1-11, pp. 827-8).
Data from United Nations Report 14, Table 8.
Conversion factor for liquified gases used (see Table 7).
Estimated by assuming ratio of United States t o world production of liquified natural gas is same as
total natural gas consumption during 1968 (560/893).
Data from United Nations Report 14, Table 9.
Assuming same average specific gravity as total supply. The conversion factor is thus 4 900/650 barrels
ton-’.
Assumes United States value, since data for other countries are lacking.
Data from United Nations Report 14, Table 9. United States value is production plus import less export.
World value is uncorrected production.
As estimated in Table 7.
Assumed as the difference between fuel uses and total supply.
To equalize values for United States total supply and total demand.
a
,
k
statisticians double counted natural gasoline
derived from natural gas for which (see below)
the U.S. production was 23.4 mmt in 1968; it
cannot be due principally to applying wrong
values of specific gravity because the values
used to convert volumetric data of the Bureau
of Mines to metric tons for fuels listed in Table
7 are precisely those prescribed by the U.N.
statisticians with the minor exception of still
gas, not mentioned in the U.N. reports.
A further difficulty is the inclusion of liquified
natural gas within the category of refined fuel
oils. (See, e.g., U.N. Report 14, p. 2.) This
product is also counted as part of natural gas
production by both the US Bureau of Mines
and the U.N. Statistical Office. During 1968,
550 million barrels of liquified natural gas
were produced in the United States ( M i n e r d s
Year Book, vols. 1-11, 1968, p. 827). Of this
Tellus XXV (1973), 2
13 - 732895
total production, 199 million barrels (about 20
mmt) were designated as “natural gasoline” and
“plant condensate”, and several other products
in small amounts [ibid., pp. 756-71. A n additional 83 million barrels were used a t the refineries (10 of which were to increase stocks).
The consumption of this 282 million barrels is
evidently tabulated by t h e Bureau of Mines
under categories such as gasoline and kerosine
because the total liquid hydrocarbon consumption of 4 900 million barrels (Table 7, col. 1)
agrees with the sum of production and import
only when the total production of liquified
natural gas is included (Table 8, col. 1). The
remaining 2 6 8 million barrels of liquified
natural gas were delivered to consumers (ibid.,
p. 7651. Only this latter portion is quoted in
Table 7 as “liquified gases” where, to agree with
the Bureau of Mines tabulation, it is increased
186
C. D. KEELING
by 71 million barrels of liquified refinery gas
(LRG) derived from crude petroleum.
The reporting of liquified gases is so complicated that I cannot be sure that the tabulations in Tables 7 and 8 are free from error.
Furthermore, lacking better information, I
have extrapolated United States production
data to arrive a world figure of 75 mmt of
liquified gas in 1968 (Table 8 ) . If the latter
value is correct, the world consumption of
hydrocarbons for non-energy uses for that year
(estimated as the difference between total
demand and fuel uses) was evidently 250 mmt.
This estimate of non-energy use is not unreasonably large. The United States has been
consuming a disproportionately large part of
the world's gasoline (55%) and natural gas
(61%). If we exclude these products, i t consumed, in 1968, only 25% of the hydrocarbon
fuels produced. I n good agreement with this
percentage, the derived figure for world consumption of hydrocarbon for non-energy uses
implies a contribution by the United States of
26%. The true U.S. contribution is not likely
to be much lower than this, but it might well
be higher. A rcasonable upper limit is set by
the U.S. proportion of world production of
liquid fuel, which, according to the data of
Table 8, was 33% (583/1777) in 1968.
The 1968 world figure for non-energy use was
then between 201 and 250 mmt in 1968, or
126 % f 14 % of the difference between productions of crude petroleum and refined oil fuels
(1956-1777 = 179 mmt). Lacking better information, I will assume that the non-energy
fraction for other years remains proportional
to the difference between crude petroleum and
refined oil fuels production, i.e. i t can be derived
by multiplying the figures listed in Table 1,
col. 8 by 126% *14%.
5. Long-term fate of incompletely burned
fossil fuel
5.1. Introduction
Of the fossil fuel products which reach the
environment uncombusted, some are rapidly
converted to CO, while others remain for long
periods in their original chemical form or otherwise incompletely combusted. Both environmental observations and chemical knowledge
of these products are far from adequate. I
therefore adopt the simple approach of deducing,
or more correctly stated, guessing the average
turn-over times (mass-flux ratios) of each class
of material before complete oxidation.
5.2. Particles
Unburned fuel as soot and ash is highly
persistent. Archeological sites show ash beds
and smoked rocks made thousands of years
ago. An increasing burden of air-borne sooty
dust is reaching the ocean floor to record in the
deep sea sediments for millions of years man's
sudden and rapid consumption of fossil fuel in
the 20th century. Goldberg (1971, p. 126)
ostimates that 25 mmt per year of industrial
pollutants are now being deposited on the sea
floor. If the amount of air-borne carbon reaches
the earlier estimated upper limit of 2 % of
the production of coal and petroleum (60 mmt,
see section 4.3.), and is dispersed uniformly
over the world's atmosphere with an average
time before fall-out of 10 days, a figure suggested by Parkin et al. (1970, p. 1791), the
average concentration a t sea level would be
0.4 x
g m-3. (Atmospheric mass 5 x lo*' g,
sea level density 1.3 x lo3 g m-") This is about
twice the figure Parkin considers possibly
typical for particulate carbon in oceanic air
(ibid., p. 1792) from a study of extensive
measurements of trade wind air on Barbados
a t 1 4 " N latitude. But the types of particles
in Barbados air samples appeared to reflect
predominately ship emissions, and this even
though few ships pass near Barbados. Also, a
considerable fraction of the world-wide carbon
dust surely falls out locally over land. That
Parkin et al. (1970) nevertheless find such high
concentrations of carbon in clean tropical air
suggests that the upper limit, suggested earlier,
of 2 % of the carbon of both coal and petroleum
is not an unreasonably high figure for the
amount remaining uncombusted for a period
of many years.
5.3. Carbon monoxide
An oceanic source of this gas has been discovered recently, but oxidation of methane and
combustion by industry appear to be the
principal sources (Junge et al., 1971; Maugh,
1972). An observed concentration of 0.1 parts
per million by volume (ppm) corresponds to a
total mass of about 5 x 1014g. If the only source
Tellus XXV (1973), 2
INDUSTRIAL PRODUCTION OF CARBON DIOXIDE FROM FOSSIL FUELS AND LIMESTONE
were anthropogenic, and of the magnitude of
2.5 x 10" g yr-1 calculated in Table 5, the turnover time would be about 2 years. If we allow
for a natural source as well, the turn-over time
is still shorter. I n any case, the low concentration in air indicates that the time is so short
that for purposes of calculating C 0 2 production,
it can be assumed that CO is immediately
converted to CO,. This oxidation probably
proceeds via bacteria in soils (Inman et al.,
1971).
5.4. Air-borne hydrocarbona
Because little is known about the fate of
other hydrocarbons after discharge, the geochemistry of methane will be reviewed first.
Radiocarbon measurements made about 1950,
before nuclear bomb testing, indicate that at
least 75 % of atmospheric CH, was then derived
from recent biological material. This biological
fraction enters the air each year a t a rate of 200
to 400 mmt yr-1 according to Ehhalt (1967).An
equal amount is probably converted to CO,
by biological processes to preserve a steady
state. Just prior to 1950, according to the radiocarbon data, a n additional 65 to 130 mmt yr-1
evidently came from coal fields and natural
gas discharges. But according to data from the
natural gas industry discussed earlier, the emission in 1968 was only 40 to 70 mmt. Before 1950
even if the fractional loss was as high as 20%,
the emission could not have been over 40 mmt.
Even allowing for additional emissions from
coal fields and for possible overestimation of
the natural emission, the value based on radiocarbon data appears to be so large as to cast
doubt on the significance of the radiocarbon
measurements. These, unfortunately, were made
on air claimed from liquification plants in
industrialized urban areas. No concentration
data to check possible contamination were
obtained on the air for which the radiocarbon
was measured. I judge that the industrial
methane flux must be between 0 and 100 mmt
per year since 1960. The total methane flux
is then between 200 and 500 mmt yr-I: Since the
air concentration is currently roughly 1.5 ppm
by volume corresponding to a total mass of
methane in the atmosphere of about 4.1 x 1 O l 6 g
(Ehhalt, 1967), the turn-over time (mass-flux
ratio) is evidently between 8 and 20 years.
Lacking better information, other hydrocarbons released to the atmosphere will be
Tellus XXV (1973), 2
187
regarded to have the same turn-over time in the
environment before oxidation as methane.
5.5. Non-fuel hydrocarbona
Agreat diversity of products and a wide
range in average lifetimes are included in this
category. To bring the problem of computing
CO, production within bounds, only three
divisions will be recognized: (1) asphalt and
road oil with very long lifetimes, (2) materials
of intermediate lifetimes used in chemical
industry, and (3) other short-lived products.
Beginning with a discussion of asphaltic
materials, most of these are still serving the
purposes for which they were originally put
into service, largely as water-proof material in
roofing and road construction. Highway engineers consider the lifetime of an asphalt
pavement to be of the order of ten years, after
which mechanical failure of the mixture of
asphalt and aggregate requires repaving or
abandonment (Vallerga et al., 1970). I n the
last fifty years almost all roads have been
repaved by adding a new surface on top of the
broken one. Only a negligible amount of asphalt
oxidizes before burial by new pavement. The
asphalt fraction (including road oil) for the
United States in 1968 was 39% by mass of
non-energy hydrocarbons consumed. This value
is higher than Brame & King (1967, p. 236)
report for the United States in 1955 (only 26 %)
but slightly less than their world figure for 1955
(41%). From these data, I estimate the relative
uncertainty in the asphalt fraction to be 1/3
(the fraction, then, is 39 13%). The figure of
39% will be assumed for the entire period
1949-1969.
An intermediate turn-over time will be
assigned to naphthas, petrochemical feed stock,
and liquified refining gases which are starting
materials for a wide range of chemical products.
These comprised 29% by mass of the United
States consumption of non-energy products in
1968. This figure cannot be readily compared
with the above mentioned data for 1955 because Brame & King (1967) combine in the
category of "others" all products except bitumen, lubricants, and waxes. The U.S. fractions
for 1968 will therefore be used as the world
average for the entire period 1949-1969. To
avoid introducing more complexity than available knowledge justifies, the turn-over time
will be considered to be the same as for methane
188
C. D. KEELING
and air-borne hydrocarbons; i.e. 8 to 20 years.
Since the estimated uncertainty in turn-over
time is considcrable, any additional error in
estimating the fraction will be neglected.
The remaining products: lubricants, waxes,
coke, and miscellaneous (32% by mass of the
total in 1968) are mostly burned during or soon
after use, and will be assumed to yield CO,
immediately.
The division between short and intermediate
turn-over times does little more than separate
the products, after accounting for asphalt, into
two nearly equal fractions, one of which is
treated like fuels and the other like air-borne
hydrocarbons. Errors in assignment occur in
both directions and it can only be hoped that
these tend to cancel. For example, 0.5 to 1.0
mmt of automotive lubricants, assigned zero
turn-over time, are discarded as unburned waste
(SCEP, 1970, p. 270), and a sizeable fraction of
insecticide (total production about 1.0 mmt
in 1968 (SCEP, 1970, p. 259)) probably persists
for several years. On the other hand, chemical
products assigned an intermediate turn-over
time, e.g. plastics, are partially burned soon
after use.
5.6. Conversion of unburned fuel products to COP
An accurate evaluation of the CO, produced
by fossil fuel combustion depends on taking
proper account of subsequent transformations
of the fraction discharged into the environment
unburned. If we for the moment leave out of
consideration that small amount which never
oxidizes, the amount of industrial CO, produced
is less than if all products were immediately
burned to completion only if rates of discharge
have recently bcen increasing. Otherwise, final
oxidation of earlier accumulations equals or
exceeds the amount of unburned products
newly discharged. I n the above context the
precise meaning of “recently” for any given
product depends on its typical rate of conversion to CO,, here expressed in terms of a turnover time. To calculate CO, production rates for
each type of discharge, I will now introduce a
simple mathematical model which is, however,
as realistic as the data justify.
The production of all types of fossil fuel
combined has been rising steadily with a doubling time of about 15 years since the end of the
1939-1945 world war. (For example, the t,otal
carbon released in 1968 was very nearly twice
that in 1953.) If p denotes the rate of production of fossil fuel, and if the rise is exponential:
p
=
p aekp t
(5.1)
where p a is the production rate a t t = 0, the
beginning of the period of interest, and
T,
=
l/k,
is the time for p to increase by a factor e (2:2.7).
I f the doubling time is 15 years, ,,z, = 15/ln 2 = 22
years.
If we now fix our attention on a specific type
of fuel discharge, the amount accumulated in
the environment since t = O is augmented each
year by an amount we will call m, where:
m
=
m,ekmt
(5.2)
analogous to (5.1) where the value of k , may
differ somewhat from the average value, k,,
quoted earlier. At the same time the accumulated amount, call it M , is reduced by various
conversion processes. Although the available
environmental data offer little information
to guide us, conversion is predominately to CO,
and probably proceeds a t a rate roughly
proportional to M . Assuming so, the rate of
loss to CO, can be expressed as k,M where
k M is the inverse of the turn-over time, zM.
The total rate of change in M is then:
dM
-= m - k M M
dt
(5.3)
Integrating from t = O to t using (5.2) with the
approximation k , N k,:
The ratio of the rate of conversion to CO, t o
the rate of input, m, is
If t is large compared to ( k P + k M ) - l , the
transient described by the factor in brackets
becomes negligible and:
R z - -k Mkp+k,
-
XU
zP,+tM
(5.6)
Tellus X X V (1973),2
INDUSTRIAL PRODUCTION OF CARBON DIOXIDE FROM FOSSIL FUELS AND LIMESTONE
Because fossil fuel consumption began early
in the nineteenth century, it might appear
reasonable to assign a value to t in equation
(5.4) of the order of 100 years. However, from
1929 to 1945 the production rate rose so slowly
that the effective value of t in 1960 is closer
t o 30 years for cases of small k, where the
transient is most significant.
The ratio is shown below for typical turnover times, t,, with k, = 1/22 years and with t
set equal to 30 years as appropriate for approximately the midpoint of the post-war years
1945 to 1969:
Table 9. Apparent fraction of discharged fuel
converted to CO, ( = R )
Turn-over
time (ty)
Calculated
by ( 5 . 5 )
Calculated
by ( 5 . 6 )
( %)
( %)
72
73
52
31
189
tion. Finally the probable error is estimated
from the upper and lower limits. For example,
for hydrocarbon emissions, t, is estimated t o
lie between 8 and 20 years, hence, using (5.6)
with t p= 22 years, 1 - R = 0.37 k 0.10. The
unconverted fraction is:
(4 +a%) x (0.37 +0.10)
=
1.5
1.1 %
fraction partial fraction
lost
unconverted
where the combined error is computed by the
law for adding systematic errors (see section
6.2. below).
For non-energy products, if A denotes the
difference between crude petroleum and refined
fuel oil given in Table 1, column 8, the unconverted amount is:
A (1.26 k0.14) [(0.39 T0.13) x 1.00 +0.29
8
20
50
49
27
I n view of the uncertainties in estimating
turn-over times in sections 4 and 5 , the relation
given by (5.6) is as close an approximation as is
justified for purposes of computing CO, production in this report.
6. Calculation of COz production
6.1. Fraction of fossil fuel converted
The data which bear on computing the
amount of fossil fuel carbon converted to CO,
each year are summarized in Tables 10 and 11.
Losses and diversions are taken into account
in estimating factors for the fraction of the
total production of fuel converted to CO,.
For those cases in which fuel lost or diverted
is later on oxidized to CO,, the rate of this
process has been estimated as follows. First, a
range in turn-over times, as discussed in the
last section, is used, with the aid of equation
(5.6), to compute upper and lower limits of the
partial fraction, 1 - R , not converted to C 0 2 .
Then the mean of the limits of 1 - R , chosen
as the most probable unconverted partial fraction, is multiplied by the total loss or diversion
to obtain an estimate of the unconverted fracTellus XXV (1973),2
x
(0.37 + O . l O ) ]
=
A(0.63 k0.27)
(6.1)
where the factor in the first parentheses is the
estimated correction to include liquified natural
gas products (see section 4.6), and the two
terms in brackets represent t h e fractional
contributions to A owing to asphalt and petrochemicals respectively. Each term is computed
as the fraction of total non-energy products
multiplied by the non-conversion factor, 1 - R,
estimated from the turn-over time. For asphalt
an uncertainty is assigned to the fraction of
total products, for petrochemicals, an uncertainty is assigned to the value of 1 - R . The
combined error shown on the line below is
computed by the law for systematic errors.
For three representative years, the unconverted
amounts expressed as fractions of the crude
petroleum production are:
1950:
1959:
1968:
(10.8 *4.7)%
( 6.2 2 2 . 7 ) yo
( 5.8 T2.5)%
Between 1959 and 1968 the fractional loss
remains nearly constant. I n view of the increasing uncertainties in computing the loss for
earlier years, the value for 1959 (rounded to
6 % ) is assumed in Table 10 to apply to the
entire period, 1949-1969, but with the relative
uncertainty raised to + 4 % to allow for the
further approximation involved.
190
C. D. KEELING
Table 10. S u m m a r y of losses and diversions of fossil fuels
Loss in combustion
-
Loss in
handling
and
transport
a23
particles
hydrocarbon
(1)
(2)
(3)
as CO
(4)
1+1
b
b
Coal
Percent of total coal production
thus diverted
Neg.a
Turn-over time of diverted
fraction
Percent of diverted fraction
not oxidized
Percent of coal production
thus not oxidized
Crude petroleum
Percent of total crude
production thus diverted
Turn-over time of diverted
fraction
Percent of diverted fraction
not oxidized
Percent of crude production
thus not oxidized
Neg.
Natural gas
Percent of total gas production
thus diverted
8+2
Turn-over time of diverted
fraction
8-20yr
Percent of diverted fraction
not oxidized
37 f 10
Percent of gas production thus
not oxidized
3 1.6
as
Total
fraction
Non-energy not
use
oxidized
(5)
(6)
Neg.
Infinite
100
l*1
If1
1+1
4f2
4fl
infinite
8 - 20 yr
< 2 yr
(see text,
section 6)
100
37+10
0
1+1
1.5fl.l
0
6+4
Neg.
b
Neg.
Neg.
8.5k6.1
3+1.6
a Negligible.
Included with crude petroleum.
6.2. Analysis of systematic errors
Principal among the difficulties in obtaining
reliable estimates of the production, carbon
content, and unburned fractions of fossil fuels
is the compiling of truly global figures. Worldwide fuel summaries could not be perfectly
determined in the best of circumstances, but
because they have been assembled primarily to
satisfy commercial needs, they are still less
complete and reliable than if they had been
gathered for scientific purposes. Even estimates
of the errors are highly uncertain.
Under these circumstances i t can hardly be
supposed that these errors are all random, and
that they obey the laws for normal distributions. For example, the figures for production
are conceivably all too low because of incom-
plete reporting, or all too high because the
mine and well head recovery data are overestimated. To verify the former is practically
impossible; if the latter is true, the data could
be improved only by lengthy checking. The
carbon contents of individual fuels quoted in
the literature appear to be possibly typicalvalues,
not accurate weighted averages. The true worldwide averages are undoubtedly time dependent
as well. For solid fuels, a small decrease in
carbon content could be seen, when separate
values for coal and lignite were combined in
Table 3 (column 7). Probably the carbon contents of the separate coal and lignite fractions
also have decreased as the average grades have
decreased in quality with increasing exploitation. Furthermore, in computing the combined
Tellus XXV (1973), 2
191
INDUSTRIAL PRODUCTION OF CARBON DIOXIDE FROM FOSSIL FUELS AND LIMESTONE
Table 11. Average factors for calculations o f CO, emissions (in ton8 of carbon)
Production
Carbon
fraction
Fuel
(1)
(2)
cod
Relative error
Lignite
Relative error
Coal and lignite
Relative error
Crude petroleum
Relative error
Natural gas
Relative error
Cement
Weight (tons)
0.70
f5.8 %
0.28
& 5.8 %
0.69
k5.8%
0.84
f3.6 %
540 g
k 5.0 %
0.1373a
f8.0 %
Weight (tons)
f-8.0 %
Weight (tons)
k8.0 %
Weight (tons)
k3.0%
Volume (m3)
k3.0 %
Weight (tons)
Fraction
oxidized
to co,
Factor
(3)
(4)
0.99
f 1.0 yo
0.99
f 1.0 %
*
0.99
1.0 %
0.915
k6.2%
0.97
k 1.6 Yo
1.00
wt. x 0.693
f 14.8 %
wt. x 0.277
k 14.8 %
wt. x 0.683
14.8 %
wt. x 0.769
f 12.8 %
vol. x 524. x
k9.6%
wt. x 0.1373
~
This figure is based on a CaO weight fraction of 64.1 % (Lea & Desch, 1940), and assumes that one mole
of CaO (mol. wt. 56.08) comes from one mole of CaCO, (at. wt. of carbon 12.01): Thus 0.1373 =0.641 x
a
(12.01/56.08).
error from all sources, the number of different
kinds of errors affecting any given type of fuel
are too few to treat the set of errors as randomly
distributed. I f the confidence to be placed in
the final result is to be as high as for the individual estimates, the possibility cannot be
excluded that the errors have accumulated
without any significant cancellation. It is
reasonable, however, to make the approximation
appropriate for small differentials, since none
of the individual errors exceed 10%.
The law for combining small systematic
errors involving continuously varying parameters is given by the formula (Parratt, 1961):
au
au
Au=-Ax+-Ay+
ax
ay
...
(6.2.1)
.
where direct measurements of x,y , and . ., with
systematic errors A z , A y , and ..., lead to a
calculated value of u according to a functional
relationship:
u k A u = ~ ( ~ t A X , y k A...)
y,
(6.2.2)
I n this study the following specific formulas
are used:
Au Ax Ay
Product: u = x y ..., - = - + - +
U
X
...
Y
(6.2.3)
Sum:
Au A x f A y
u = x + y , -=
u
xty
Tellus XXV (1973), 2
(6.2.4)
Inner product: u = 1 i
xy,
&
u
u
YAXf x A y
l+xy
(6.2.5)
Because the distribution of errors is not
known, the confidence to be placed in assigned
errors following the symbol k cannot be
accurately specified. I have attempted to assign
error estimates which are large enough to give
a 90 to 95 % probability that the true value lies
between the limits u + A u and u - A u , but even
such a confidence limit is only a rough guess.
On the basis of the relative errors shown in
Table 10 the overall error when all fuels are
combined is approximately 13% after 1960. It
was somewhat higher in earlier years because
of a higher proportion of solid fuels. The published production figures for years before 1949
are likely to be systematically too low because
of incomplete reporting. The carbon contents
of solid fuels, however, were probably higher
than recently. Errors arising from using the
fixed factors in Table 11 to compute emissions
in earlier years thus may tend to cancel, but
this cannot be proved, and the data before 1949
could well be in error by 20 % or more.
6.3. Carbon dioxide production from cement
manufacture
The kilning of limestone to obtain calcium
oxide for the manufacture of portland cement
produces carbon dioxide as a byproduct. Since
192
C. D. KEELINU
Table 12. World production of carbon dioxide
from cement
Source of dataa
no.
(1)
tion
(2)
Cement
production
lo6
barrelsb
(3)
(4)
(5)
1949
1950
1951
1952
1953
1954
1955
1956
1957
1958
1959
1960
1961
1962
1963
1964
1965
1966
1967
1968
1969
1953
1954
1955
1956
1957
1958
1959
1960
1961
1962
1963
1964
1965
1966
1967
1968
1968
1968
1968
1968
1969
280
267
274
296
302
272
287
303
391
371
347
319
271
438
273
247
247
247
247
247
255
674
782
879
945
1052
1143
1275
1381
1448
1556
1726
18561956
2 102
2 217
2 437
2 543
2 722
2 832
2 985
3 169
15.8
18.3
20.6
22.1
24.6
26.8
29.8
32.3
33.9
36.4
40.4
43.4
45.8
49.2
51.9
57.1
59.5
63.7
66.3
69.9
74.2
Year of
publics- Page
Year
CO,
production
lo6 tons
carbon
Publication date and page no. refer to Vol. I of
Minerals Year Book for each year cited.
1 barrel equals 376 Ib. or 0.17055 metric tons.
a
*
1945 the world production of cement has risen
rapidly, quadrupling between 1950 and 1969.
The annual release of CO, from kilning is now
2 % of that from fossil fuel combustion.
The annual production of cement and associated CO, released since 1948 are shown in
Table 12. Uncertainties in the cement data
and in the conversion factor to CO, have not
been estimated because they are too small to
influence the uncertainty in total industrial
CO, production.
6.4. Tabulated results, 1860 to 1969
The annual CO, production from industrial
activity for 1929, 1937, and the period 1949
to 1969 is shown in Table 13, based on data
listed in Tables 1 and 12, and the conversion
factors of Table 11. I n Table 14 the annual
production of CO, from fossil fuel is extended
back to 1860 based on fuel production data
tabulated hy the United Nations (1955, pp.
28-29). Overlapping data for 1929, 1937, and
1949 to 1953 agree within 1%. Beginning with
the year 1949, CO, from limestone is included.
The annual increase in atmospheric CO,
mixing ratio if all emitted CO, remained in the
air, A p , is calculated according to the formula:
Ap =
wt. of C in fossil fuel
produced in one year (g) xmol.wt. of air x lo6
at. wt. of C x total mass of dry air (g)
=wt. of C ~ 2 8 . 9 6x lO6/(12.Ol ~ 5 . 1 1 x9 loz1)
= 4.711
x
x wt. of C (g)
where the molecular weight and mass of dry
air are quoted from the recent study by Verniani
(1966).
For the period before 1860, only fragmentary
production data are available, but one can
approximately deduce the CO, emissions by
assuming an exponential growth a t the same
rate as found for 1860 to 1900. On the basis of
the values in Table 14, the production of CO,
before t , = 1860 is given by the relationship
Pfuel
= 92 eo.ol:15<t-td
x
lofitons of carbon
(6.1)
where t is the calendar year.
Similarly, if we assume a long term exponential growth in CO, emissions from cement
manufacture a t the rate found for 1949 to 1969,
the production of CO, before 1, = 1950 is given
by:
Pcernent =
18e0.0785(L- t ? )
x
loRtons of carbon
(6.2)
This relation agrees fairly closely with a graph
by Baxter & Walton (1970, p. 218) based on
data extending back to about 1920.
The cumulative output of CO, from fossil
fuel based on the values of Tables 13 (Trend A )
and 14 is 1.092 x 1017 g of carbon (4.00 x 1017 g
of CO,) for 1860-1969. This is 17.7% of the
amount of CO, in the atmosphere in the late
nineteenth century if we accept the estimate of
Bray (1959) that the atmospheric mixing ratio
a t that time was about 290 ppm. When cement
is included (assuming relation (6.2) to hold
before 1950) the result is increased to 1.103 x 1017
g of carbon, or 17.9% of the atmospheric CO,.
The total for 1800 to 1969 is 1.123 x 10'' g
Tellus XXV (1973), 2
INDUSTRIAL PRODUCTION O F CARBON DIOXIDE FROM FOSSIL FUELS AND LIMESTONE
193
Table 13. World production of C O , from fossil fuel and cement from 1929-1968a ( i n units of 108 t o m
of carbon)
Year
Coal
Crude
petroleum
Natural gas Cement
Total
Fraction of
atmosphere x lodb
964.4
958.9
1 008.1
1096.2
1 137.2
1 134.5
1139.9
1137.9
1234.2
1288.8
1334.6
1383.8
1435.0
1496.6
1381.0
1415.9
1474.6
1528.6
1549.0
1577.0
1 506.7
1550.4
1591.4
163.0
225.3
370.7
414.5
476.8
501.4
529.1
553.7
606.7
659.8
693.6
711.3
766.7
825.9
879.0
952.0
1022.0
1 104.3
1 183.5
1285.0
1378.0
1504.2
1621.1
29.9
40.9
88.6
102.7
119.5
129.4
136.8
143.6
156.7
169.8
185.5
200.2
223.7
244.7
264.1
288.7
315.4
344.8
368.9
401.4
431.8
469.5
514.0
15.8
18.3
20.6
22.1
24.6
26.8
29.8
32.3
33.9
36.4
40.4
43.4
45.8
49.2
51.9
57.1
59.5
63.7
66.3
69.9
74.2
1 157.3
1 225.1
1483.1
1631.7
1 754.0
1 787.4
1830.4
1861.9
2 027.4
2 150.7
2 247.6
2 331.7
2 465.8
2 610.5
2 569.9
2 705.8
2 863.9
3 034.7
3 161.0
3 327.1
3 382.8
3 594.0
3 800.7
0.55
0.58
0.70
0.77
0.83
0.84
0.86
0.88
0.96
1.01
1.06
1.10
1.16
1.23
1.21
1.27
1.35
1.43
1.49
1.57
1.59
1.69
1.79
711.3
766.7
825.9
200.2
223.7
244.7
36.4
40.4
43.4
2 294.1
2 388.7
2 483.5
1.08
1.13
1.17
~~
1929
1937
1949
1950
1951
1952
1953
1954
1955
1956
1957
1958
1959
1960
1961
1962
1963
1964
1965
1966
1967
1968
1969
Corrected to “Trend A”’
1958
1959
1960
1346.2
1357.8
1369.4
Based on production data of Tables 1 and 12, and conversion factors of Table 11. To convert to tons of
CO, multiply by 3.664.
Moles of CO, per mole of dry air in the earth’s atmosphere.
‘ See text, page 177.
a
(or 18.2%) if we assume (6.1) to hold before
1860. On this same basis, the period before 1800
contributes negligibly (less than 0.2%) to the
accumulated CO, production up to 1969.
6.5. Comparison with other compilationa
World production data for fossil fuels have
been published for selected recent years (1950,
1955, 1960-62) by the Organization for Economic Cooperation and Development in Paris
(OECD, 1966, Table 2 of Annex 1).For member
countries, which comprise most of the highly
industrialized nations outside of the communist
trade area, the data were compiled from original
sources and thus offer a check on the U.N.
compilations. World totals quoted by the OECD
for the five years combined are 1.0% higher
than the U.N. figures of Table 1. For individual
Tellus XXV (1973), 2
years and fuels, they differ a t most by 4 %
from the U.N. data. This close agreement
probably reflects the use of essentially the same
source data by both agencies, even for OECD
member countries.
The amount of CO, attributable to fossil fuel
has been estimated in several previous studies.
Decade averages, copied directly from the
original papers, are shown in Table 15, expressed
as percent of the amount of CO, which Revelle
& Suess (1957) estimated to be in the atmosphere in 1950. All these published studies
assumed 100% conversion of raw material t o
CO,. The values of Broecker et al. (1971) and
Revelle (1965) are 1 0 % and 1 4 % higher,
respectively, than the new values derived in
this work. Broecker’s figures for the carbon
fractions except lignite (for which he used 0.42
instead of 0.28) are almost identical with those
194
C. D. KEELING
T a b l e 14. World production of CO, from fossil fuel from 1860 to 1953 ( i n units of 10' t o m of carbon)'
Year
(1)
Coal
(2)
1860
1861
1862
1863
1864
1865
1866
1867
1868
1869
1870
1871
1872
1873
1874
1875
1876
1877
1878
1879
1880
1881
1882
1883
1884
1885
1886
1887
1888
1889
1890
1891
1892
1893
1894
1895
1896
1897
1898
1899
1900
1901
1902
1903
1904
1905
1906
1907
1908
1909
1910
1911
1912
1913
1914
1915
1916
1917
1918
91.5
96.7
96.1
103.5
112.5
119.0
125.7
134.7
133.3
138.1
141.0
157.4
170.9
182.5
177.3
182.8
184.9
188.8
188.9
199.1
217.5
234.0
251.3
268.8
269.9
263.8
264.4
282.8
305.8
310.5
328.9
342.2
345.5
337.6
352.9
371.7
382.4
400.0
421.5
462.2
485.6
497.2
508.5
559.5
563.0
594.2
641.6
708.7
670.9
700.2
732.2
746.2
785.7
842.9
752.7
741.2
798.4
842.2
828.1
Lignite
(3)
1.7
1.8
2.0
2.1
2.4
2.6
2.6
2.8
3.0
3.2
3.4
3.9
4.3
4.8
5.3
5.3
5.5
5.5
5.7
6.0
6.4
6.8
7.1
7.6
7.8
8.0
8.3
8.6
9.2
9.8
10.8
11.5
11.7
12.2
12.5
13.6
14.5
15.7
16.7
17.7
19.8
21.3
20.9
21.7
22.6
24.1
25.8
28.3
29.3
29.8
30.0
31.2
34.5
35.7
33.5
34.3
36.5
37.4
39.2
Crude
petroleum
(4)
0.1
0.2
0.3
0.3
0.2
0.3
0.4
0.4
0.4
0.5
0.6
0.6
0.7
1.2
1.2
1.1
1.2
1.7
1.9
2.5
3.2
3.4
3.8
3.2
3.8
3.8
5.0
5.0
5.5
6.5
8.1
9.6
9.4
9.7
9.4
10.9
12.1
12.8
13.2
13.8
15.8
17.7
19.1
20.5
23.0
22.7
22.5
27.8
30.1
31.5
34.5
36.3
37.1
41.3
43.0
45.4
48.1
53.2
53.4
Natural
gas
(5)
0.0
0.0
0.0
0.0
0.0
0.0
0.0
0.0
0.0
0.0
0.0
0.0
0.0
0.0
0.0
0.0
0.0
0.0
0.0
0.0
0.0
0.2
0.3
0.4
0.6
0.8
1.o
1.3
1.5
1.7
2.0
2.0
2.1
2.2
2.3
2.4
2.7
3.0
3.2
3.5
3.7
4.1
4.4
4.7
4.9
5.6
6.1
6.41
6.3
7.5
8.0
8.1
8.9
9.2
9.3
9.9
11.8
12.5
1 J.3
Total
(6)
Fraction of
atmosphere x
(7)
93.3
98.7
98.4
106.0
115.1
121.9
128.7
137.9
136.7
141.8
145.0
161.9
175.9
188.4
183.8
189.2
191.6
196.0
196.5
207.6
227.1
244.4
262.5
280.0
282.1
276.4
278.7
297.7
321.9
328.5
349.7
365.4
368.7
361.6
377.1
398.6
411.6
431.5
454.6
497.3
524.9
540.3
552.9
606.4
613.4
646.6
696.1
771.2
736.6
769.0
804.8
821.8
866.2
929.0
838.4
830.8
894.8
945.3
932.0
0.04
0.05
0.05
0.05
0.05
0.06
0.06
0.06
0.06
0.07
0.07
0.08
0.08
0.09
0.09
0.09
0.09
0.09
0.09
0.10
0.11
0.12
0.12
0.13
0.13
0.13
0.13
0.14
0.15
0.15
0.16
0.17
0.17
0.17
0.18'
0.19
0.19.
0.20
0.21 .
0.230.25
0.25
0.26
0.29
0.29
0.30
0.33
0.36
0.35
0.36
0.38
0.39
0.41
0.44
0.39
0.39
0.42
0.45
0.44
lL{,I 8
loeb
Tellus XXV (1973), 2
INDUSTRIAL P~ODTJCTION OF CARBON DIOXIDE FROM FOSSIL FUELS AND LIMESTONE
195
Table 14 (cont.)
Year
Coal
Lignite
Crude
petroleum
Natural
gas
Total
Fraction of
atmosphere x
(1)
(2)
(3)
(4)
(5)
(6)
(7)
36.9
43.6
46.6
50.6
45.3
47.9
51.8
52.0
56.2
60.8
64.4
54.6
50.3
47.1
48.3
53.0
57.0
56.8
69.8
73.1
80.9
88.4
91.7
92.5
96.4
87.6
51.9
70.9
75.9
82.9
92.0
99.9
109.9
117.2
124.7
59.6
76.3
82.6
92.2
108.8
108.7
114.4
117.5
134.7
141.6
158.9
151.2
145.7
139.0
151.6
160.1
174.3
188.4
215.3
209.9
219.6
224.5
214.6
202.2
221.5
249.2
263.0
289.1
318.4
359.9
358.4
402.2
455.2
479.1
506.0
11.7
12.6
10.4
12.0
15.8
17.9
18.9
20.8
22.8
24.9
30.3
28.4
27.6
25.7
25.9
29.4
30.1
34.9
38.8
37.0
39.8
42.9
45.4
49.4
55.1
59.8
65.5
67.8
75.0
84.5
89.2
103.2
122.2
131.4
138.0
1919
1920
1921
1922
1923
1924
1925
1926
1927
1928
1929
1930
1931
1932
1933
1934
1935
1936
1937
1938
1939
1940
1941
1942
1943
1944
1945
1946
1947
1948
1949
1950
1951
1952
1953
720.7
826.4
688.3
735.8
835.4
824.0
821.3
815.7
883.7
863.5
918.3
843.3
744.6
662.0
693.0
754.0
770.3
866.3
902.3
841.3
892.6
944.6
985.4
990.3
991.0
955.6
823.3
842.7
952.2
990.3
930.0
1007.6
1 056.1
1038.8
1039.5
828.9
958.9
828.0
890.6
1005.3
998.5
1006.4
1006.0
1097.5
1090.9
1171.9
1077.5
968.2
873.8
918.8
996.5
1031.7
1146.4
1226.2
1 161.4
1232.9
1300.4
1337.1
1334.4
1364.0
1352.2
1203.6
1270.5
1421.5
1517.5
1469.6
1613.0
1743.5
1766.5
1808.1
Based on production data of Tables 1 and 11, and
of CO,, multiply by 3.664.
Moles of CO, per mole of dry air in the earth’s atmosphere.
a
of this work, while Revelle’s are somewhat
higher. In comparison, the older figures of
Revelle & Suess (1957) are 28 % higher than the
new values. These authors did not state how
they converted fossil fuel production of the
United Nations (1955) to CO, emissions. Since
the fraction of carbon in petroleum and natural
gas is fairly well established and these fuels
anyway contributed only about 30% of the
total carbon combusted in the 1940 and 1950’s,
the difference between the figures of Revelle
and Suess and those from the updated study
of Revelle (1965) is probably a result of their
assuming a carbon fraction for coal of about
88 %, a value appropriate to hard coal.
The high values of Bolin & Eriksson (1959)
Tellus XXV (1973), 2
loeb
0.39
0.45
0.39
0.42
0.47
0.47
0.47
0.47
0.52
0.51
0.65
0.51
0.46
0.41
0.43
0.47
0.49
0.54
0.58
0.55
0.58
0.61
0.63
0.63
0.64
0.64
0.57
0.60
0.67
0.71
0.69
0.76
0.82
0.83
0.85
of Table 10. To convert t o tons
are derived from a not-too accurate exponential
f i t of the data of Revelle & Suess, but equally
high are the recently published values of Bolin
& Bischof (1970) which are based on estimates
by the OECD (1966). Because the authors quote
no conversion factors, one cannot be sure how
they obtained 4.4% as compared with 3.5% of
Revelle (1965) when using almost identical
data for fuel production. It is, however, possible
that they assumed 100% of fuel tonnage to
be carbon, since this assumption yields a value
of 4.38%.
I n another recent computation, Baxter &
Walton (1970) used United Nations data and
somewhat higher conversion factors than Revelle
(1965). Their values are presented on a logarith-
196
C. D . KEELING
Table 15. GO, added to the atmosphere by fossil
fuel combustion expressed as fraction of amount
in the atmospherea
Journal Reference
This work
Broecker et al. (1971)
Revelle (1965)
Revelle & Suess (1957)
Bolin & Eriksson (1959)
Bolin & Bischof (1970)
~
1940-1949
%
(1)
1950- 1959
%
(2)
3.08
3.39
3.49
3.90b
4.40b
4.40
2.10
2.30
2.39
2.71
3.17
~
~
x 1018 g CO, (0.41 x 10’’ g C) as calculated by
Revelle & Suess (1957), assuming that dry air
contains 300 ppm CO, and that the mass of the
atmosphere is 5.2 x loz1g.
Based on estimates of future fossil fuel usage.
a 2.35
mic graph from which I cannot read off accurate
values. They state, however, that their estimates “are within 5 % of the less detailed
figures calculated by Revelle & Suess (1957)”.
Their calculated values are likely to be too
high because they used high conversion factors
for coal and lignite (0.80 and 0.73, respectively).
Since the variability in the above estimates
principally stems from different values for the
carbon fraction of fuels, it would be of interest
to look closely into the criteria used by the
different authors to estimate these factors.
This is not possible, however, because none of
the authors stated how they obtained their
values. Baxter & Walton (1970) appear to have
been informative when they list four references
serially after a list of five fuel factors, but one
of these references is to a three volume treatise
which I have searched without finding any
reference to average carbon fraction of fuels,
while another reference is to a paper by Rohrman, Steigerwald & Ludwig (1967) who themselves simply “assumed” values for coal, crude
oil, and natural gas, all of which differ from
values used by Baxter & Walton. A recent
authoritative appraisal of carbon fractions in
fuels is evidently not be found.
7. Concluding remarks
Considerable effort was required to make even
this first step towards obtaining accurate values
of the global emission of CO,. Much more work
will be required if the presently estimated error
13 % are to be substantially reduced.
limits of
The global scope of this work demands massive
documentation and will require the cooperation
of the statisticians who compile the production
data on fossil fuels.
Especially needed are correct regional averages of the carbon content of fuels, and of losses
and diversions outside the United States. The
recommendation of the Study of Man’s Impact
on Climate [SMIC, 1970, p. 2441 that “continual
computation [be maintained] of past records
and updating of forecasts of the global consumption of fossil fuels in combustion” ought t o
be expanded to include determination of emission factors.
Ultimately these factors should be computed
country by country from reliable inland studies
of currently determined fuel compositions and
emission factors for carbon dioxide, carbon
monoxide, hydrocarbons, and air-borne particles. Handling losses also ought to be monitored.
The major concern of fuel statisticians until
now has been t o document production and
consumption of materials as they relate t o
energy consumption, and thus to provide a
baseline for anticipating future energy needs.
This goal should be expanded to reflect a
concern for the emission of byproducts related
to energy consumption. Documentation of such
emissions would provide a baseline for anticipating environmental changes arising from
man’s need for energy. The demand for energy
is certain to increase and the problem of disposing of byproducts will also grow as an ever
larger and more crowded human population
strives to improve its standard of living on
every continent of our globe.
Added in proof
As shown by Rotty (commentary to appear
in a future issue of Tellus) the foregoing computations neglect to count CO, produced by
flaring of natural gas. According to Rotty’s
analysis of flaring, the CO, production values
in the sixth column of Table 13 for 1960 through
1969 should be uniformly increased by 2.1 %.
This percent increase is probably a reasonable
basis for correcting data for earlier years, but
a more consistent approach, equally concordant
with the original data, is to increase the crude
petroleum factor in column 4 of Table 11 from
0.769 to 0.813, and thus to increase the CO,
emission values for crude petroleum by 5.7 %.
Tellus XXV (1973), 2
INDUSTRIAL PRODUCTION OF CARBON DIOXIDE FROM FOSSIL FUELS AND LIMESTONE
Readers are urged to apply one or the other of
these corrections before using values in the last
two columns of Tables 13 and 14.
Acknowledgments
I am indebted to Drs M. King Hubbert,
Roger Revelle, Lester Machta, Arthur A.
Orning, and M. D. Schlesinger for advice, and to
Dr Robert Bacastow and Messrs Alexander
197
Adams and Daniel Milder for help with the
compilations. 1 am expecially indebted to Dr
Ralph M. Rotty for suggesting improvements
relating to almost every aspect of this paper
and for having taken the unusual trouble of
checking all of the source data and the calculations. This work was supported by the Atmospheric Sciences Division of the National Science
Foundation under Grants GA-13839 and GA31324X.
REFERENCE S
Baxter, M. S. & Walton, A. 1970. A theoretical
approach to the Suess effect. Proc. Roy. SOC.,
London, Series A , 318, 213-230.
Bolin, B. & Bischof, W. 1970. Variations in the
carbon dioxide content of the atmosphere in the
northern hemisphere. Tellus 22, 431-442.
Bolin, B. & Eriksson, E. 1959. Changes in the carbon
dioxide content of the atmosphere and sea due to
fossil fuel combustion. I n The atmosphere and the
sea i n motion, Rossby Memorial Volume (ed. B.
Bolin), pp. 130-142. Rockefeller Institute Press,
New York.
Brame, J. S. S. & King, J. G. 1967. Fuel, solid,
liquid and gaseous, 6th ed. 518 p. Edward Arnold,
London.
Bray, J. R. 1969. An analysis of the possible recent
change in atmospheric carbon dioxide concentration. Tellus 11, 220-230.
Broecker, W. S., Li, Y. H. & Peng, T. H. 1971.
Carbon dioxide-man’s unseen artifact, Chapter
11 in Impingement of M a n on the oceans (ed. D.
Hood), pp. 287-324. Wiley-Interscience, New
York.
Callendar, G. S. 1938. The artificial production of
carbon dioxide and its influence on temperature.
Quarterly J. of the Royal Meteorological SOC.64,
223-240.
Callendar, G. S. 1957. The effect of fuel combustion
on the amount of carbon dioxide in the atmosphere, Tellus 9, 421-422.
Callendar, G. S. 1958. On the amount of carbon
dioxide in the atmosphere. Tellus 10, 243-248.
Ehhalt, D. H. 1967. Methane in the atmosphere.
J . Air Poll. Control Assoc. 17, 518-519.
Fonselius, S., Koroleff, F. & Wiirme, K. E. 1956.
Carbon dioxide variations in the atmosphere.
Tellus 8, 176-183.
Goldberg, E. D. 1971. Atmospheric dust, the sedimentary cycle and man. Comments on Earth
Sciences: Geophysics, I , 117-132.
Inman, R. E., Ingersoll, R. B. & Levy, E. A. 1971.
Soil; a natural sink for carbon monoxide. Science
172, 1229-1231.
Junge, C., Seiler, W. & Warneck, P. 1971. The
atmospheric W O and 14C0 budget. J. Ueophys.
Res. 76, 2866-2879.
Keeling, C. D. 1960. The concentration and isotopic
abundances of carbon dioxide in the atmosphere.
Tellua 12, 200-203.
Tellus XXV (1973), 2
Keeling, C. D., Ekdahl, C. A., Guenther, P. R.
Waterman, L. S. & Chin, J. F. S. 1973a. Atmospheric carbon dioxide variations at Mauna Loa
Observatory, Hawaii. Tellua 25, No. 5.
Keeling, C. D., Adams, J. A., Ekdahl, C. A. &
Guenther, P. R. 1973 b. Atmospheric carbon
dioxide variations at the South Pole. Tellus 25,
No. 5.
Lea, F. M. & Desch, C. H. 1940. The chemistry of
cement and concretes. Arnold and Go., London.
Maugh, T. H. 1972. Carbon monoxide: Natural
sources dwarf Man’s output. Science 177, 338-339.
Minerals Year Book, Bureau of Mines, United States
Department of the Interior. U.S. Government
Printing Office, Washington, D.C.
NAPCA, 1970. Summary of emissions in the United
States, 1970 edition, 49 pp. Bureau of Criteria
and Standards, National Air Pollution Control
Admin., Raleigh, N.C., USA.
OECD, 1969. Energy policy, problems and objections,
Organization for Economic Cooperation and
Development, Paris. 188 pp.
Parkin, D. W., Phillips, D. R. & Sullivan, R. A. L.
1970. Airborne dust collections over the North
Atlantic. J . Geophys. Res. 75, 1782-1793.
Parratt, L. G., 1961. Probability and experimental
errors i n science. 255 pp. Wiley, New York.
Revelle, R. 1965. Atmospheric carbon dioxide,
Appendix Y4 of Restoring the Quality of our
Environment, a Report of the Environmental
Pollution Panel, President’s Scientific Advisory
Committee, The White House, pp. 111-133.
Revelle, R. & Suess, H. E. 1957. Carbon dioxide
exchange between atmosphere and ocean, and the
question of a n increase of atmospheric CO, during
the past decades. Tellus 9 , 18-27.
Robbins, R. C., Borg, K. M. & Robinson, E.
1968. Carbon monoxide in the atmosphere. J.
A i r Poll. Control Assoc. 18, 106-110.
Rohrman, F. A., Steigerwald, B. J. & Ludwig,
J. H. 1967. Industrial emissions of carbon dioxide
in the United States: a projection. Science 156,
931-2.
SCEP, 1970. Man’s impact on the global environment assessment and recommendations for action.
Report of the Study of Critical Environmental
Problems (SCEP), 319 pp. The MIT Press,
Cambridge, Mass.
SMIC, 1971. Inadvertant climate modification. Report
198
C. D. KEELING
of the Study of Man's Impact on Climate (SMIC),
308 p. The MIT Press, Cambridge, Mass.
United Nations, 1951-1969. World energy supplies.
Statistical Papers, Series J., Published by the
Statistical Office, U.N. Dept. of Economic and
Social Affairs, New York.
United Nations, 1955. World energy requirements i n
1975 and 2000, 33 pp. International Conference
on Peaceful Uses of Atomic Energy, Published
by U.N. Depart. of Economic and Social Affairs,
New York.
Vallerga, B. A., White, R. M. BE Rostler, K. S.
1970. Changes in fundamental properties of asphalt
during service i n pavements. F i n d report by
Materials Research and Development Inc., Oakland, California, Contract FH-11-6147, Publication No. PB. 190841, available from the National
Technical Information Service, Springfield, Va
22151, VSA.
Verniani, F. 1966. The total mass of the earth's
atmosphere. J . Qeophys. Res. 71. 385-391.
ARYOHMCb Y r J I E P O A A rIPOMblLLIJIEHHOF0 I I P O M C X 0 3 K ~ E H M RI43
BCHOXAEMbIX BMAOB TOIIJIMBA
Tellus XXV (1973), 2