Note added to readers of the electronic version:
Because the printed version of this publication is exhaused and will not
be reprinted, the authors have decided to offer this publication in
electronic form .
Note that fonts and page breaks deviate slightly from the original printed
version. The original version was written in W ordPerfect for M S-DO S.
Conversion to W ordPerfect for W indows and further to the Adobe PDF
form at has unfortunately, for som e unknown reason, changed som e of
the graphic expressions. Anyway, we hope that the text is legible.
The authors
-1-
CONTENTS
P reface to S econd E dition. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2
S um m ary. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3
1.
Introduction. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5
2.
The C O 2 budget. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 9
2.1
C O 2 flow s and reservoirs. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 9
2.2
C O 2 equilibria. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 11
3.
A ir concentrations of C O 2 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 16
3.1
C ontem porary m easurem ents. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 16
3.2
N ineteenth century concentrations. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 22
4.
Tree rings as indicators of CO 2 in the atm osphere. . . . . . . . . . . . . . . . . . . . . . . . . . . . 26
5.
C O 2 in glacier ice. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 28
5.1
C hanges in original gas com position.. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 29
5.1.1
Liquid in ice at low tem peratures. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 29
5.1.2
N itrogen/oxygen/argon ratios. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 33
5.1.3
D epletion of C O 2 in surface snow .. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 34
5.1.4
E ffect of carbonates. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 34
5.1.5
C hanges of air inclusions in ice. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 35
5.1.6
M elt layers. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 38
5.1.7
E ffects of drilling.. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 39
5.1.8
C onclusions. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 39
5.2
M easured C O 2 levels in glacier ice. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 40
5.3
C onclusions. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 51
6.
H y drogen and oxygen isotopes in glaciers.. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 52
7.
Lag betw een C O 2 levels and tem perature changes. . . . . . . . . . . . . . . . . . . . . . . . . . . 54
8.
G lobal tem perature records. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 55
9.
A ir tem peratures and glaciers at high latitudes. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 60
10.
C oncluding rem arks .. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 63
11.
R eferences. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 65
12.
S am m endrag (N orw egian sum m ary). . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 75
-2-
PREFACE TO SECOND EDITION
The supply of the first edition of this w ork (N orsk P olarinstitutt R apportserie Nr. 59) w as exhausted
b y th e e n d of 1991. In this second edition w e review new publications available to us sinc e
publication of the first edition. W e have also taken into account som e of the com m ents and critical
suggestions m ade by readers.
Feedback from our readers has indicated that w e addressed a controversial issue. Various aspects
of the "greenhouse" warm ing hypothesis have been criticized before, but to our know ledge, our report
is the first critical review of the C O 2 studies in glaciers.
T h e glacier CO 2 data are often interpreted as representing the com position of the pre-industria l
atm osphere, and they are im portant input param eters in clim ate change calculations. The verification
o f the reliability of the glacier data is therefore of fundam ental im portance for validation of th e
hypothesis on a global w arm ing caused by hum an activities.
O ur thanks are due to E . B oyle, G .R . D em arée, J. Frøyland, I.Th. Rosenqvist, and J. S chw ander
for criticism and useful suggestions.
O slo, A pril 1992
-3SUMMARY
The projections of m an-m ade clim ate change through burning of fossil carbon fuels (coal, gas, oil)
to CO 2 gas are based m ainly on interpretations of m easured C O 2 concentrations in the atm osphere
and in glacier ice. These m easurem ents and interpretations are subject to serious uncertainties.
D om inant factors in the E arth's surface C O 2 cycle are the ocean, in addition to m ineral equilibria.
D ue to their vast buffer capacity, they stabilize the geochem ical equilibrium of CO 2 gas betw een the
hydro-, atm o-, litho- and biosphere. Radiocarbon ( 14 C ) studies indicate that the turnover tim e of
dissolved organic carbon in the upper ocean is a few decades. This suggests that CO 2 produced by
burning the Earth's w hole fossil carbon fuel reservoir w ould be dissolved in the ocean before reaching
the double concentration of its current atm ospheric level.
The 19th century m easurem ents of C O 2 in the atm osphere w ere carried out w ith an error of up to
100% . A value of 290 ppm v (parts per m illion, by volum e) was chosen as an average for the 19th
century atm osphere, by rejecting "not representative" m easured values which differed m ore than 10%
from the "general average for the tim e". This introduced a subjective factor in the estim ates of the
pre-industrial level of C O 2 in the atm osphere.
The M auna Loa (H aw aii) observatory has been regarded an ideal site for global C O 2 m onitoring.
H ow ever, it is located near the top of an active volcano, which has, on average, one eruption every
three and a half years. There are perm anent CO 2 em issions from a rift zone situated only 4 km from
the observatory, and the largest active volcanic crater in the world is only 27 km from the
o bservatory. These special site characteristics have m ade "editing" of the results an e s ta b lis h e d
procedure, w hich m ay introduce a subjective bias in the estim ates o f th e "tru e" values. A sim ilar
procedure is used at other C O 2 -observatories. There are also problem s connected to the instrum ental
m ethods for m easurem ents of atm ospheric C O 2 .
The C O 2 concentrations in air bubbles trapped in glacier ice are often interpreted a s p re vious
atm ospheric concentrations, a s s u m ing that the com position of the air in the bubbles rem ained
unchanged. This w as based on another assum ption: liquid does not exist in ice below a m ean annual
te m perature of about -24EC , and no changes due to diffusion m ay be expected. H ow ever, it w a s
recently found that liquid can be present in Antarctic ice at tem peratures as low as -73EC . N um erous
studies indicate that, due to various chem ical and physical processes, the C O 2 content in ice can be
largely enriched or depleted in com parison with the original atm ospheric level. In the air inclusions
from pre-industrial ice the CO 2 concentrations were found to range betw een 135 and 500 ppm v.
M ethods using dry extraction of C O 2 from crushed ice release only about half of this gas present
in the ice. C O 2 in air inclusions can penetrate the ice by diffusion or dissolution into the liquid present
at the ice grain boundaries, at a rate different from rates of other gases in the air. A problem for the
determ ination of C O 2 levels in gas inclusions is the form ation of solid CO 2 clathrates (hydrates). Other
gases in air also form clathrates, but at different tem peratures and pressures. This leads to im portant
-4changes in the com position of the inclusion air at different core depths and indicates that glacier ice
cannot be regarded as a steady state m atrix suitable for observation of long-term atm ospheric trends.
Thus, the results of C O 2 determ inations in air inclusions in ice cannot be accepted as representing
the original atm ospheric com position.
A nother difficulty in this respect is a speculative assum ption that air is 90 to 2800 years younger
than the ice in w hich it is trapped. W ithout this assum ption the CO 2 concentration in air recovered
from 19th century ice is the sam e as now. A tm ospheric N 2 / O 2 / A r ra tio s in trapped air are not
preserved. Instead the ratio s agree w ith those from aqueous solubility data.
85
K r and
39
Ar
m easurem ents indicate that 36 to 100% of gas from the ice cores are contam inated by am bient air.
P aleo-tem perature calculations based on light stable isotope ratios (D /H and
18
O / 16 O ) in ice have
large uncertainties. After the discovery of liquids between ice crystals in the deeply frozen A ntarctic
ice, considerable isotopic exchange and frac tio n a tion should be expected in the ice, m aking
calculated paleo-tem peratures m eaningless if phase changes occurred in the presence of a m obile
fluid phase.
A ttem pts have been m ade to calculate the paleoatm ospheric C O 2 content from
13
C/ 12 C carbon
stable-isotope ratios in tree rings. It is concluded here that the C O 2 conte n t in th e a tm osphere
calculated from such carbon isotope analyses cannot be considered a valid tool in paleoclim atology,
nor can it be used as evidence of changing atm ospheric C O 2 levels.
The so-called increasing "greenhouse effect" signal, i.e. anthropogenic increase of the global air
tem perature, which was claim ed to have been observed during the last decades, is not confirm ed
by recent studies of long tem perature series.
In the A rctic, according to m odel calculations, this w arm ing should be m ost pronounced. How ever,
cooling rather than w arm ing has been recorded in this region during the last tw o decades. Glacier
balance studies provide evidence fo r a re c e n t decrease in glacier retreat, and for an increased
accum ulation over the polar ice caps, corresponding to a sea level low ering of about 1 m m per year.
-51. INTRODUCTION
O nly about half the incom ing solar energy is absorbed by the E arth's surface. The rest is scattered
back and to som e extent absorbed by the atm osphere (clouds included), or reflected by the ground.
The E arth itself radiates at wavelengths m uch lon g e r th a n th ose of solar radiation. This E arth
radiation, unlike the solar radiation, is strongly absorbed in the atm osphere. The absorbtion is m ainly
caused by wate r v a p o r and clouds, but also by som e trace gases. Only a very sm all part of the
radiation em itted by the ground escapes directly to space. In this w ay the atm osphere is heated, and
returns radiative energy to the Earth's surface, w here it is again absorbed and re-radiated. Thus a
rem arkable exchange of therm al energy takes place betw een the ground and the low er atm osphere.
T h e s e processes, som ew hat m isleadingly called the "greenhouse" effect, are responsible for th e
relatively high m ean surface tem perature on the E arth, of about 14EC . If the E arth had no atm osphere
the corresponding tem perature would be about -18EC .
The term "greenhouse effect" is now often used to nam e a predicted increase in the tem perature
of the low er atm o s p h e re a s a consequence of m an's release of C O 2 and other trace gases to the
atm osphere. This predicted additional effect w ill in the follow ing be referred to as an increasing
"greenhouse effect" or "greenhouse" warming.
In the Earth's atm osphere dry air consists of nearly 78% (by volum e) nitrogen (N 2 ), about 21%
oxygen (O 2 ) and about 1% argon (Ar). In hum id air the water vapor content varies from about 3% in
the tropics to a sm all fraction of this quantity in the polar regions. C arbon dioxide (C O 2 ) is just a trace
com ponent, w ith a concentration of about 0.035% (= 350 ppm ), but it plays an im portant part in plant
and anim al life processes. The concentration of C O 2 varies with tim e and place. It has been found,
for exam ple, that the concentration m ay double during one single day over a w heat field (Fergusson,
1985).
D uring the 1970s and the first half of the 1980s several clim atic m odel com putations predicted that
for a hypothetical doubling of the average atm ospheric C O 2 concentration during the next 60 years,
the average global tem perature will increase by 1 to 5EC (see e.g. review by B raathen et al., 1989),
that the polar regions w ill w arm m ore than the low er latitudes, up to 8 to 1 0 EC (S chneider, 1975;
M anabe and W etherald, 1980), and that the seasonal variations will be greatest in the north polar
regions (R am anathan et al., 1979). These m o d els also predicted considerable changes in the
geographical distribution of precipitation. A t the end of the 1980s m ore sophisticated m odels revised
th e e a rlier predictions substantially, decreasing the net im pact on the clim ate and changing its
geographical distribution. A recent study estim ated a 1.2EC increase in the surface and tropospheric
tem perature due to d oubling the atm ospheric C O 2 (Lorius et al., 1990), assum ing no feedback
processes.
T o the present natural global atm ospheric flow of C O 2 , m an's burning of fossil carbon m ay a d d
som ew here betw een 0.1 and 3.6% , a ccording to different estim ates. C O 2 is one of about 40 trace
"greenhouse gases" present in the atm osphere (Ram anathan et al., 1985). W ater vapor contribute
the m ost to the total "greenhou s e e ffect" of the atm osphere of about 150 W /m 2 (R aval and
R am anathan, 1989). A cco rd in g to K ondratyev (1988) H 2 O contributes about 62% , CO 2 21.7% , O 3
-67.2% , N 2 O 4.2% , C H 4 2.4% , and other gases 2.4% to the m ean "greenhouse e ffe c t" o f the
atm osphere.
A d o u b ling of C O 2 w ould increase its "greenhouse" contribution to about ~4 W /m 2 (R aval a n d
R am anathan, 1989 a). Landsberg (1974) estim ated that only 3% decrea s e in atm ospheric w ater
vapor, and a 1% increase in cloudiness can com pensate the w arm ing from an anticipated C O 2
doubling (other conditions held constant). A s a w hole, the influence of clouds on atm ospheric
tem perature is still an unsolved problem (e.g. S chlesinger and M itchell, 1987). The predictions of C O 2
doubling are based on an assum ption that all past hum an activities have contributed about 21% of
the current atm osp h eric C O 2 , the level of which is supposed to be 25% higher than in the preindustrial period (IP C C , 1990). This assum ption is based on glacier studies. A s w ill be seen later on
these studies do not provide a reliable basis for such an estim ate.
The level of atm ospheric C O 2 depends on constantly changing therm odynam ic equilibria betw een
its sources and sinks. O ceanic flow s of this gas in and out of the global atm osphere are im portant
for the CO 2 budget. E ven very sm all natural fluctuations of these oceanic flow s can m ask the m anm ade C O 2 inputs into the global atm osphere.
S everal studies have sugge s te d that radiative heating of 4 W /m 2 caused by the doubling of
atm ospheric CO 2 w ould le a d to a global w arm ing of 3.5 to 5EC (H ansen et al., 1984; W ilson and
M itchell, 1987; W ashington and M eehl, 1984; W etherald and M anabe, 1988). The total present global
m ean w arm ing due to all trace "greenhouse" gases added by m an of about 2 W /m 2 , is below the
estim ated natural v a ria tion of about ± 5 to 10 W /m 2 in the global net radiation (R aval and
R am anathan, 1989). The positive (w arm ing) forcing by clouds is abo u t 30 W /m 2 (R aval and
R am anathan, 1989). This "greenhouse effect" of clouds is approxim ately fifteen tim es larger than that
resulting from a hypothetical doubling of CO 2 (increase from -2 to 4 W /m 2 ). A new estim ate of
R a m a n a than et al. (1989 b) suggests that "the CO 2 concentration in the atm osphere has to be
increased m ore than tw o orders of m agnitude to produce a "greenhouse effect" com parable to that
of clouds". The negative cloud forcing (due to a high albedo at the upper cloud surface) is about -50
W /m 2 . From such fig u res one gets a net cloud forcing of about -20 W /m 2 (30 m inus 50), i.e. m uch
higher than the claim ed m an-m ade positive forcing of CO 2 doubling.
R am anathan et al. (1989 b) dem onstrated that the clouds have a large net cooling effect o n the
E arth, w hich will offset the possible increasing "greenhouse effect" w arm ing. This is because an
increase in the global tem perature will increase the am ount of clouds in the troposphere, introducing
a s trong negative radiative feedback. It is claim ed that the total past anthropogenic "greenh o u s e "
forcing (due to CO 2 and other trace gases) betw een 1850 and 1985 should cause a global surface
w arm ing of 0.8 to 2 .4 EC
(R am anathan et al., 1989 b). H ow ever, no such warm ing has been
observed, which m ay indicate that the estim ate of the "greenhouse gases" increase is incorrect or
that the negative cloud forcing of about -20 W /m 2 (or som e other negative forcings) is sufficiently
large to stabilize the increasing "greenhouse effect" warm ing. This latter supposition was confirm ed
by Slingo (1989) w ho found that the radiative forcing by doubled CO 2 concentrations can be balanced
by m odest increases in the am ount of low clouds.
-7W igley et al. (1989) revived the idea posed by M itchell (1975) that S O 2 -derived cooling m ay offset
considerably the "greenhouse" w arm ing. S O 2 originates from dim e thylsulfide from the oceans
(Charlson et al., 1987), volcanic em issions, and from m an-m ade sources. The cooling effect is partly
due to the absorption of incom ing solar radiation by sulfuric acid in the stratosphere and partly due
to an increase of cloud condensation nuclei in the atm osphere. The latter effect is due to H 2 S O 4 and
sea-salt aerosols (Latham and S m ith, 1990) and serves to "brighten" clouds (increase their albedo),
thereby reflecting part of the incom ing solar radiation back into space. S atellite data now confirm that
the stratocum ulus clouds (one of the m ost com m on cloud types on E arth, and the variety m ost likely
to be affected by an increasing num ber of condensation nuclei) are indeed considerably brighter in
the lee of regions of m ajor anthropogenerated S O 2 em issions (Cess, 1989).
W igley (1989) argued that m an-m ade S O 2 is sufficiently large to offset signific a n tly th e global
w arm ing that m ight result from the "greenhouse effect", and to cool the Northern Hem isphere relative
to the S outhern H em isphere, because m ost of the m an-m ade S O 2 em issions occur in the Northern
H em isphere. W igley (19 8 9 ) s upposed that the m an-m ade S O 2-derived (negative) forcing m ight
explain the inconsistency betw een G eneral C irculation M odel (G C M ) predictions of current warm ing
and observations. To substantiate this W igley cited two sulfate records from ice cores collected in
southern Greenland show ing up to three-fold increase during the twentieth century. The tem perature
in this region is high enough to allow sum m er m elting, which m a y le a d to changes in chem ical
com position of snow and ice (Jaw orow ski et al., 1992). H ow ever, seven other studies in the A rctic
and five studies in Antarctica dem onstrated no increase of sulfate or acidity in snow and ice during
the past century. T h e s e studies indicate that there w ere covariations of the sulfate content in
precipitation from the Southern and Northern Hem isphere in relation to m ajor volcanic events, and
that during the last decades the concentration of sulfate in precipitation in the A rctic was sim ilar to
that in Antarctica (Jaw orow ski, 1989). Thus W igley's hypothesis is not substantiated.
A m ain feature of the predictions of alm ost all clim ate m odels is a relatively large warm ing at high
latitudes. Therefore polar regions m ay be assum ed to be the m ost prom ising ones for detection of
any current in c re a s in g "greenhouse effect" warm ing. Tem perature and to som e degree glacier
records can be used to check these m odel predictions. A discussion of com puter m odelling with the
help of GC M is, however, beyond the scope of this report. Cess et al. (1989) com pared 14 different
m odels of this kind and showed that from the sam e input data the m odels produced results which
varied greatly, i.e., both cooling and w arm ing of the clim ate. A lso in C ess et al. (1991) the net effect
of snow feedback produced by 17 m odels differed m arkedly, ranging from cooling to w arm ing. In this
latter paper it w as dem onstrated that the conventional explanation that a w arm er E arth will have less
snow cover, resulting in a darker planet, absorbing m ore solar radiation, is overly sim plistic.
If the present GC M m odeling is fit for predicting future clim ate, it should also reproduce the past
clim ates. How ever, according to S treet-Perrott's (1991) analysis of such m odeling, this point is still
not reached. The ocean is a dom inating factor in determ ining the Earth's clim ate. How ever, in these
m odels the enorm ous heat capacity of the ocean water m ass and the oceanic currents are still not
taken into account in a satisfactory w ay (M oene, 1991).
-8A s will be seen from the discussion below , the hypothesis of an im m inent clim atic change is based
on data subject to serious uncertainties and inconsistencies. These uncertainties should be factored
into policy decisions in view of the staggering costs of im plem entation of "anti-greenhouse" decisions
on a global scale. In the U nited S tates alone these costs m ay reach 3.6 trillion US dollars (Passel,
1989). Im plem entation of the C O 2 tax of 500 U S dollars per m etric ton of carbon w ould increase the
price o f crude oil about 3.7 tim es (to m ore than 60 dollars per barrel) and of utility coal about 8.3
tim es (to m ore than 276 dollars per short ton) (A nonym ous, 1992). This m ight have serious negative
social consequences both for developed and third world countries reaching beyond the 21st century.
S uch consequences should be w eighed against the very uncertain predictions of environm ental
effects of an increase in atm ospheric C O 2 .
In m ost cases scientists are aw are of the w eak points of their basic assum ptions and sim plifications
needed to interpret the results of m easurem ents or to create m odels. How ever, these uncertainties
are m ostly ignored or banished to a subordinate clause w hen the results are presented by politicians
or m ass m edia . In th e process of form ing environm ental policy the prelim inary hypotheses are
transform ed into "reliable facts" w hen presented to the public. The m agnitude of "norm al" natura l
reservoirs, fluxes, and variations are not presented and not com pared to claim e d "a b n o rm a l"
anthropogenic contributions. A m ore balanced view is certainly needed.
The m ost im portant basis of the hypothesis of m an-m ade clim atic w arm ing due to burning of fossil
carbon fuels are the m easurem ents of C O 2 in air and in glacier ice, hydrogen and oxygen isotopes
in glacier ice, carbon isotopes in tree rings, and 100-150 years long atm ospheric tem perature
records. In this paper w e critically review these m easurem ents and their interpretations, in order to
te s t th e n o w adays widely accepted postulate that "the change in atm ospheric C O 2 is not just a
fluctuation of nature, but is predom inantly the consequence of the activities of m ankind - chiefly the
burning of fossil fuels such as coal, gas and oil" (R ow land and Isaksen, 1988). W e also com pare the
q u a n tities of anthropogenic contributions w ith know n fluxes of natural reservoirs, and discus s a ir
tem perature and glacier balance records, w hich should reveal signals of an increasing "greenhouse
effect".
-92. THE CO 2 BUDGET
2.1 CO 2 FLOWS AND RESERVOIRS
M os t o f th e recent attem pts at predicting an increasing "greenhouse effect" have assum ed a
doubling of the present atm ospheric CO 2 level, w hich is about 350 ppm , due to fossil fu e l burning
during the next 30 - 60 y e a rs . T o estim ate the reliability of this basic assum ption, the size of the
various sources and sinks of atm ospheric CO 2 should be taken into account.
It is usually accepted that fossil carbon fuel burning contributes about 5 GT C (carbon equivalents
in Gigatonnes = 10 15 g) per year, i.e. about 3.6% of the global natural CO 2 flux of 169 GT C per year
(Iansiti and N iehaus, 1989). H ow ever, Fergusson (1985) estim ated the m an-m ade contribution to be
only about 0.1% . Fluctuations in the natural flux of C O 2 (Table 1) are generally higher than this m anm ade contribution.
The yearly C O 2 exchange betw een atm osphere and ocean, and betw een atm osphere and
biosphere am ounts to about 23% of the estim ated pre-industrial atm ospheric C O 2 c ontent. In
contrast, the present annual m an-m ade input due to fossil fuels am ounts to 0.8% of atm ospheric C O 2
(Oeschger and S iegenthaler, 1988). This m eans that a sm all fluctuation in the natural exchange rate
w ould m ask the fossil fuel contribution.
If all fossil fuel resources (approxim ately 7000 G T C) w ere bu rn e d im m e d iately, the current
atm ospheric content of CO 2 (about 700 G T C ) w ould suddenly increase by a factor of 11, and would
then dissolve in th e o c e a n and be precipitated as carbonates in the bottom deposits (W alker and
D rever, 1988). S uch an assum ption is, of course, rather unrealistic, as the CO 2 residence tim e in the
atm osphere is only about 5 years (B olin and K eeling, 1963; S tum m and M organ, 1970; B roecker and
P eng, 1974; Sundquist, 1985), and as the existing resources are supposed to be exhausted in
approxim ately 300 years. A ccording to G orshkov (1982) the ocean is so powerful a sink for CO 2 that
after burning all existing fossil fuels, the C O 2 contribution to the atm osphere can not increase by m ore
than 35-40% .
C O 2 consum ed annually by the photosynthesis of land plants give fluxes in the range 10 - 70 tim es
higher than produced by m an; photosynthesis by m arine plants give fluxes in the range 50 to 250
tim es higher (R e v e lle and S uess, 1957). S uch large natural sources and sinks w ould m ost likely
m as k the effect of fossil carbon fuel burning, w hich accounts for just a tiny fraction of the global
atm ospheric flux of this gas. Just sm all fluctuations in the dissolved m arine carbon reservoir (38,000
G T C; B olin, 1989), the sedim entary carb o n ate carbon reservoir (60,000,000 G T C; W alker and
D rever, 1988), the vast carbon reservoir of the E arth's interior (constantly being tapped by volcanic
activity), or short-term fluctuations such as due to the E l N iño - S outhern O scilla tion (E N S O )
phenom enon occurring every few years (about 2 GT C p e r event; G audry et al., 1987), w ould
probably obscure the fossil fuel contribution of about 5 GT C per year.
A s recently found in a m odelling com putation, for sim ulations run w ith tw ice the present-day C O 2
levels, strong negative feedbacks appear due to carbon storage in terrestrial biota and soils (P rentice
-10Table 1. C urrent reservoirs of carbon at the E arth's surface and annual fluxes of CO 2 (expressed as
carbon equivalents in Gigatonnes = 10 15 g) into the atm osphere.
GT of carbon
References
RESERVOIRS
S edim ents 1
60,000,000
W alker and Drever, 1988
M arine dissolved
organics
1,000
B olin, 1989
M arine dissolved
inorganics
38,000
B olin, 1989
Fossil fuels (exploitable)
A tm osphere
S oil
Terrestrial biom ass
M arine biom ass 2
7,200
727
W alker and Drever, 1988
R ose et al., 1983
1,300
P rentice et al., 1990
834
P rentice et al., 1990
42
ANNUAL FLUXES
Iansiti et al., 1989
N A TU R A L
O cean
Land
TOTA L
106
63
169
M A N -M A D E
Fossil fuels and
land use
1
2
6
N e a r s u rfa c e
C a lc u la te d fro m a to ta l m a s s o f m a rin e b io ta o f 1 0 1 8 g (N ria g u , 1 9 7 8 ) a n d fro m c a rb o n c o n c e n tra tio n in th e s e
b io ta a s s u m e d to b e 0 .0 4 g ra m c a rb o n p e r g ra m w e t w e ig h t (G E S A M P , 1 9 8 3 ).
-11and Fung, 1990). These negative feedbacks rem ove 235 G T of carbon from the atm osphere. From
a carbon cycle m odel it w as estim ated that the m arine reservoir contains, in the form of dissolved
carbonic acid, at least 37% of the fossil fuel CO 2 produced in the past (B roecker et al., 1979). If the
s a m e fraction w ere assum ed for the future releases, it w ould correspond to about 87 G T C . T h e
terrestrial biota, soils, and m arine inorganic carbon sinks would then be large enough to rem ove 322
G T of carbon from the atm osphere. This is m ore than enough to rem ove the anthropogenic am ount
introduced into the atm osphere during the next 5 0 years, at the current consum ption rate of fossil
fuels.
The enorm ous sink of m arine biota and of m arine dissolved organics is not accounted for in the
current m odels of the global carbon cycle (e.g. O eschger et al., 1975; Siegenthaler and Oeschger,
1987). Therefore such carbon cycle m odels should be regarded as incom plete.
The C O 2 content in the atm osphere is dependent on the natural clim atic changes through
m odulation of the ocean upw elling in low latitudes. An intensified upw ellin g o f d e e p w ater rich in
calcium , phosphorous, potassium , and other biogenic com ponents rises the rate of form ation of
calcium carbonate a n d re d u ces the transport of C O 2 from the ocean to the atm osphere. Thus the
atm ospheric C O 2 content m ay be controlled by the clim ate (K ondratyev, 1988). C ooling and warm ing
of ocean water w ill also influence the CO 2 flux. A ccording to Takahashi (1961) heating of sea w ater
by 1EC w ill increase the partial pressure of atm ospheric CO 2 by 12.5 ppm during upw elling of deep
w ater. For exam ple the 12°C w arm ing of the B enguela Current should increase the CO 2 partial
pressure by 150 ppm . In reality som e of this CO 2 is precipitating as C aC O 3 .
The rise of biogenic C a 2 + a n d expulsion of CO 2 from upw elling of relatively warm water will
increase the therm odynam ic activities of C a 2+ and C O 2 , and facilitate precipitation of calcium
carbonate. This precipitation is not com plete, so the atm ospheric C O 2 will rise (see next section), as
observed by Takahashi (1961).
2.2 CO 2 EQUILIBRIA
Tab le 1 s h o w s that the ocean is the dom inant factor in the CO 2 cycle of the Earth's surface.
Therefore, w e discuss briefly the geochem ic a l e q u ilibria that govern the relationship betw een
atm ospheric and oceanic C O 2 . The inorganic part of the CO 2 cycle in the atm osphere / hydrosphere
/ lithosphere system is buffered w ith respect to pH by carbonate equilibria (see below ). The salinity
of the ocean is held constant by erosional and depositional processes and by m ineral equilibria. All
parts of the system , including C O 2 in the atm osphere, are therefore therm odynam ically buffered and
w ill tend to establish a chem ical equilibrium at a given tem perature. For a change in tem perature,
new chem ical equilib ria apply, and a new value for the partial pressure of C O 2 w ill be established
(Rubey, 1951; M acIntyre, 1970). It appears from the w ork by S m ith and Jones (1985) that w ind is
capable of pum ping C O 2 into the w ater, thereby establishing an increased partial pressure of CO 2 in
the surface water. This is due to pressurization of bubbles generated by breaking waves.
-12The oceans to a depth of about 4 km are supersatura te d w ith re s pect to calcium carbonate
(B roecker et al., 1979). This w ould facilitate precipitation of calcium carbonate for any additional input
of C O 2 through the atm osphere/ocean interface, and thereby oceans w ill consum e any excess C O 2
in the atm osphere. In the global carbon cycle m odels this is not accounted for (e.g. Oeschger and
S iegenthaler, 1975).
C O 2 in the atm osphere is in chem ical equilibrium with carbonic acid dissolved in the hydrosphere
(sea, lakes, rivers, etc.) (e.g. Ohm oto and R ye, 1979; Gonfiantini, 1981; M ozeto et al., 1984; Etcheto
and M erlivat, 1988; H orita, 1989), w hich again is in chem ical equilibrium with calcium carbonate in
w ater, in lim e shells of aquatic organism s, and in lim estone (see e.g. R ubey, 1951; G arre ls a n d
Thom pson, 1962; G arrels and C hrist, 1965; P ytkow icz, 1967; S tum m and M organ, 1970; P lum m er
et al., 1978; P lum m er and B usenberg, 1982; Talsm an et al., 1990). S everal chem ic a l re a c tions
stabilizing this atm osphere/ hydrosphere equilibrium have been w orking at least during the last 600
m illion years (Holland, 1984).
The inorganic dissolved carbon in the ocean (aq) is exchanged betw een atm ospheric C O 2 (g) and
solid calcium carbonate (s) by the follow ing chem ical reactions:
P artial reactions:
C O 2 (g) º CO 2 (aq)
C O 2 (aq) + H 2 O º H 2 C O 3 (aq)
H 2 C O 3 (aq) º H + + H C O 3 - (aq)
H C O 3 - (aq) º H + + C O 3 2- (aq)
C O 3 2- (aq) + C a 2+ (aq) º CaC O 3 (s)
N et reaction:
C O 2 (g) + H 2 O + C a 2+ (aq) º CaC O 3 (s) + 2 H +
In the current global carbon cycle m odels the last partial chem ical reaction is neglected. A ny
additional C O 2 entering the ocean from the atm osphere w ill have the potential of precipitating calcium
carbonate according to the P rinciple of le C hâtelier (average ocean depth 3.8 km ; average calcite
saturation depth 4 km ). This is w hy the vast sedim entary C O 2 reservoir has been accum ulated on the
E arth's surfa c e throughout its history. The ultim ate source is C O 2 constantly degassed from the
E arth's interior. The atm osphere represents just a sm all short-term C O 2 reservoir in this pro c e s s .
W ithout oceans and sedim ents the partial pressure of atm ospheric C O 2 on E arth would be several
tens of atm ospheres, like on V enus.
The C O 2 of the Earth's interior is stored in the form of solid carbonates. A considerable am ount of
C O 2 is expected to be stored in the m antle in the form of m agnesite M gC O 3 (Katsura and Ito, 1990).
M antle C O 2 m oves in the form of carbonated m agm a. Such a m elt m ay dissolve up to 8 w eight % C O 2
(E ggler and M ysen, 1976) at 45 kbar (1 bar = 10 5 P a), at about 125 km depth. The CO 2 transported
upward would be released from the carbonated m agm a as volcanic gases and fluid inclusions when
the m agm a reaches a depth of less than 70 to 90 km , where the carbonates becom e unstable (W yllie
-13and H uang, 1976). W hen the m agm a erupts at the surface it can only hold 0.01 to 0.001 weight %
C O 2 dissolved (H arris and A nderson, 1983), im plying that up to 99.99% of the CO 2 in m antle-derived
m agm as will be degassed to the atm osphere. In this w ay C O 2 has been successively extracted from
the m antle by volcanism throughout the Earth's history, and is still being degassed today.
O nly in the upper 75 m layer of the ocean there is enough calcium to bind m ore than 3000 G T C .
This is a large potential sink for the annual anthropogenic 6 GT C . One should note that this reservoir
is continuously replenished from other parts of the ocean and from the lithosphere. Adding CO 2 to
the w ater w ill increase its buffer capacity and enhance its ability to m oderate the atm ospheric C O 2
level (B utler, 1982). The m axim um buffer capacity is achieved at 2.5 to 6 tim es the present partial
pressure of atm ospheric C O 2 , depending on tem perature and alkalinity. A ccording to M aier-R eim er
and H asselm ann (1987) the borate system increases the ocean storage capacity for CO 2 by m ore
than 20% (com pared w ith a borate-free ocean).
The vast calcium carbonate buffer is not the o n ly buffer acting in the atm osphere/hydrosphere/
lithosphere system . The Earth has a set of inorganic geoc hem ical equilibrium system s w orking
sim ultaneously. The equilibrium system anorthite CaA l 2S i 2O 8 - kaolinite A l 2S i 2O 5(O H ) 4 at the pH
o f s e a w ater has a buffer capacity one thousand tim es higher than that of a 0.001 M ca rb o n a te
solution (S tu m m a n d M organ, 1970). In addition, there are clay m ineral buffers, and the
calcium silicate + C O 2 <--> calcium carbonate + silica buffer (e.g. M acIntyre, 1970; Krauskopf, 1979).
These serve as "security nets" under the m ain buffer controller: the C O 2 (gas) - H C O 3 - (aqueous) C aC O 3 (solid) buffer system (S egalstad and Jaworow ski, 1991). This suite of m ineral buffers attain,
in principle, an infinite buffer capacity (S tum m and M organ, 1970). A s K ram er (1965) expresses it:
"A few sim ple calculations will show that only very large additions or subtractions of carbon dioxide
(from the lithosphere) could overwhelm these equilibria".
C urrent global carbon cycle m odels are m ade to fit the assum ption that the level of C O 2 in the preindustrial atm osphere w as about 280 ppm v, and that due to a "buffer factor" the ocean can rem ove
only about 10% of the atm ospheric CO 2 added by m an's activities (e.g. Siegenthaler and Oeschger,
1987). This "buffer factor" w as calculated by assum ing that the chem ical interaction of atm ospheric
CO 2 is lim ited only to the reactions C O 2 <-> H C O 3 - <-> C O 3 2- in the 75 m thick upper ocean layer, and
by neglecting other seawater species and buffer system s, and by assum ing that C O 2 rem oval will be
lim ited to this upper layer.
In the carbon cycle m odels it is assum ed that the ventilation of the deep ocean w ater is very slow
and that the deep circulation is effective on tim e scales of 100 - 1000 years (IP C C , 1990). H ow ever,
recent m easurem en ts o f
14
C and
39
A r (w ith the half-lives of 5730 and less than 270 years,
respectiv e ly ) in the oceanic w ater dem onstrated that the turnover tim e of the Atlantic deep-w ater
below 4250 m is about 30 years, and that the inflow rate of the north east A tlantic deep water is about
10 14 m 3 per year (S chlitzer et al., 1985). O nly this inflow transfers about 10 GT total CO 2 per year.
The assum ption on the low pre-industrial C O 2 and on the "buffer factor" produce m odeling results
w hich are incom patible w ith atm ospheric m easurem ents and actual anthropogenic em issions of C O 2 .
-14The m odeling results indicate that about 50% of the m an-m ade C O 2 rem ained in the atm osphere
since the m iddle of 18th century, and that only 10% was rem oved by the ocean. Thus the c a rb o n
cycle m odels are unable to account for the "lacking" 40% of m an-m ade C O 2 . A ccording to Broecker
et al. (1979) 25 to 100 % of fossil fuel C O 2 released in the atm osphere is not accounted for by the
current carbon cycle m odels. This dem onstrates a serious w eakness of the m odels and indicates that
the assum ptions on which the m odels are based (i.e. low pre-industrial atm ospheric level, and low
capability of the ocean to rem ove m an-m ade C O 2 from the atm osphere) m ay be wrong.
The solubility of C O 2 in w ater can be expressed by H enry's Law C onstant, K H (D rum m ond, 1981;
H enley et al., 1984; B arta and B radley, 1985). This constant num erically expresses the decreasing
solubility of CO 2 in water for increasing tem perature (up to about 200 EC ) and the equilibrium
partitioning of C O 2 betw een atm osphere and w ater (1:32 at 25°C ; Krauskopf, 1979). A s tem perature
rises, less CO 2 m ay rem ain in the water. There fo re a w a rm er ocean will degas CO 2 to the
atm osphere. This "therm ally driven solubility pum p" accounts for approxim ately 70% o f th e to tal
ocean degassing, while the rem aining 30% is accounted for by the organic nutrient "biologic pum p",
according to Volk and Liu (1988). The therm al solubility pum p and the biological pum p increase the
atm ospheric level of C O 2 w hen the sea and air tem peratures rise due to a natural clim atic fluctuation.
The powerful negative feedback m echanism s, e.g. due to an in c re a s e in cloudiness, low er the
tem perature and keep the naturally fluctuating global air tem perature in balance (see Chapter 1).
It has been estim ated that approxim ately 4000 GT of CO 2 wa s flu x e d from the ocean via the
atm osphere to the continental biosphere during transition fro m a g lacial to an interglacial stage
(F a u re , 1990). Other natural carbon fluxes (w eathering, volcanism , carbonate sedim entation an d
dissolution, degassing by m etam orphism , etc.) are not included in this figure. The current natural C O 2
flux betw een the atm osphere and ocean is about 100 GT C per year, and between the atm osphere
and the terrestrial biosphere about 63 G T C per year. A nthropogenic releases of C O 2 of about 6 G T
C are sm all in com parison to these natural fluxes.
The biological activity in the ocean, together w ith gravitational forces, act as a pum p for keeping
C O 2 out of the atm osphere (e.g. S akshaug, 1990). G orshkov (1982) estim ated that due to fixation of
m an-m a de carbon by oceanic phytoplankton the m axim um atm ospheric C O 2 concentration could
never exceed about 390 ppm v. If these biologic activities w ere rem oved, the partial pressure of C O 2
w ould be increased by a fa ctor of 5 (Eriksson, 1963). Hence, variations in m arine biologic activity
alone could account for larger variations in the am ount of atm ospheric C O 2 th a n anthropogenic
contributions from burning fossil carbon fuels at the current rate.
T h e equilibrium partition coefficient for the C O 2 distribution betw een atm osphere and o c e a n is
approxim ately 1:50 (R evelle and Suess, 1957; S kirrow, 1975). This partition coefficient can be used
to set an upper lim it for how m uch the C O 2 concentration will rise in the atm osphere if all available
fossil carbon fuel (coal, petroleum , gas) w ere burned. In order to perm anently double the current level
of C O 2 in the atm osphere under chem ical equilibrium conditions, the atm osphere m ust be supplied
w ith approxim ately 51 tim es the present am ount of C O 2 if equilibrium should be attained. In order to
keep the partition coefficient (air:sea = 1:50) constant at the double am ount of C O 2 in the air, the new
-15ratio would have to be 2:100. In other w ords an increase of one unit in air leads to an increase of 50
units in the sea; a total of 51 units has to be supplied.
A s can be see n in T a b le 1 , a ll available fossil fuel carbon am ounts to 11 tim es the am ount of
carbon in the atm ospheric C O 2 . Therefore, m ankind does not have enough fossil fuel to double the
current level of atm ospheric C O 2 under equilibrium conditions, all other factors held constant. If the
total fossil fuel reservoir of 7200 GT C w ere burned during the next 300 ye a rs , o nly the dissolved
organics (carbon pool of about 1000 G T C ) w ould consum e all m an-m ade C O 2 , due to the fact that
this period covers 6 to 15 turnovers of the upper-ocean pool of dissolved organic carbon, based on
radiocarbon ( 14 C ) studies (Toggw eiler, 1990; Druffel and W illiam s, 1990; Rau, 1991). How ever, the
vast oceanic dissolved inorganic carbon reservoir of 38,000 G T C indicates that the sea is a m uch
m ore pow erful sink for the atm ospheric C O 2 . H ence, it is unlike ly th a t p e rm anent doubling of the
am ount of atm ospheric C O 2 is attainable by hum an activities. S im ilar doubts have been expressed
by A belson (1990), and earlier by Gorshkov (1982) (see also K ondratyev 1988).
The transfer of C O 2 from the atm osphere to the hydrosphere is facilitated by leaching with cloud
droplets. The average diam eter of a cloud droplet is 20 µm (B attam , 1979), and its m ass is 4.2 x 10 -9
g. The m ass of ~5 x 10 20 g of global annual precipitation corresponds to about 10 29 droplets. The total
surface of these droplets is about 10 14 km 2 , i.e. ~275,000 tim es greater than the surface of the ocean.
The high solubility of C O 2 in w ater at low tem perature and enorm ous surface of droplets exposed to
atm ospheric C O 2 m ust contribute to a flow of inorganic carbon from the atm osphere to the
hydrosphere. The total am ount of C O 2 , w hich m ight be dissolved in the global annual precipitation,
m ay reach ~0.3 GT C O 2 .
A nother sink for CO 2 is w eathering of silicate m inerals. It has been dem onstrated by W alker et al.
(1981) that the partial pressure of C O 2 in the atm osphere is buffered, over geologic tim e scales, by
a negative feedback m echanism in which th e ra te of w eathering of silicate m inerals (follow ed by
deposition of carbonate m inerals) depends on surface tem perature. The higher the tem perature the
greater the rem oval of C O 2 from the atm osphere by deposition of carbonates. This negative feedback
is an im portant factor in the long-term stabilization of the global surface tem perature. S chw artzm an
and V olk (1989) have suggested that biota am plify this negative feedback.
H olland (1984) has tested whether the atm ospheric C O 2 content has b e en extrem ely low or
extrem ely high during the E arth's history. A t low partial pressure of C O 2 gyp s u m C a S O 4 ! 2H 2 O
w ould be form ed at the expense of calcium carbonate. This has not been found in m arine sedim ents.
A t high partial pressure of C O 2 dolom ite CaM g(C O 3 ) 2 would be expected to precipitate directly from
the ocean instead of calcium carbonate (aragonite or calcite). This has not been found either. The
conclusion is that the atm osphere/hydrosphere equilibria operating at the Earth's surface today, have
been w orking through hundreds of m illions of years, even during periods of intense volcanic release
of CO 2 . A correlation betw een increased volcanic production and increasing m arine carbonate
sedim entation has been found by B udyko et al. (1987).
-163. AIR CONCENTRATIONS OF CO 2
3.1 CONTEMPORARY MEASUREMENTS
A n im portant com ponent o f th e "g reenhouse" warm ing hypothesis is the analysis of C O 2
conce n trations in the atm osphere. The first large scale m easurem ents were started in 1955 in
S candinavia (B ischof, 1960). S ince 1958 system atic m onitoring of C O 2 has been m ade at the M auna
Loa Observatory in H aw aii (B acastow at al., 1985) and later at several other stations (B oden et al.,
1990). One should note that nondispersive infrared techniques now used at M auna Loa and other
stations are not direct chem ical m easurem ents. The results m ay be influenced by the presence of
other "greenhouse" gases in air sam ple s w ith absorption bands overlapping those of CO 2 . This is
suggested by results from 19 S candinavian stations (B ischof, 1960) in w hich a sudden increase in
C O 2 concentration was observed after a chem ical m ethod w as replaced by the infrared (IR ) technique
in 1959. Other IR -absorbing gases than C O 2 have continuously increased their abundances in the
global atm osphere. This could have given continuously increasing and too high "C O 2 readings" at
M auna Loa and other stations using the infrared technique. Independent non-instrum ental chem ical
analyses of the reference gases and flask sam ples of the atm osphere have not been seen reported,
and should certainly be required.
The annual m ean concentrations reported from the M auna Loa observatory increased from 315.55
p pm in M arch 1958 to 351.45 in January 1989 (P ales and K eeling, 1965; K eeling et al., 1 9 8 9 ;
Thoning et al., 1989; B oden et al., 1990). The M auna Loa data have been regarded as representative
for the global concentration of C O 2 in the atm osphere. This seem s to be rather doubtful, due to the
fa c t that the site is exposed to vast local natural em issions of C O 2 , and also CO 2 from m an-m ad e
sources.
The published results of the M auna Loa m easurem ents indicate that the atm ospheric C O 2 load has
system atically increased about 10% during the past 30 years. Together w ith concentrations of C O 2
found in air b u b bles trapped in glacier ice, these results have often been used as a proof that the
atm ospheric C O 2 level has increased by 25% since about 1850 (e.g. Schneider, 1989; IP C C , 1990).
The predictions that the atm ospheric C O 2 level will d o uble around the year 2030 are based on
extrapolating the com bined results of glacier and air m easurem ents, and on the assum ption that the
25% inc re a s e is s olely due to m an-m ade sources. C om bining the glacier data w ith atm ospheric
m easurem ents into one sm ooth curve was m ade possible only by assum ing that the air entrapped
in the ice is 95 years younger than the age of ice in w hich the air w as entrapped. How ever, this
assum ption was found to be incorrect (Jaw orow ski et al., 1992; see also discussion in 5.2). W ithout
this assum ption the CO 2 concentration in air recovered from the 19th century ice is the sam e as that
at about 1980 in the M auna Loa record.
The m easurem ents in Scandinavia were carried out by a chem ical m ethod, different from that used
at M auna Loa. Therefore, it is especially interesting to com pare them . For 19 stations in S candinavia
the total annual m ean C O 2 concentrations were 326 ppm v in 1955, 321 ppm v in 1956, 323 ppm v in
1957, 315 ppm v in 1958, and 331 ppm v in 1959 (B ischof, 1960). The first M auna Loa annual m ean
-17for 1959 was 315.83 ppm v, 316.75 ppm v for 1960, 317.49 for 1961, 318.30 for 1962, and 318.83
for 1963. There w as an apparently decreasing trend in S candinavia during the first four years before
introducing the infrared technique, w ith a m arked rise after its introduction, and a steadily increasing
trend at M auna Loa w here only the infrared technique w as used. No increasing trend in the CO 2 air
concentrations betw een 1957 and 1961, m easured by the infrared technique, was observed in
S candinav ia a t altitudes of 1000 to 3000 m eters (Bischof, 1962). The decreasing trend in
S candinavia could hardly be due to errors in the analysis, w hich had an accuracy not m uch different
from that of the technique used at that tim e a t M a u n a L o a . The cause of the inconsistency of the
S candinavian and M auna Loa data rem ains unclear.
A s the M auna Loa data are extensively use d a s representative for the average global air
concentration of CO 2 , we discuss here the accuracy of the M auna Loa m easurem ents, to illustrate
the difficulties involved in estim ating levels of C O 2 in the atm osphere.
The observatory is located at the slope of the active M auna Loa volcano, w hich has had on the
average one eruption every three and half years since 1832 (Encyclopaedia B ritannica, 1974; Sim kin
et al., 1981). Follow ing an eruption in 1975, the M auna Loa volcano rem ained at rest until M arch
1984, w hen about 220 m illion tons of lava covered an area of about 48 km 2 . P re-eruption activity had
been occurrin g s in ce about 1980 (K oyanagi and W right, 1987; K oyanagi et al., 1987). The C O 2
c o ntent of volcanic gases em itted, associated w ith various types of lava, w as reported by Rub e y
(1951). The concentration of C O 2 in the gases em itted from the M auna Loa and K ilauea volcanos of
H aw aii reaches about 47% . This is m ore than 50 tim es higher than in volcanic gases em itted in m any
other volcanic regions of the world. The reason fo r th is is the alkaline nature of this volcanism ,
strongly associated with m antle CO 2 degassing. The Kilauea volcano alone is releasing about 1 M T
C O 2 per year, plus 60 - 130 kT S O 2 per year (Harris and A nderson, 1983).
The observatory is also exposed to perm anent C O 2 vents from the volcanic caldera and a rift zone
situated only 4 km upslope from the observatory (P ales and Keeling, 1965), and from som e distant
sourc e s d o w n s lo p e (K eeling et al., 1976). P ales and Keeling (1965), in their description of
m ethodology and the sam pling site, did not m ention that the world's largest active volcanic m ass,
K ilauea, with the largest and m os t a c tiv e v olcanic crater on E arth (5 km long and 2 km w ide) is
situated only 27 km southeast from the M auna Loa observatory. Frequent eruptions of this volcano
occurred during the 1960s and 1970s. C O 2 em ission from K ilauea also occurs in non-eruption periods
(Decker and K oyanagi, 1983; Decker et al., 1987). E m issions of up to 5000 tons of CO 2 per day were
recorded from the sum m it crater of this volcano in non-eruption periods (Gerlach and Taylor, 1990).
M ore recently, increased activity of K ila u e a begun in January 1983 and continued throughout
1984. There were 16 m ajor gas-charged eruptions in 1984, w ith fountains of lava several hundred
m eters high, an d w ith an average production of lava of about 10 m illion tons per episode. A word
"vog" (from "volcanic fog") has been coined on the island of Haw aii to define the volcanic haze that
has been hanging over the island since Kilauea's latest eru ptive phase began in 1983. This "vog"
consists of w ater vapor, C O 2 , and S O 2 . The conditions m ight resem ble a m ild city sm og (B endure and
Friary, 1990). S uch conditions should influence the CO 2 readings at M auna Loa O bservatory. The
-18question arises how the air at M auna Loa can give a representative average global atm ospheric C O 2
level.
To account for the influence of volcanic em issions from the neighboring 10 km long rift zone and
caldera at M auna Loa, Pales and K eeling (1965) calculated an increase in C O 2 concentration of 2
ppm for a certain "w eather type", w hich is about three tim es higher than the observed 0.68 ppm
a v e ra g e increase per year. The eruption events of the M auna Loa and K ilauea volcanoes, o r fo r
quiescent em ission of C O 2 from the gigantic K ilauea crater, w ere not discussed by these authors.
E leven years later K eeling et al. (1976) m entioned the prolonged period of K ilauea activity w hich
com m enced in N ovem ber 1967 and ended in M arch 1971. In M arch 1971 a locked chain gate was
erected across the road to the M auna Loa observatory 0.5 km from the CO 2 intakes, to control the
autom otive traffic. A current tourist guide instructs tourists: "P ark in the lot below the weather station
[because] the equipm ent use d to m easure atm ospheric conditions is highly sensitive to exhaust"
(B endure and Friary, 1990). (The chain w as not in use and cars w ere parked im m ediately under the
C O 2 intakes w hen one of the present authors visited the site in M arch 1992).
A n exam ple of the variations of hourly average atm ospheric concentration of C O 2 durin g three
consecutive days at M auna Loa is show n in Figure 1. It dem onstrates that it seem s extrem ely difficult
at this locality to determ ine values representing global concentrations. This point is further illustrated
by Figure 2, where the effect of data selection at the Cape M atatula Sta tio n in Sam oa (another
volcanic island, w ith H aw aii-type hot-spot volcanism ) is presented. The description of the m ethods
used at both statio n s fo r elim ination of irregularities to get a representative value confirm s this
conclusion.
The effects of different kinds of local vegetation on the concentrations of C O 2 in air have not been
satisfactorily treated, and a num ber of features of the published curves for atm ospheric variation of
C O 2 through approxim ately the last 30 years have not been explained (K eeling et al., 1989). P ales
and K eeling (1965) discuss the depletion of atm ospheric C O 2 by a tropical forest downhill from
M auna Loa O bservatory, w hich was supposed to cause "dips" in their readings. Grassland production
of CO 2 m ay have a large influence on CO 2 levels in the air, as w ell as respiration and decay of
organic m atter. D aily fluctuations of C O 2 concentration over a grassland w ere found to reach 40 ppm ,
and the seasonal variations (betw een June and S eptem ber) reached about 25 ppm (S pittlehouse and
R ipley, 1977). D aily variations over a w heat field can alm ost double the am oun t o f C O 2 in the air
(Fergusson, 1985). P ales and Keeling (1965) noticed the existence of C O 2 sources at the M auna Loa
observatory itself, such a s e x h a usts of a diesel engine-driven generator and autom obile pollution
w hich have becom e a problem (Keeling et al., 1976). In view of these points of criticism , the claim
of P ales and K eeling (1965) that "the observatory is thus an excellent, if not ideal, site for m easuring
C O 2 in the upper air" seem s to be overly exaggerated.
Throughout the whole period of m easurem ents the results w ere "edited" (an expression used by
B acastow et al., 1985) to account for lo c al disturbances causing both higher and low er C O 2
concentrations. A s P ales and Keeling (1965) stated, the measurem ents are clearly locally influenced.
The authors applied "om issions of variable periods from the daily averages" to elim inate both high
-19-
Figure 1. Hourly average atm o s pheric CO 2 concentration at M auna Loa (Haw aii). V ertical bars
indicate periods rejected from the records, as local C O 2 releases w ere suspected. H orizontal arrow s
indicate "steady periods" supposed to give a m ore representative concentration. (A fter Keeling et al.,
1976.)
-20-
Figure 2. R esults of the C O 2 d a ta selection procedure for 1984 at C ape M atatula, S am oa. A :
unselected hourly averages. The vertical lines represent the full range, w hile the boxes represent the
range of the m iddle half of the data. Indicated are the 25th percentile, m edian, and 75th percentile
points. B : average daily cycle for the entire year of unselected data. C: average daily cycle for the
entire year after selection. (A fter W aterm an et al., 1989).
-21and low readings. In Figure 1 about 80% of the readings w ere om itted. In addition, days without data
constitute 17% of the record. For these lacking days the values were assigned by linear interpolation
(Keeling et al., 1976).
The values that w ere om itted w ere defined as "m easurem ents that occur w hen the air trace show s
significantly m ore variation than the reference gas" (B acastow et al., 1985). W hen estim ating daily
a v e ra g e s the "too high" or "too low " values w ere elim inated by "visual inspection". This clearly
introduced a subjective factor in the estim ates of tem poral trends.
O ther long term m easurem ents of atm ospheric C O 2 w ere recorded at stations not exposed to vast
local volcanic em issions of C O 2 . These m easurem ents show an increasing long-term trend sim ilar
to that of the M auna Loa data, but w ith different am plitudes for the seasonal changes. How ever, an
"editing" process sim ilar to that used for the M a u n a L o a re c ord was applied also to the data from
these other stations to account for (assum ed) local disturbances.
It is difficult to understand that the effect of a powerful natural injection of 6 G T of CO 2 (equivalent
to 2 G T C ) by the El N iño - S outhern Oscillation (E N S O ) 1982-1983 event (G audry et al., 1987), and
of nine other E N S O events, are not directly visible in the M auna Loa and th e S o u th P o le records
(S iegenthaler, 1990). On the other hand, it has b e e n c la im ed that the records detect the annual
anthropogenic C O 2 releases, w hich are just 2-3 tim es higher than the E N S O events. This m ay m ean
that the "editing" procedure produces records not reflecting im portant large-scale CO 2 variations in
the atm osphere. The anthropogenic em issions are sm all com pared to natural CO 2 fluxes, and it
cannot be excluded that the reported increase in atm ospheric C O 2 is due to fluctuation of the natural
fluxes. The tacit assum ption in the "greenhouse" warm ing hypothesis that the natural CO 2 fluxes are
constant, is unrealistic.
In order to test the M auna Loa C O 2 data w e com pared them w ith fossil fu e l C O 2 em ission data
com piled by B oden et al. (1990) (Figure 3). For this com parison w e assum e that the M auna Loa data
represent the average global atm ospheric C O 2 concentration. The a n n u al C O 2 em issions into the
atm osphere from fossil fuel burning and cem ent production increased alm ost continuously from about
2.6 GT C per year in 1960 to 5.9 G T C per year in 1988. The rate of increase w as less betw een 1973
and 1975, and turned into a decrease between 1979 and 1983. The annual increases of the
atm ospheric CO 2 m ass inferred from the M auna Loa data (referred to as "M auna Loa" atm ospheric
C O 2 m ass) revealed a totally different character (Figure 3 A ). These increases fluctuated irregularly
b e tw e e n 1.0 GT C in 1963 to 5.5 GT C in 1988, with highs and low s not related to in d u s tria l
em issions (e.g. in 1963, 1969-72, 1973-74, 1981-83, and 1986).
D uring the so called "petroleum crisis" in 1973 - 1975 the annual increase of the CO 2 atm ospheric
m ass, inferred from M auna Loa data ("M auna Loa" atm ospheric C O 2 m ass), decreased from about
4.5 to about 1 GT C w hereas the real fossil fuel CO 2 em issions w ere nearly constant at that tim e
(Figure 3 A ). T h e s u dden deep drop betw een 1973 and 1974 in year to year changes of annual
increases of "M auna Loa" atm ospheric C O 2 m ass, w hen the changes in real C O 2 em issions from
anthropogenic sources were stable (F igure 3 B ), m ay be an effect of the subjective factor in the
"editing" of the M auna Loa data.
-22That the M auna Loa data are not representative for the m an-m ade em issions is strongly suggested
by the character of the interannual changes in the M auna Loa data, which are highly different from
the w ay the fossil fuel C O 2 em issions vary from year to year. The equilibration betw een C O 2
concentration in the atm osphere and the dissolved inorganic carbon in the sea is very short (about
three quarters of a year according to Bolin, 1982). Therefore one m ight expect that m ost of the annual
m an-m ade perturbation in atm ospheric C O 2 w ould be visible in the M auna Loa data in the sam e, or
in the next year, provided the m an-m ade C O 2 fluxes are not m asked by the natural fluxes. Figure 3
B show s sharp upw ard and dow nw ard interannual changes in the "M auna Loa" atm ospheric C O 2
m ass. The am plitude of these changes reaches up to about 5 .5 G T C per year. These dram atic
"M auna Loa" C O 2 changes are not reflected in the m ore steady annual em issions of fossil fuel C O 2,
w hich has a m axim um am plitude of 0.28 GT C . In about half of the years the "M auna Loa" C O 2 m ass
changed in a direction opposite to that of fossil fuel CO 2 em issions.
These discrepancies suggest that the M auna Loa record was strongly influenced by som ething
else than fossil fuel C O 2 . A s indicated before, one possibility is the natura l flux of C O 2 (e.g. C O 2
outgassing from the ocean and the interior of the E arth).
3.2 NINETEENTH CENTURY CONCENTRATIONS
In the 19th century the m easured C O 2 concentrations in the atm osphere ranged from -250 to 550
ppm v (Fonselius et al., 1956). O n the basis of these m easurem ents (m ainly carried out at M ontsouris
observatory near Paris) the pre-industrial level of CO 2 was estim ated by Callendar (1940, 1958) to
be 292 ppm . A ccording to Callendar (1958) the 19th century data do not "show a significant trend
be tw e e n 1 8 7 0 a n d 1900", i.e. in a period when the annual em issions of CO 2 from fossil fuels
increased by a factor of 3, from 0.15 to 0.5 GT of carbon (Elliott, 1983).
C allendar's estim ate of 292 ppm for pre-industrial C O 2 atm ospheric level was obtained by applying
a selection m ethod. This m ethod w as questioned by Slocum (1955), w ho indicated that w ithout such
selection these data averaged 335 ppm . S locum (1955) pointed out that from a set of 26 19th century
averages C allendar rejected 16 that w ere higher than the global average of 292 ppm , and only 2 that
w ere low er. O n the other hand, from the 20th century set Callendar rejected 3 averages that were
low er than his global average of 317 ppm , a nd none that was higher. This bias in the selection
m ethod is dem onstrated in the set of 19th century data com piled by Fonselius et al. (1956) (Figure 4).
The M ontsouris data show large shifts of m onthly m eans (>10 ppm from one m onth to the next)
and they show no seasonal variations. This is incom patible with m odern records show ing m uch less
daily scatter and a c le a r s easonality (W aterm an, 1983). The m ost obvious jum p occurred at
M ontsouris betw een June and July 1890, when a change of 27 ppm w as reported. This change is
about the sam e as the increase observed at the M auna Loa observatory during 22 years from 1958
to 1980. S om e of the M ontsouris m easurem ents indicate extrem ely high or low values (355 ppm and
243 ppm , respectively). This has been interpreted as evidence of analytic a l a n d sam pling
shortcom ings (Stanhill, 1983).
-23-
Figure 3. M auna Loa representation of global atm ospheric C O 2 m ass, and fossil fuel C O 2 em issions.
A : A nnual increase in "M auna Loa" atm ospheric C O 2 m ass and annual em issions of fossil fuel C O 2.
B : Y ear to year c h anges in "M auna Loa" atm ospheric C O 2 m ass increase and fossil fuel C O 2
em issions; the ordinate represents the difference betw een one year and the pre v ious year. The
"M auna Loa" atm ospheric C O 2 m ass w as calculated assum ing that 1 ppm v C O 2 corresponds to an
atm ospheric m ass of 2.12 GT C . Calculated from data in B oden et al. (1990).
-24In 1880 and 1881 M üntz and A ubin determ ined C O 2 in the center of P aris. The m easured
concentrations averaged 62 ppm m ore than those at the sam e period at M ontsouris. Of their ten nonurban sites, seven reported CO 2 concentrations greater th a n at M ontsouris, the m ean difference
being 12 ppm . In Paris the concentrations recorded in 1880 and 1881 averaged 325 ppm (W igley,
1983).
It is difficult to believe that a m onthly m ean of 355 ppm for Decem ber 1878 and a m ean of 243 for
A pril 1880, reported for the M ontsouris O bservatory, are truly representative of global background
values (W aterm an, 1983). The M ontsouris in v estigators w ere probably not running any blanks or
duplicate m easurem ents that could shed light on the precision of their m easurem ents.
The chem ical m ethod used at M ontsouris was sim ilar to the procedure used by M üntz and A ubin.
This m ethod could lead to erroneously low results due to incom plete stripping of the CO 2 from the
airstream by a reaction w ith K O H . Another source of error was probably a reaction of the reagents
on the laboratory a p p a ratus (W aterm an, 1983). The enorm ous tem poral scatter of the M ontsouris
results (of 27 ppm ) and the geographical scatter of the M üntz and A ubin data (of 40 ppm ) do not
represent the "natural noise", but w as due to sam pling and analytical shortcom ings, and indicates
that these data are less reliable than 20th century m easurem ents.
It is interesting to note that there is a te n dency to choose a low average value for the C O 2
concentration in 19th century atm osphere because "the m ost com pelling support for a 270 ppm preindustrial CO 2 level com es from direct m easurem ents of C O 2 in the ice cores" (W igley, 1983). B ut
those w ho m easured CO 2 in the ice cores also preferred to select as true the low er values, because
they were "w ithin the range of the estim ated [by C allendar, 1958] pre-industrial atm ospheric content
of 290 ppm " (B erner et al., 1978).
This dem onstrates that a s u b jective factor biased the C O 2 values chosen by the 20th century
researchers, and resulted in a too low estim ate of its pre-industrial level.
-25-
Figure 4. A verage atm ospheric C O 2 concentrations m easured in the 19th and 20th century. E ncircled
are the values used by C allendar (1958). R edraw n after Fonselius et al. (1956).
-264. TREE RINGS AS INDICATORS OF CO 2 IN THE ATMOSPHERE
The stable carbon isotope ratio, 13 C / 12 C , in tree rings w as used to calculate the CO 2 contents of the
atm osphere in the past. The results of S tuiver et al. (1984) are often used (even in official docum ents)
to draw a curve show ing th e C O 2 content in the atm osphere during historic tim e (e.g.
M iljøverndepartem entet, 1988-1989). The original paper by S tuiver et al. (1984) and other w orks (e.g.
D eines, 1980) dem onstrate that contem poraneous trees from the sam e period show large variations
in their carbon isotope ratios. S tuiver et al. (1984) em phasize that after norm alizing and averaging
the data from different trees, there w as still a rem aining variability in the data resulting in a standard
error of ± 16 ppm (assum ed 68 % confidence level) in the calculated CO 2 contents (analytical and
sam pling errors not included). For th e c o m p u tational uncertainty alone, a 95 % confidence level
w ould give ± 31 ppm . A large part of the variation calculated by S tuiver et al. (1984) is less than this
statistical uncertainty.
R ecent system atic studies on stable isotope geochem istry in plants cast serious doubts on the
usefulness of this m ethod in paleoclim atology because of isotopic inhom ogeneity in the plants
them selves (e.g. Y akir et al., 1989) and post-photosynthetic m odifications in the wood cellulose (e.g.
D eN iro and C ooper, 1989). A system atic study by Tans and M ook (1980) of carbon isotopic ratios
of the com plete circle of a single tree ring revealed that the carbon isotope variations were up to 3.5
perm il.
The m axim um variation in δ13 C given by S tuiver et al. (1984) for the last 1800 years was 3 perm il,
m ost of the variation was w ithin 2 ± 1 perm il. The calculated curve for the atm ospheric C O 2 variation
for the last 1800 years from carbon isotope ratios in tree rings is hence based on isotope variations
w hich are less than the variations expected in a single tree ring. This fact supports objections to the
use of tree ring isotope analysis in paleoclim atology.
The δ13 C m ethodology calculates the C O 2 contents of the atm osphere throu g h tim e b y isotopic
m ass balance calculations in a sim plified physical box m odel (e.g., O eschger et al., 1975) betw een
the atm osphere, biosphere and the sea at constant tem perature. The ratios of light stable isotopes,
like the carbon isotopes, are strongly changed by tem perature. This is the reason w hy such isotopes
find their m ajor use in science as paleotherm om eters. The surface tem perature of the Earth has
varied considerably throughout the last 1800 years (see Figure 9). A tem perature change of a few
degrees C can account for the observed carbon isotope variations by S tuiver et al. (1984). V ariations
in rainfall would have influenced photosynthesis and hence the resulting carbon isotope ratios of the
tree cellulose (governed to a large extent by biologic kinetic isotope effects). The m odel also neglects
C O 2 releases (or accum ulations) from the lithosphere and the E arth's interior, the largest reservoirs
of carbon involved. Furtherm ore, the m odel assum es that the global sourc e of C O 2 alw ays has a
constant carbon isotopic com position. W ith so m a ny sources and sinks of carbon, each with its
d is tin ct carbon isotopic signature (O hm oto and R ye, 1979), this assum ption m ay b e a n
oversim plification.
Finally, the carbon isotopic com position of corals and contem poraneous trees do not correlate
(N ozaki et al., 1978). H ow ever, they should correlate if all the assum ptio n s w e re valid and the
-27m ethods suitable. It m ust be concluded that the CO 2 c o n tents in the atm osphere calculated from
carbon is o to p e a n a lysis of tree rings cannot be considered a valid tool in paleoclim atology, and
cannot be used as evidence of changing atm ospheric C O 2 levels.
-285. CO 2 IN GLACIER ICE
S tudie s of C O 2 concentrations in glacier snow and ice are a cornerstone of the increasing
"greenhouse effect" hypothesis. R esults from these studies are the m ost im portant input param eters
in the m odels of the global carbon cycle, and are used as the basis for estim ates of clim atic change
caused by hum an activities (e.g. IP C C , 1990). Therefore, they deserve a w ider discussion.
The early determ inations of C O 2 in air bubbles trapped in glacier ice, carried out in the 1950s to
the 1970s, recorded m uch higher concentrations in the pre-industrial sam ples than the m ean value
of about 290 ppm estim ated by C allendar (1958) for the nineteenth century a ir. It w as during the
1980s that the C O 2 levels in the pre-industrial ice began to be interpreted as being in agreem ent with
the Callendar estim ate.
The first CO 2 studies in glacier ice were initiated at th e U n iv ersity of Oslo, and the first
m easurem ents w ere carried out at the S torbreen glacier in Jotunheim en, Norway (Coachm an et al.,
1956; Coachm an et al., 1958 a; Nutt, 1959). These studies in d ic a te d that the C O 2 content in air
bubbles in the pre-industrial ice ranged betw een 200 and 2900 ppm , and that in the older parts of the
glacier it was reduced by a factor of three due to leaching by liquid w ater. In sam ples of old ice from
the Greenland ice cap, C oachm an et al. (1958 b) found that the oxygen values w ere close to those
of the present-day atm osphere, but values for C O 2 w ere tw ice as high as today. They concluded that
the atm osphere m ight have b e e n ric h e r in carbon dioxide at the tim e this ice was form ed. It w as
recognized in these early studies that the original com position of the air trapped in the ice is changed
due to various physical and chem ical processes (Scholander et al., 1961).
C oachm an et al. (1958 a) stated that the loss of C O 2 from the ice m ight occur during m elting, both
at the site of ice form ation and at the older part of the glacier. They also noticed that the ice could be
contam inated by organic dust, w hich subsequently becam e oxidized, enriching the CO 2 of the air in
the bubbles. Hem m ingsen (1959) observed a great m obility of CO 2 in the glacier ice, which he
supposed to be an effect of diffusion in the intercrystalline liquid brine film s. S uch film s and veins of
liquid exist in ice even at extrem ely low A ntarctic tem peratures (see below ).
A lder et al. (1969) found that C O 2 can be adsorbed at the ice surface and rem ain adsorbed even
in vacuo, and that this adsorption is not reversible. This effect m ight decrease the level of this gas in
the air bubbles, and also increase its content in the ice by adsorption from the laboratory air. A lder
et al. (1969 ) a n d S c h o lander et al. (1961) found evidence that, in the Greenland ice, oxygen is
rem oved by oxidation of organic m atter. This m ay lead to an increase of CO 2 in the ice.
These early studies posed th e question whether glacier ice is a reliable m atrix for study of the
com position of the ancient atm osphere. A s w ill be seen later, after three decades the question is still
entirely open.
-295.1 CHANGES IN ORIGINAL GAS COMPOSITION
S tauffer et a l. (1984) supposed that, am ong the factors influencing the CO 2 levels in the air
bubbles, are the presen ce of clathrates (gas hydrates), m elt layers, and m icrobubbles. To this list
S tauffer and Oeschger (1985) added adsorption of a ir c o m p o n e n ts on the firn grain surfaces and
enclosing the air during snow flake form ation, zone refining processes parallelling crystal grow th,
interaction of C O 2 w ith the ice itself, fracturing of the ic e , chem ical reactions betw een C O 2 and
im purities trapped in the ice, and oxygen exchange betw een C O 2 and w ater. They found that the
increase of the C O 2 concentration parallels the dust content in all the m easured ice cores. They also
recognized a dependence of C O 2 levels on the acidity of the ice: the sam ples show ing the highest
pH values had the low est C O 2 contents. B ut perhaps the m ost im portant factor, changing the original
com po s itio n o f a tm ospheric air trapped in the glaciers, is liquid water present in the ice even at
extrem ely low tem peratures.
5.1.1 LIQUID IN ICE AT LOW TEMPERATURES
The A ntarctic and Greenland ice cores, used for historica l s tu d ie s o f C O 2 changes in the
a tm o s p h e re, are usually regarded as a steady state closed system in w hich the origina l
concentrations of this g a s are preserved indefinitely. The m ain argum ent used in support of this
assum ption is that the existence of a liquid ph a s e in th e polar ice can be neglected at or below a
m ean annual tem perature of -24EC (e.g., B erner et al. 1978; Raynaud and B arnola, 1985). This liquid
phase w ould enable diffusion processes to change the original com position of air contained in the
ice. How ever, liquids are present in A ntarctic ice at m uch lo w er tem peratures (see below ), so this
argum ent now seem s invalid.
B erner et al. (1978) stated that the high content and wide variations of CO 2 observed in ice could
be u n d e rs to o d if, during the sintering processes of snow to ice, a liquid phase was present. They
believed that w ith m ean annual tem peratures at a polar ice cap of -24EC , the influence of the liquid
phase can be neglected. This is w hy the later C O 2 s tu d ie s in the ice were carried out only on the
polar ice caps. This opinion is, h o w e v e r, n o t c o rrect, in view of the evidence that at low A ntarctic
tem peratures both m elt layers and intercrystalline liquids occur in the ice (see discussion below ).
S cholander et al. (1961) listed several processes that could change the original com position of gas
inclusions in ice. O ne of their m ost im portant findings w as an obse rv ation of enrichm ent of C O 2 ,
oxygen, and argon (relative to nitrogen) in air bubbles, by freezing them out from the liquids present
in the firn or ice containing dissolved atm ospheric gases. They also noticed that supercooled fog,
carrying dissolved gases, leads to C O 2 enrichm ent when freezing on the surface of ice sheets. This
is what w ould be expected in view of the changes in gas ratio s w h e n air dissolves in water. On a
m olar basis the solubility of CO 2 in water at a tem perature of 0EC is 73.5 tim es higher than that of
nitrogen, the solubility of oxygen is 2.0 tim es that of nitrogen, and the solubility of argon 2.4 tim es
higher than that of nitrogen. Therefore, at 0EC the concentration of o x ygen in the air dissolved in
-30w ater is 35% , an d n o t 21% as in the atm osphere (H odgm an et al, 1962; W east et al., 1989). Gas
ratios different from those in atm ospheric air should be characteristic of prim ary gas inclusions (see
below ) in ice.
The dissolution of air gases in the liquid contained in ice is likely to start w ith the form ation of snow
crystals in the atm osphere. The liquid containing sulfuric acid is incorporated into snow crystals as
a film on a solid nucleus or added by rim ing, and it is expected to rem ain as a liquid on the outside
of the grain crystals (W olff et al., 1988). Therefore, it seem s that at least a part of enrichm ent of C O 2 ,
O 2 and A r observed in glacier ice occurs during the atm ospheric history of the snow crystals.
In later studies of ice cores from both central Greenland and A n ta rc tic a (see e.g. Neftel et al.,
1982; R aynaud and B arnola, 1985), m elt layers w ere com m only observed. A lso Boutron (1986) notes
the possibility of disturbing the old ice deposits in Greenland and A ntarctica b y percolating water
during the sum m er m onths. D uring the A ntarctic sum m er, m eltw ater w as found at a depth of about
1 m in the ice sheet at the station Troll, D ronning M aud Land (A ntarctica), 250 km inland, where the
s um m er air tem perature was below -20EC (H agen, 1990). V eins of liquid are also present in th e
capillary netw ork betw een ice crystals, even at extrem ely low tem peratures in the Antarctic ice-sheet
(M accagnan, 1981; M ulvaney et al., 1988).
A cid-water m ixtures in layers at ice grain boundaries have been found to rem ain liquid dow n to the
eutectic point of -73EC (W olff et al., 1988). In Antarctic ice high aqueous concentrations of 4.9 M
sulfuric acid (H 2 S O 4 ) w ere found in veins at triple-junctions, w here the ice grains m eet (M ulvaney et
al., 1988). U sing isotopic ( 14 C ) dilution m ethod S hchennikova et al. (1957) found that the solubility
of CO 2 in concentrated H 2 S O 4 increases w ith decreasing tem perature and increasing concentration.
A t 20°C and above 85 w eight % concentration of acid the CO 2 solubility w as found to be about 0.9
m l/m l, i.e., about the sam e as in distilled w ater. This indicates that C O 2 from air bubbles trapped in
the ice can penetrate into the ne tw o rk of intercrystalline acid brine and its concentration in the air
inclusions m ay decrease w ith tim e.
W hen alkaline dust neutralizes acids in the ice, the resulting salt-H 2 O m ixtures m ay cause freezing
point depressions. The extent of freezing depression is dependent on the type and concentration of
salts in the intercrystalline brines. In this w ay, the brine can keep its liquid state down to -50EC or
even low er (R oedder, 1984; S hepherd et al., 1985; Oakes et al., 1990). W hen freezing salt aqueous
solutions, pure ice crystallizes, leaving a m ore saline solution. In th e sea ice the liquid brine was
observed dow n to a tem perature of -70EC (W eeks and A ckley, 1982). This effect m ay influence the
behavior of gases in the deep ice strata. The high content of Na at V ostok (A ntarctica) at depths of
about 500, 850, and 2000 m coincided w ith the low CO 2 content recovered from the air inclusions.
The low content of N a at the depths of about 1420, 1770, and 1850 m coincided w ith the increasing
C O 2 concentrations (Barnola et al., 1987).
The im portance of the discovery of a liquid phase in ice at extrem ely low tem peratures is difficult
to overestim ate. It suggests that in glacier ice there is a quasi infinite netw ork of liquid-filled veins
through w hich dissolved substances can m igrate during long periods of tim e. Thus the polar ice is
com posed of water in liquid, solid and gaseous states which are interchanging during the long lifetim e
of the ice sheets; a large scale displacem ent of m ass is involved in this interchange (Jaw orow ski et
-31al., 1992). Therefore, the validity of A ntarctic and G reenland ice cores as reliable objects for study
of chem ical and isotopic com position o f th e a tm o sphere of the past epochs is therefore highly
questionable.
C onsiderable know ledge of gas and fluid inclusions in solids has been gained from the geological
sciences (e.g. H o llister and C raw ford, 1981; Roedder, 1984; Shepherd et al., 1985; Bodnar et al.,
1985). The fundam ental principle underlying all gas/fluid inclusion studies is that gases and fluids
inside the inclusions are representative portions of the gases and fluids present in the phase from
w hich the host m ineral grew (provided the gas /flu id is hom ogenous at the tim e of trapping). S uch
inclusions are called prim ary inclusions.
C hanges m ay occur inside the inclusion s a fte r they were trapped, even if the inclusion volum e
rem ains constant. The te m perature m ay change and give rise to phase changes or chem ical
reactions within the inclusion phases or w ith the inclusion walls. S olids m ay precipitate from
dissolved species in the inclusion liquids, gases or vapors. Then, the m easured characteristics m ay
not be the sam e as the original ones.
The in clusion volum e m ay change after inclusion trapping, and a chem ical re-equilibration
internally betw een phases or w ith the inclusion w all m ay take place after the change in tem perature,
pressure or volum e. S olid, liquid and gas inclusions can readily m igrate through crystals, and even
m etals, under the in fluence of strong therm al gradients (S hepherd et al., 1985). There is also
evidence that the inclusions m ay leak, even from such sturdy m aterials as quartz, giving wrong
inform ation about the C O 2 content of the inclusions (Bakker and Jansen, 1990).
Inclusions form ed after the host crystal has been form ed are regarded as secondary inclusions.
These are com m only form ed in post-crystallization fractures initiated during m echanical or therm al
stress. These cracks are later sealed, form ing characteristic trails of secondary inclusions. S uch
inclusions are unrelated to and not representative of the orig in a l v o latile environm ent during the
grow th of the host m ineral. In a sim ilar way, pseudosecondary inc lu s io n s m a y develop, but the
fracturing and "healing" take place before th e crystal grow th has term inated. Pseudosecondary
inclusions should reflect the volatile environm ent during the last grow th stages of the host m ineral,
but not necessarily the initial stages of grow th. These factors are not taken into account in the papers
reporting recent studies of C O 2 in polar ice. E ven such fundam ental fact as the total disappearance
of the air bubbles in deep ice (see below ), which are used for paleoclim ate reconstructions (e.g.
B arnola et al., 1987), is not discussed or m entioned in these papers.
The upper, several tens m eter deep part of the polar ice sheets is com posed of porous snow and
firn. Firn gradually changes into solid ice (at a depth between about 60 and 100 m eters) in which the
p o res are occluded and form "air bubbles". B elow about 900 to 1200 m eters the air bubble s
disappear in ice. In the ice cores recovered from deep bubble-free ice, the secondary gas cavities
reappear, after relaxation of the load pressure. In these new secondary cavities the com position of
gases cannot be the sam e as in the original air bubbles (Jaw orow ski et al., 1992). A s the com plete
o c c lusion of the pores lasts from tens to thousands of years, depending on local conditions (e.g .
precipitation rate), the air in the ice w as assum ed to be m uch younger than the ice (see e.g. IP C C ,
1990). If this speculative assum ption w ere correct, the air in the firn should freely com m unicate with
-32the atm os p h ere dow n to the firn-ice transition at about 60 to 100 m eter depth, where the final
occlusion occurs. How ever, in the polar firn there exists a dense, com plex structure of im perm eable
ice crusts (see below ) w hich precludes such free com m unication. The assum ption was first proposed
by B erner et al. (1980 ) a n d th e n e laborated by S chwander and S tauffer (1984).The latter authors
supported this supposition by citing the results from
39
A r dating o f Greenland ice (Loosli, 1983).
H ow ev e r, th e re s ults in Loosli's (1983) paper indicate the opposite: at 70 m depth at D ye-3,
G reenland, an ice core show s the age of the
sam e as the age of ice, determ ined by the
39
35
A r gas trapped in the bubbles to be ~70 years, i.e. the
S i m ethod.
The assum ption on the different age of air and ice w as posed at a tim e when the concentrations
of C O 2 w ere found to be sim ilar for the air bubbles from the ice deposited in pre-industrial period and
for the 20th century atm osphere (D elm as et al., 1980; S tauffer et al., 1981; N eftel et al., 1982; Neftel
e t a l., 1983). For detailed discussion of the assum ption on the different age of air and ice s e e
Jaw orow ski et al. (1992).
The ice is not a rigid m aterial. In the ice sheets the ice is constantly subjected to interaction with
liquid water, gaseous/liquid/solid phase changes, recrystallization and plastic deform ation. P ressure,
tem perature, and density of the ice are constantly changing, and com pression and disappearance
of air bubbles will occur. S econdary cavities are form ed in the decom pressed, bubble-free ice cores,
together with m acro- and m icro-fractures. The gas in the secondary cavities and in the air bubbles
com m unicate with th e atm osphere through these fractures, which are successively "healed" by
annealing.
In the very cold (belo w -5 0°C ) firn strata the absorption of solar radiation leads to intense
sublim ation and m elting of ice crystals, with a loss of H 2 O m ass. This leads to im portant structural,
chem ical and isotopic changes in the ice (see discussion in Jaworow ski et al., 1992). Under these
circum stances the ice neither behaves as a clo s e d s y s te m nor as a rigid host m aterial for air
inclusions. No textural evidence has been presented that the original pseudosecondary inclusions
have been preserved, and that no new secondary inclusions have been form ed. W e expect that the
secondary inclusions dom inate in deep ice core sam ples (the drilling m ay even destroy the integrity
of the rem aining orig in al pseudosecondary inclusions present), and that their validity for the
paleoatm ospheric research m ay be questioned until such evidence exists. It is also know n that ice
at great depths in the ice sheets loses all its original air inclusions (see 5.1.5). This suggests that the
gases diffused into the ice structure. This leads to problem s connected with later release of these
gases into secondary gas c a v itie s reappearing in the relaxed ice cores, and m akes it difficult to
distinguish betw een air originally trapped in the ice structure and air occluded in the air bubbles.
The release of gases from inclusio n s in m in erals is com m only achieved by decrepitation (e.g.
W alder and S egalstad, 1987) or crushing (e.g. O 'R eilly et al., 1990). The sam e m ethods have been
used for inclusions in ice. The m ajor differences betw een rock silicates and ice as a host m ineral for
the inclusions are the m uch low er m elting point for ice and higher vapor pressures. The solubilities
of g a s e s in the H 2 O phases are m uch higher than in, for instance, quartz. Furtherm ore, ice is
physically not a very good container for the gaseous inclusions.
-33Free CO 2 is a very c o m m on m ajor constituent of inclusions in a w ide range of geological
environm ents. On crushing a sam ple in the presence of liquid w ater or w ater vapor a large fraction
of the CO 2 form s aqueous carbonic acid (R oedder, 1984). In glacier ice the m easured CO 2 in the gas
phase will not represent the total C O 2 in the gas inclusions, because m uch CO 2 will be dissolved in
the (not analyzed) cold w ater, both present in the ice and produced at crushing. It seem s that this has
not been considered by S tauffer et al. (1981), when they described the "m issing CO 2 " during their ice
m elting runs. B arnola et al. (1987) reported that the presence of water vapor in inclusions gave low
values for C O 2 w hen crushing the ice. They therefore added a flat "best estim ate" of 5 ppm to the
low er C O 2 concentrations found. U nfortunately the papers usually give no inform ation on the C O 2 /
N 2 / A r ratios. H ence it is not possible to inspect the data as to the validity of their c la im e d
atm ospheric paleocom positions.
5.1.2 NITROGEN / OXYGEN / ARGON RATIOS
For estim ation of the age of the air trapped in the upper 80 m of A ntarctic and Greenland firn and
ice Oeschger et al. (19 8 5 ) assum ed that "the entrapm ent of air in ice is essentially a m echanic
process ... w hich occurs w ith no differentiation of the gas com ponents". In support of this assum ption
they also stated that N 2 / O 2 / A r ratios in ice are the sam e as in the am bient atm osphere. How ever,
this assum ption, w hich is crucial for age estim ates, is not in agreem ent with the analytical results.
In the cold A ntarctic enviro n m e n t a t th e B yrd S tation (Antarctica), and at Cam p Century
(G reenland), R aynaud and D elm as (1977) found that the oxygen content in gas trapped in the ice
w as reduced by 0.83 - 1.2% in com parison w ith atm ospheric air (i.e. an effect opposite to th a t
expected from the solubility data; H odgm an et al., 1962; W east et al., 1989) and enriched by 0.75
to 1.2% in argon. This is also opposite to the earlier findings of Scholander et al. (1961). In the firn
sam ples from the Pionerskaya and V ostok stations in Antarctica Raynaud and Delm as (1977) found
the concentrations of both these gases reduced in relation to nitrogen. The error of m easurem ent of
these gases w as reported to be 0.5% . The authors stated that "par rapport a la com p o s itio n
atm ospherique ... les effets observes sont donc faibles m ais sig n ific atifs et indiquent un
appauvrissem ent relatif des teneurs en O 2 et Ar". It is therefore difficult to understand w hy Oeschger
et al. (1985) interpreted the R aynaud and D elm as (1977) finding as evidence that "N 2 / O 2 / A r ratios
in ic e originating from very cold areas with no sum m er m elting show that, within experim en ta l
uncertainty, the m easured ratios agree w ith those in air". This statem ent O eschger et al. (1985) used
in support of an assum ption that the gas trapped in the ice is about 100 years younger than the
frozen w ater w hich encom passes it. H ow ever, the early data of the Oeschger group dem onstrated
that the A r / O 2 ratio in various parts of a glacier m ay reach a value of 0.95% , w hich is m uch low er
than the 1.19% in the am bient air (B erner et al., 1977). A ccording to B erner et al. (1977 ), the A r
c o n te n t in the ice (0.948% ) is slightly higher than in the atm osphere (0.934% ) and the O 2 conte n t
slightly low er (20.34% and 20.95% , respectively).
-34S tauffer et al. (1981) have m easured the Ar / O 2 ratio in the old Greenland ice, but refrained from
presenting the results. The value of this ratio was given by S tauffer et al. (1985) for ice sam ples from
D ye-3 (G reenland) collected betw een 0 and 35 m depth. In these sam ples they found extrem ely high
C O 2 c o n c e n tra tio n s , up to 28,000 ppm . In the surface layer the ratio of N 2 / O 2 w as 51.6 % (while
being 26.8 % in the atm osphere), N 2 / A r was 2.6 % (1.19 % in atm osphere), and N 2 / C O 2 was 4.4
% (0.042 % in atm osphere). This m eans that the en tra p m e n t o f a ir in ice leads to a large
differentiation of the gas com ponents. The N 2 / O 2 / Ar ratio in the glacier ice is of great im portance
for estim ating the validity of this m atrix for th e s tudies of tem poral changes of gases in the global
atm osphere. The observed changes of this ratio in the ice m ean that this m atrix is not suitable for
studying the original com position of the ancient atm osphere.
5.1.3 DEPLETION OF CO 2 IN SURFACE SNOW
A n im portant finding of R aynaud and D elm as (1977) was the observation that in surface firn (up
to 1 m depth) at the P ionerskaya and V ostok stations the concentration of CO 2 in the interstitial air
w as 160 to 240 ppm , respectively, w hereas at that tim e in the atm ospheric air this concentration was
reported to be 310 ppm . This dem onstra te s th a t, even in snow that was not subject to longer
firnification and firn-ice transition processes, the CO 2 content could have been reduced by up to 150
ppm , i.e. about 48% low er than in the am bient air of the sam e age. This im portant field experim ent
w as never repeated in the later C O 2 studies.
The striking feature of the glacier data used as an evidence for a recent m an-m ade C O 2 increase
is that all of them are from ice deposited not in the last decades but in the 19th century or earlier. In
these studies no info rm a tio n w a s presented on the recent concentrations of CO 2 in firn and ice
deposited in the 20 th c e n tu ry . The results of CO 2 determ ination in the pre-industrial ice are not
com pared with the CO 2 content in recently deposited snow, firn or ice but w ith its current levels in the
atm osphere. To justify such com parisons an assum ption was needed that the entrapm ent of air in
ice is purely a m echanical process, involving no chem ical differentiation of gases. H ow e v e r, as
appears from the discussion in this report, and as w as dem onstrated by Jaw orow ski et al. (1992), this
assum ption is w rong.
5.1.4 EFFECT OF CARBONATES
In 13 sam ples (of H olocene and W isconsin age) from an ice core from C am p Century (G reenland)
R aynaud and D elm as (1977) found an average C O 2 concentration of 3500 ppm , and an average C O 2
concentration of 1300 ppm in a core from V ostok (Antarctica). These values w ere determ ined in
w hole sam ples of ice, and not only in the air bubbles. The authors added som e drops of sulfuric acid
to the ice before m elting it. This probably added CO 2 to the original gas com position, by reactions of
acid with carbonates present in the sam ples. S ulfuric acid and sulfates are com m only present in the
-35polar ice. They m ay possibly react with carbonates in situ, as w ell as after sam pling of the cores of
old ice. This m ay lead to a w id e ra n g e of C O 2 values, as found in m any studies of these cores.
D elm as et al. (1980) suggested that the presence of carbonates in the ice m ay influence th e C O 2
m easurem ents.
H igh concentration of dissolved organic carbon (up to 0.32 m g per kg) w as found in the Greenland
snow by Tw ickler et al. (1986). These authors noticed that, in contrast to oxygen, the diffusion of C O 2
in th e ic e w a s m easurable. This could change the original concentration of this gas in the air
inclusions.
5.1.5 CHANGES OF AIR INCLUSIONS IN ICE
S tauffer et al. (1984) supposed that 20% of the observed tem poral changes in CO 2 content in air
inclusions ("bubbles") is due to enrichm ent in m icrobubbles. In the interpretation of their results from
the Dye-3 (G reenland) core they assum ed a 4 0 p p m s u rp lus CO 2 from the m icrobubbles, and
subtracted it from the m easured values. W ithout this subtraction, the CO 2 level which they found in
the approxim ately 30,000 - 40,000 year old ice w as about 340 pp m , w h ic h is s im ilar to the C O 2
content of the present atm osphere.
In polar glaciers, notably in A ntarctica, air bubbles m igrate and form inclusion clusters leaving
outcrops of bubble-free ice that are several hundred m eter long. It seem s that gas inclusions m igrate
from less to m ore bubbly areas (S tehle, 1 9 6 7 ). M ig ration of bubbles was also reported by M aeno
(1967). A t the B yrd S tation (Antarctica) shrinkage of bubbles with increasing depth was observed to
be faster below 800 m than expected from hydrostatic pressure; at a depth of 1100 m or m ore, no
bubbles w ere visible. After several m o n th s s to ra ge of ice cores, bubbles started to form due to
decom pression (Gow and W illiam son, 1975). M aeno (1967) also observed disappearance of air
bubbles in the ice, w hich he interpreted as due to dissolution of air m olecules into ice or due to m ass
tran s p o rta tion through diffusion of som e kind of defects in the ice crystal. B oth dissolution and
diffusion are quite probable in the open-lattice configuration of an ice crystal (M aeno and K uroiva,
1967).
The internal pressure in air bubbles trapped in ice ranges from 2 to 20 atm ospheres (S cholander
a n d N utt, 1960). Langw ay (1958) found that, in the ice from a deep core in G reenland, the in itia l
pressures of m ore than 13 atm ospheres relaxed after a year's storage, a period shorter than usually
encountered in glacier core studies. S uch high internal pressures m ay influence the diffusion of C O 2
in the ice in situ and in the cores, and facilitate the creation of clathrates (hydrates) of gases in the
intercrystalline liquid. C racking of ice occurs due to the sheeting phenom enon caused by relaxation
o f the geostress w hen the ice cores are rem oved from deep layers (for detailed discussion s e e
Jaw orow ski et al., 1992). P ressurized gases m ay then escape from the bubbles through m icrocracks.
The air b u b b le s c o llected from ice by the dry extraction technique contain half or less of C O 2
present in the ice. This low concentration, and not the content in the w hole ice sam ple, is taken into
account in the estim ates of the C O 2 level in the pre-industrial atm osphere (see Chapter 5.2).
-36A striking illustration of the effects of glacial processes on the CO 2 content in ice, recovered from
great depths is presented in Figure 5. The air and C O 2 content fluctuations in the ice are m ore or less
p a rallel betw een ~900 and 1130 m depth. B ut below this depth a dram atic decreas e in th e C O 2
concentration w as found, not accom panied by a sim ilar decrease in air content (B erner et al., 1980).
A t corresponding depths all air bubbles disappear in the ice.
The total disappearance of air bubbles w as observed by G ow and W illiam son (1975) at the B yrd
S tation (A ntarctica) in ice from depths below 1100 m eters. The authors found that air w as present
in th e b u b b le -fre e ice, as after decom pression of the ice core during storage, the secondary gas
cavities were form ed again. It is interesting to note that at Cam p Century (Greenland) and the B yrd
S tation (N eftel et al., 1982) a decrease in C O 2 content also occurred at the depth where all original
air bubbles disappear, i.e. at about 1200 m .
A t a certain depth, w hen the gas pressure exceeds the dissociation pressure, the gas is converted
to the c la th ra te h y drate (solid) form , provided that there is enough liquid H 2 O . A t the Byrd Station
(A ntarctica) this pressure is reached at a depth of 800 m (M iller, 1969). A t Vostok the predicted depth
is 400 m , however sporadic bubbles w ere observed below 800 m (K orotkevich et al., 1 9 7 8 ). The
dissociation pressure of the CO 2 clathrate is about 13 to 20 tim es, depending on tem perature, low er
than that of N 2 and O 2 , a n d s ta rts at about 5 bars below -15°C (Takenouchi and K ennedy, 1965;
M iller, 1969, 1973). This m eans that at the sam e tem perature the C O 2 clathrates will be form ed at
m uch shallow er depth than clathrates of the m ajor com ponen ts o f a ir. This phenom enon is a
plausible cause for the decrease of the C O 2 concentration in ice below 70 m eters, observed in
G reenland a n d A n ta rc tic cores by B erner et al. (1980), N eftel et al. (1982), N eftel et al. (1985),
R aynaud and B arnola (1985) and W ahlen et al. (1991).
-37-
Figure 5. C hanges of air and C O 2 content in glacier ice from C am p C entury (Greenland) versus depth
(after Berner et al., 1980). N ote the decrease in C O 2 content below ~1130 m , where air bubbles were
reported to disappear.
-38S p a c ia l fra ctionation of gases m ay occur due to selective form ation of clathrates and different
solubilities of th e d iffe re n t air gases. This is coupled w ith the m igration of liquids in the extensive
capillary network of the glaciers. After decom pression of the cores the different com ponents of air
reenter the gaseous phase at different rates, form ing new secondary air bubbles. At low tem perature
at this stage m ore C O 2 m ay rem ain in solution than other gases, due the higher retrograde solubility
of C O 2 ($74, $35, and $ 31 tim es higher solubility than that of N 2 , O 2 , and A r, respectively, on an
equal m olar basis) w hereas m ost of the other air constituents enter the secondary air bubble. The
disso ciation pressure depends on the tem perature. At greater depths in the polar ice caps, where
tem perature rises, one m ay expect release of gases from the clathrates. How ever, the clathrates
rem ain stable because the pressure increases. The gases rem aining in solids will be subject to
diffusion and chem ical processes. They m ay m igrate in the intercrystalline liquid for a tim e sufficient
to change their original com position. After release of the geostress the clathrates will dissociate in
the recovered cores, at various rates, depending on the differences between the original tem perature
and pressure in situ and during storage.
A sudden decrease in the CO 2 concentration from 250 to 180 ppm occurred in the Vostok core at
a depth of 400 m (B arnola et al., 1987). S uch a depth can be predicted for this location for conversion
of CO 2 to the clathrate form (Takenouchi and K ennedy, 1965; M iller, 1969). The CO 2 decrease w as
associated w ith drastic structural and other physical changes in the ice, occurring both in situ and in
the core (Jaw orow ski et al., 1992).
5.1.6 ICE LAYERS
The ice layers can not only change the chem ical com position of the gases in the ice, but they are
also im portant im perm eable barriers for the penetration of gases from the atm osphere into the deeper
layers of firn. In A ntarctica such im perm eable layers w ere observed and extensively studied dow n
to a depth of m ore than 100 m (Kotlyakov, 1961; Gow, 1968, W atanabe, 1977; Korotkevich et al.,
1978; Rep p , 1 9 7 8 ; N eftel et al., 1985; Raynaud and B arnola, 1985). The processes leading to
form ation of ice crusts in extrem ely cold A ntarctic ice sheets, due to absorption of solar radiation, are
described by G ow (1968) and review ed by Jaw orow ski et al. (1992). R aynaud and B arnola (1985)
stated that air in the A ntarctic firn cannot be w ell m ixed w ith the atm osphere due to the existence of
these im perm eable layers w hich isolate the air in the firn from the free atm osphere even at shallow er
depth. It is therefore astonishing that the assum ption that air bubbles (w ith the sam e concentration
of C O 2 as in 1983 atm osphere) in ice from a depth of m ore than about 70 m , deposited in the 19th
century could represent the com position of th e 1 9 8 3 a tm osphere (N eftel et al., 1985) w as so
credulously accepted. In this assum ption, w hich is based on an observation that the air pores
becam e closed at a depth of 70 m and on porosity m easurem ents (B erner et al., 1978; S chw ander
and S tauffer, 1984), the sealing effect of the im perm eable layers w as not taken into account, even
though such layers w ere observed by C O 2 students (e.g., N eftel et al., 1985) and extensively studied
-39by others. The physical and chem ical processes which m ay change the origina l c o m p o sition of
atm ospheric air along its long m igration route in porous m aterial to such a depth were also neglected.
5.1.7 EFFECTS OF DRILLING
D rilling the ice cores is an extrem ely brutal procedure, w hich leads to im portant changes in the
structure and com position of ice. A s discussed by Jaw orow ski et al. (1992) in the very m om ent of
d rillin g th e a dense cracking of ice occurs due to the sheeting phenom enon, caused by rap id
decom pression of the load pressure in the ice. In all ice cores from below a depth of 100 m eters a
dense stratification of horizonta l m a c ro - and m icro-cracks occurs. The visible, m acro-crack strata
appear at about 2 to 5 m m intervals. Through these fractures the drilling fluid (a m ixture of diesel oil
or kerosene and trichlorethylene), highly contam inated w ith heavy m etals, acids, and particulates,
pen e tra te s deeply into ice, during recovery of the cores from the boreholes. A t the surface the
m acro-cracks are soon "healed" by annealing, sealing th e c o n ta m inants inside the cores. The
"healed" fractures are clearly visible even in cores stored for m any years in freeze-room s. Extrem ely
high degree of contam ination of cores of ancient polar snow, with hundred of thousand tim es higher
elem ent concentrations than in the surface snow , w ere often reported, including cores from Cam p
C entury (G reenland), V ostok, and B yrd Station (A ntarctica), which are regarded as classic cores for
C O 2 and isotopic studies (N g and P atterson, 1981; N eftel et al., 1982; K udryashov et al., 1984 a and
b; P etit et al., 1990). C racking and contam ination can lead to dram atic changes in the chem ical and
isotopic com position of the ice. These effects are discussed in detail by Jaw orow ski et al. (1992).
5.1.8 CONCLUSIONS
The physical phenom ena discussed above, and chem ical reactions betw een C O 2 and chem ical
species dissolved in the intercrystalline liquid, m ust change the proportion of gases in the secondary
air inclusions (trapped in fractures and betw een ice crystals), a s w e ll a s in the prim ary inclusions
(gases originally dissolved in water, trapped w hen the water froze to ice), ps eudosecondary gas
inclusions (trapped in pores betw een the ice crystals, and in the "air bubbles"), and in secondary gas
cavities (form ed after the decom pression of cores), com pared to the o rig in a l atm ospheric
com position. Therefore, the concentrations of gas species (like CO 2 ) determ ined in the air bubbles
and secondary gas cavities from G reenland and Antarctic cores, e.g. from the V ostok core (Barnola
et al., 1987), cannot be regarded as representing original atm ospheric concentrations of gas species
in the ancient atm osphere.
-405.2 MEASURED CO 2 LEVELS IN GLACIER ICE
The m ajority of early C O 2 m easure m e n ts were m ade by m elting ice sam ples weighing up to
several kilogram s, from w hich C O 2 w as extracted after the ice was m elted and resulting water boiled.
Later, sm aller sam ples (usually ~300 g) w ere used, and extraction w as carried out first during a short
tim e, 7 - 15 m inutes, before ice had com pletely m elted. The second extraction was carried out during
several h ours. The first fraction was interpreted as representing CO 2 contained in the air bubbles
trapped in the ice. The second fraction was believed to represent the total content of this gas in the
ice. The "dry" extraction of C O 2 from air bubbles by shaving the ice sam ples in a closed container
w as used in the very first C O 2 studies in ice as w ell as the m icroanalysis of CO 2 in single air bubbles
in ice (C oachm an et al., 1956). S ince 1980 a m ethod consisting of crushing the ice sam ples in low
tem perature, was introduced (D elm as et al., 1980). W ith this m ethod the gas was supposed to be
extracted only from the air bubbles, and the weight of sam ples used by various authors ranged from
1 to 1400 g.
The early determ inations of CO 2 in m elted ice sam ples produced a very w ide range of results, with
peak values reaching several thousands ppm , found after several hours of extraction (i.e. in the total
volum e of gas contained in the ice structure). The short-te rm e x tra c tion (up to 15 m inutes) from
m elted sam ples, and crushing the sm all cubes of ice, produced m uch low er results (i.e. less than half
of total volum e results).
W e are aw are of only one study carried out on snow cores sealed at the m om ent of collecting. All
other cores w ere stored unprotected during up to 16 years before analysis. In the sealed snow cores
R aynaud and D elm as (1 9 7 7 ) dem onstrated that C O 2 concentration in the air trapped in the near
surface snow at very cold Vostok and Pionerskaya sites (A ntarctica) w as up to about 50% low er than
in the atm osphere (Table 2).
The first determ inations carried out at a tem perate Norw egia n g lacier showed rather high
concentrations of C O 2 in the ice, sim ilar to those found later on polar ice caps. On the Greenland ice
cap Scholander et al. (1961) found C O 2 air concentrations ranging from about 100 to 900 ppm in 20
sam ples of approxim ately 2700 years old ice (M urozum i et al. 1969). Old ice sam ples from other
localities in this region had CO 2 concentrations from a few ppm to m ore than 1500 ppm . S cholander
et al. (1961) found large differences betw een the chem ical com position of single air bubbles from the
sam e ice sam ples. This suggests that, w h e n u s ing sm all sam ples with a m ass near one gram
containing few air bubbles, the analytic uncertainty is increased.
In an A ntarctic ice core M atsuo and M iyake (1966) found a C O 2 concentration range of 280 to 2900
ppm . R aynaud and D elm as (1967) m easured C O 2 concentrations ranging betw een 1700 and 7400
ppm in a 108.5 to 1345 m deep ice core from Cam p Century, Greenland, and betw een 900 ppm and
1800 ppm in a 71 to 300 m deep core from B yrd Station, A ntarctica.
D elm as et al. (1980) found high concentrations of CO 2 in ice from two A ntarctic cores. At a depth
betw een 113 and 670 m the concentrations ranged from 210 to 740 (average 440) ppm in gas from
the m elted sam ples. These values were d e te rm in e d not in the original ice sam ples, in which the
authors found concentrations an order of m agnitude higher (1050 to 6100 ppm ), but in sam ples that
-41w ere "copiously" rinsed w ith w ater or ethylalcohol. This procedure, w hich m ight change the original
chem ical com position of the sam ples, w as used because the authors supposed that the carbonates
present in the sam ple m ight have influenced the CO 2 readings.
From the sam e ice cores Delm as et al. (1980) released about 75% of the gas by pulverization of
the frozen "dry" sam ples. In 22 sam ples from one core, at various depths betw een 37 and 893 m ,
they found C O 2 concentrations random ly ranging betw een 160 and 360 ppm . In a second core, in 13
sam ples from various depths betw een 37 and 301 m , they found concentrations ranging betw een 160
and 350 ppm . The low est values we re m e a sured in the cracked parts of the core, which "lost
im portant volum es of gases". In both cores exactly the sam e average concentration of 250 ppm C O 2
w as calculated.
V ery high concentrations of C O 2 ranging betw een about 300 and 2350 ppm were found by Neftel
et al. (1982) in the air bubbles in several hundred year old ice from a D y e-3 core in southern
G reenland. A t C am p Century in another region of G reenland, they found C O 2 concentrations ranging
from about 140 to 470 p p m in s e v e ra l hundred to 40,000 years old ice. A t the Byrd Station
(A ntarctica) in ice of sim ilar age they found C O 2 concentrations ranging betw een about 100 and 500
ppm . They s ta te d th a t only the low est values (w hich they fail to specify) "best represent the C O 2
concentrations of the o rig in a lly trapped air", and assum ed that the higher readings were due to
contam ination of the sam ples with a m ixture of diesel oil and trichlorethylene, used as a drilling fluid,
w hich penetrated the cracks in the core. B ut they did not discuss the possibility that the low readings
m ight be due to the sam e sam pling artifacts. From this description, the cores sam pled with such a
technique should not be qualified for CO 2 determ inations.
In a very cold region at the N orth C entral S tation (G reenland) N eftel et al. (1982) foun d in 2 2
sam ples from a 13-cm -long part of core from a depth of 103 m a low average C O 2 value of 271 ppm
(the detailed results and range not presented). This value is m uch low er than in ice sam ples from
another cold region in Greenland (Cam p Century), covering a period between 500 and 5000 years
ago. In these la tter sam ples CO 2 concentration ranged up to about 440 ppm , w ith the average of
about 380 ppm . H ow e v e r, the low value of 271 ppm , w hich is from a 13 cm long core, and not an
average from all determ inations in Greenland (w hich is sim ila r to , o r h igher, than the present
atm ospheric level) is often used in clim atological estim ates as representative for the pre-industrial
period (e.g. Schneider, 1989).
R aynaud and B a rn ola (1985) m easured the C O 2 concentrations in air bubbles from a core
collected at the D57 S ite in Antarctica. The ice w as sam pled from the core at various depths below
89 m . The m easured annual snow accum ulation rate at this region, upslope the sam pling site, ranged
betw een 4.4 and 45 cm of ice equivalents. The authors do not report the age of the ice, which,
however, can be calculated from the assum ed accum ulation rate values. The accum ulation rate of
45 cm per year corresponds to an ice age of about 1849 A.D . at the depth of about 89 m , and 1622
A .D . at the depth of about 197 m . The authors were aware of the occurrence of im perm eable layers,
and therefore it is not clear from the presented evidence w hy they assigned an age of 1940 A.D . for
the air bubbles in ice from a depth of 89 m , and 1560 A .D. from 197 m depth. They stated, how ever,
-42that the age of bubbles "m ay be, in fact, older by several tens of years ... b e c a u s e the air can be
isolated" (w ith im perm eable ice layers).
In the air bubbles preserved in the ice from 89 m depth Raynaud and Barnola (19 8 5 ) fo u n d an
average C O 2 concentration of 288 ppm , and in the ice from 197 m depth a concentration of 271 ppm .
They interpreted the 17 ppm v c hange as dem onstrating an anthropogenic increase of C O 2 in the
atm osphere. The authors did not account for a loss of CO 2 from the bubbles du e to diffusion in
intercrystalline liquids, and due to num erous other factors discussed in Chapter 5.1.5.
The obvious draw back of the R aynaud and B arnola (1985) paper is a lack of data from the first few
m eter depth of firn, w hich would serve as a contem porary reference. A s discussed in C hapter 5.1.3,
in the 1 m thick layer of surface firn in A ntarctica, the recorded concentrations of CO 2 w ere up to
~50% low er than in the atm ospheric air at the sam e tim e. N o statistical evaluation of the tem poral
tre n d of data w as presented. The authors com pared their results w ith the current M auna L o a
atm ospheric m easurem ents, and stated that their results "indicate clearly the increase in atmospheric
C O 2 due to burning of fossil fuels, and probably also due to the anthropogenic influence on the
biosphere over the nineteenth and tw entieth centuries". This conclusion is not c o n s istent w ith the
presented results, w hich probably represent a random fluctuation of CO 2 concentrations in the ice,
due to natural physical and chem ical processes at the snow /air interface in the atm osphere and in
glaciers. From the difference of 17 ppm betw een the m ean C O 2 concentration of 288 ppm (st. dev.
6.3) and 271 ppm (st. dev. 8.5); recorded in ice from 89 m and 197 m depth, respectively, it is not
possible to judge if this difference is significant. R aynaud and B arnola (1985) were unable to support
by any factual evidence their assum ption that som e of their air sam ples w ere from the twentieth
century.
S im ilar flaw s in estim ation of the age of air in the bubbles, neglecting the enrichm ent and depletion
processes in snow and ice, lack of analysis of near-surface layers, and not accounting for exchange
betw een C O 2 in the air bubbles and in the ice itself, can be found in the N eftel et al. (1985), P earm an
et al. (1986), B arnola et al., 1987 and W ahlen et al. (1991) papers. Neftel et al. (1985) m easured the
C O 2 content in air bubbles from an ice core collected in 1983 at the Siple Station, A ntarctica. The
accum ulation rate at that site was about 50 cm (w ater equivalent) per year. A t a depth of 7 m they
found a 2 to 10 m m thick im perm eable ice layer form ed from m elt w ater. They also noted that at a
depth betw een about 68 and 69 m tw o im perm eable layers occur. A t this depth, corresponding to an
ice age of 1891 A.D ., they found an average C O 2 concentration of 328 ppm (range not indicated), and
interpreted this as an evidence of an effect of fossil fuel burning in the 20th century.
N eftel et al. (1985) assum ed that in the ice deposited about 1891 A .D . the air w as trapped betw een
1962 an d 1983 A .D . They did not m easure the age of air. A m ong argum ents in support of this
speculative assum ption, they used a value of the C O 2 concentration in the ice from 1891 A .D . of 328
ppm v which was the sam e as m easured in the S outh P ole atm osphere in 1973/74. This seem s to
be circular logic. The coincidence sim ply indicates that in 1891 the atm ospheric C O 2 level was the
sam e as in 1974.
-43N eftel et al. (1985) concluded that around 1750 A .D . the atm ospheric concentration w as 280 ppm ,
and has increased since, because of hum an factors, by 22.5% to 345 ppm in 1984. This conclusion
is based on a speculative assum ption of the age of air trapped in the ic e , o n c irc u la r logic, and
neglecting the physical and chem ical processes in the ice sheets and artifacts in the decom pressed
and contam inated ice cores.
The S iple core is tre a te d as a classical proof that the pre-industrial CO 2 concentration in the
atm osphere was about 70 ppm v low er than now (e.g. IP C C , 1990). One should note, however, that
this core was expos e d to p o s t-coring m elting. The m elting, which m ust have caused im portant
changes in the gaseous com positio n , w a s n o t reported by N eftel et al. (1985) (see discussion in
Jaw orow ski et al., 1992). The data fro m this core were adjusted to overlay exactly the recent
atm ospheric c o ncentrations at M auna Loa (Siegenthaler and Oeschger, 1987). A figure
dem onstrating this adjustm ent is reproduced in countle s s p u b lications (e.g. in IP C C , 1990). The
overlaying was achieved by assum ing that the age of the trapped air w as 95 years younger than the
ice. W ithout this speculative assum ption the M auna Loa and S iple data do not agree at all. A s was
indicated by Jaw orow ski et al. (1992), rather than representing the past atm ospheric changes, the
results show how the CO 2 concentra tio n s in the Siple core decrease w ith the increasing load
pressure up to about 15 bars, due to clathrate form ation, diffe re n tia l dissolution of gases in the
intercrystalline liquid brine, and other processes in the ice sheet and in the ice core.
Ice cores can be easily contam inated by am bient air during field w ork (Loosli, 1983 ). This
contribution can reach 36 to 100% , as indicated by the m easurem ents of
85
K r and
39
A r, radioactive
gas isotopes produced in nuclear explosions and reactors and by cosm ic radiation. This suggests
that other am bient gases, including CO 2 , m ay contam inate the ice c o res during field work and
storage, and that there exist opened routes through w hich the gases can escape from the ice cores.
O eschger (1989) presented a photograph show ing handling of an ice core used for CO 2 studies in
the presence of persons sm oking cigarettes. This indicates that the contam ination problem s are not
seriously considered in gas analysis of ice cores. U nder such circum stances ice is exposed to higher
than am bient CO 2 air concentration, which could change the original com position of the ice core
sam ples. This possibility w as noted by A lder et al. (1969) and S tauffer et al. (1981) w ho noticed "that
ice, and especially firn, sam ples in contact with an atm osphere enriched in CO 2 get contam inated".
It is interesting to note that R aynaud and Barnola (1985) found C O 2 concentrations in ice low er
than that of the present atm osphere, w hile Neftel et al. (1985) found C O 2 concentrations in ice sim ilar
to that of the present atm os p h e re . How ever, both groups reached the sam e conclusion: that their
results dem onstrate an anthropogenic increase of atm ospheric C O 2 . This suggests that there exists
a tendency of a unidirectional interpretation of analytical results.
-44TA B LE 2. CO N C E N TR A TION S OF C O 2 IN GLA C IE R ICE (ppm by volum e).
Age of ice
Locality
P I3
S torbreen 6
Concentration
~200-2900
(N orw ay)
C oachm an et al., 1956
C oachm an et al., 1958 a
G reenland
PI
Reference
C oachm an et al., 1958 b
11 localities
~100-2500
S cholander et al., 1961
Tuto Tunnel
~100-800
S cholander et al., 1961
PI
A ntarctica
280-2900
M atsuo and M iyake, 1966
H olocene and
G reenland
2700 B P
4
W isconsin
1700-7400
A ntarctica
>100 B P
Few years B P
R aynaud and Delm as, 1977
B yrd Station
P ionerskaya
6
Few years B P
V ostok
180 B P
G reenland
PI
A ntarctica
900-1800
6
160
240
400-1200
w hole ice 5
air bubbles
210-740
6
G reenland
200-800
40,000 B P
A ntarctica
230-400
5000-30,000 B P
G reenland
200-700
4
5
6
B erner et al., 1978
D elm as et al., 1980
160-335
P I up to
3
R aynaud and Delm as, 1977
P I p re -in d u s tria l, e x a c t a g e n o t g iv e n .
Y e a rs b e fo re p re s e n t.
S a m p le s w a s h e d in w a te r o r a lc o h o l.
In g a s in c lu s io n s a fte r d ry e x tra c tio n , o th e rw is e in m e lte d ic e .
B erner et al., 1980
S tauffer et al., 1981
-45TA B LE 2 - continued.
Age of ice
Location
180 B P
G reenland
180 B P
G reenland
1000?-30,000 B P
A ntarctica
Concentration
6
Reference
250-1000
S tauffer et al., 1981
300-550
S tauffer et al., 1981
200-520
S tauffer et al., 1981
G reenland
D ye-3 6
>100 B P
300-2350
N eftel et al., 1982
600-40,000 B P
C am p Century
6
140-470
N eftel et al., 1982
~680 B P
N orth Central 6,7
271
N eftel et al., 1982
100-500
N eftel et al., 1982
135-275
S tauffer et al., 1984
247-295
R aynaud and B arnola, 1985
279-328
N eftel et al., 1985
278-281
N eftel et al., 1985
268-326
P earm an et al., 1986
191-296
B arnola et al., 1987
A ntarctica
B yrd Station 6
600-40,000 B P
G reenland
30,000-40,000 B P
D ye-3 6
A ntarctica
D 57 6
>~300 BP
A ntarctica
S iple 6
1663-1891 A D
110-820 B P
S outh Pole
6
A ntarctica
Law D om e 6
1530-1900 A D
A ntarctica
4050-163,670 B P
6
V ostok
G reenland
1310-1720 A D
G IS P 2
W ahlen et al., 1991
7
In o n e 1 1 c m th ic k la y e r a t 1 0 3 m d e p th .
272-310
-46-
F ig u re 6 . E rrors of C O 2 m easurem ent in air bubbles in ice cores from Greenland and A ntarctic a
(m odified after Oeschge r e t al., 1985) and claim ed reported increases in CO 2 levels in recent ice
sam ples due to fossil fuel burning (A : P earm an et al., 1986; B : R aynaud and B arnola, 1985; C : N eftel
et al., 1985). S olid line indicates ±1 stan d a rd d eviation (assum ed to include ~68% of the
observations), broken line indic a tes ±2 standard deviations (assum ed to include ~95% of the
observations).
-47A n exam ple of an unusual selection of results is a paper by P earm an et al. (1986). The authors
collected 74 ice sam ples from an A ntarctic ice core, from w hich they rejected 32, because the C O 2
con c e n tra tio n s d iffered by m ore than 7 ppm from "at least one other sam ple of com parable age".
They stated that betw een 1850 and 1900 an increase in CO 2 concentration of 13 ppm occurred, and
that this corresponds to a n e t g lo b a l C O 2 input of 48 GT, which cannot be explained by the sm all
am ount of fossil fuel consum ed during this period.
Finally, one should com pare the range of analytical uncertainty with the tem poral differences in
C O 2 concentration in ice used recently to dem onstrate the atm ospheric increase of this gas. S uch
tem poral difference in the case of R aynaud and Barnola (1985) was 17 ppm , in the case of Neftel et
al. (1985) 49 ppm , and in the case of Pearm an et al. (1986) 13 ppm . According to Oeschger et al.
(1985) the "errors" (at an assum ed 68% probability level) of the m easurem ents of CO 2 in air trapped
in ice cores from Greenland and A ntarctica range betw een 11 a n d 2 4 p p m . A t an assum ed 95%
probability level the "errors" reach about 47 ppm . It is clear from these considerations that the claim s
of a recent increase in atm ospheric CO 2 content, found in the glacier gas inclusions, is based on
rather spurious evidence. The "errors" of m easurem ents at a 95% probability level are larger than or
close to the claim ed rise in atm ospheric CO 2 (see Figure 6).
C lassical exam ples of C O 2 m easurem ents in the G reenland and A ntarctic ice are given in Figures
7 and 8. These m easurem ents show typical long-term (6000 to 30,000 years B .P .) random variations
betw een 200 and 700 ppm (Figure 7), and short-term random variations in a 90 cm long core of 180
years old ice, of betw een ~280 and 1000 ppm (Figure 8). It w ould be difficult to believe that these
variations represent the real changes in C O 2 levels in the atm ospheric air. R ather, they resulted from
the physical and chem ical processes in the glacier ice, or are due to artifacts during sam pling and
analyses of the ice. A s m ay b e s e e n in F ig ures 5 and 6, after a 15 m inutes extraction the
concentration of C O 2 recovered from the m elted ice w as m uch low er than after a 7 hour extraction.
N 2 , O 2 and A r do not reveal such behavior. Only ~50% of the CO 2 is extracted after the m elting
process is term inated. M any hours a re needed in order to m ake the extraction com plete. Dry
extraction by crushing the frozen sam ples releases the air contained in bubbles but does not allow
m easurem ents of the total CO 2 content (S tauffer et al., 1981).
S tauffer et al. (1981) rather incorrectly assum ed that, in spite of these difficulties, the C O 2
concentrations m easured in the air released from the ice, after 15 m inutes extraction, represent the
atm ospheric content w ith an accuracy of 22 ppm . W ith this inappropriate m ethod glaciological studies
provided the first confirm ation of the Callendar (1958) value of 292 ppm C O 2 level for the 19th century
atm osphere. H ow ever, the "w et extraction" studies also provided valuable inform ation. It was
dem onstrated that the ice itself is not C O 2 -free. D epending on conditions c h anging both in the ice
sheets and in the cores, ice rich in CO 2 m ay be either a source or a sink for the CO 2 trapped in the
air bubbles.
The concentrations of CO 2 in air recovered from ice with the dry gas e xtraction m ethod ranged
between 140 and 2350 ppm (Table 2). A s was dem onstrated by Jaw orow ski et al. (1992) these data
represent the effects of about tw enty physical and chem ical processes occurring in the ice sheets
-48a n d in the ice cores, rather than the original atm ospheric concentrations. The value of 280 p p m ,
w id e ly accepted on the basis of glacier studies as the pre-industrial atm ospheric CO 2 level,
apparently results from such processes, invalid assum ptions and arbitrary rejection of high readings.
W e are not discussing the validity of analytical m ethods u s e d in the current CO 2 studies in
glaciers. W e criticize the quality of ice as a closed system , which is an absolutely essential criterion
for its use to reconstruct the com pos ition of the pre-industrial atm osphere. W e also criticize the
m ethodology of sam pling, and biased interpretation of results. Ice is neither a rigid m aterial, nor m ay
be regarded as a closed system , suitable for preserving the original chem ical a n d is o to p ic
com position of atm ospheric gas inclusions. Ice core drilling is an extrem ely brutal procedure leading
to drastic changes in the ice sam ples, precluding their reliability for gas analyses. The concentrations
reported from the pre-industrial ice, no m atter whether they were higher, equal to, or low er than the
present atm osphere, were alw ays interpreted as indicating an anthropogenic increase of atm ospheric
C O 2 . In this interpretation differentiation processes in the ice w ere neglected or downplayed. In fact
they were never thoroughly studied, and this is one of the reasons w hy the glaciological studies w ere
not able to provide a reliable reconstruction of the C O 2 level in pre-industrial an d a n c ient
atm ospheres.
-49-
Figure 7. C oncentrations of C O 2 in a ir from the C am p Century (G reenland) and the Byrd Station
(A ntarctica) ice cores. W hite dots represent CO 2 concentrations after 15 m inutes extraction, black
dots after 7 hour extraction. A fter S tauffer et al., 1981.
-50-
Figure 8. Concentration of C O 2 in a 90 cm long piece of the C am p C entury (G reenland) ice core. The
low er curve represents the values after 15 m inutes extraction from m elted ice, and the results after
dry extraction. The upper curve show s the C O 2 content after 7 hours extraction. A fter Stauffer et al.,
1981.
-515.3. CONCLUSIONS
(1)
The results of C O 2 m easurem ents in air bubbles in pre-industrial ice ranged betw een 140
and 550 ppm . In ice from the 19th century and before, these concentrations were found
to be either the sam e, or considerably higher or low er than the present atm ospheric levels.
The "errors" of m easurem ents at a 95% probability le v e l a re larger than or close to the
claim ed rise in atm ospheric C O 2 . H e n c e a rise, fall, or steady level of C O 2 in the
atm osphere cannot be established from the available CO 2 gas inclusion data. The value
of about 290 ppm , often accepted as representing the average global concentration of C O 2
in the pre-industrial atm osphere, is highly uncertain and speculative.
(2)
E ven the recent snow deposits do not represent the original C O 2 level in the atm osphere.
A ir trapped in near s u rfa c e A n tarctic snow has about 50% low er C O 2 content than the
am bient atm osphere.
(3)
C hem ical and p h ysical processes during the early atm ospheric history of the snow
crystals, and in the glacier firn and ice, can both decrease or increase the C O 2 content in
air trapped in the ice, relative to the original atm ospheric concentration. P robably the m ost
im portant cause of these c h a n g e s is the presence of liquid in snow and ice, even at
extrem ely low tem peratures. Because of its retrograde solubility CO 2 easily dissolves in
cold w ater and in concentrated acid brines from an intercrystalline capillary netw ork. C O 2
m ay form solid clathrate. C O 2 m ay also m igrate from the air bubbles into other gaseous
inclusions in the solid ice. Other air com ponents have different solubilities in w ater than
C O 2 , and different clathrate dissociation pressures. This leads to changes in the chem ical
and isotopic com position of gases trapped in ice.
(4)
M oreover the sam pling and analytical m ethods used for C O 2 determ ination in glacier ice
m ay change the original content of this gas in the sam ples, due to incom plete release of
the gas, loss of C O 2 , or due to contam ination.
(5)
The establishm ent of the true pre -industrial level of C O 2 in the atm osphere is of great
im portance. H owever, the current polar ice sam pling techniques are not able to provide
a reliable m aterial for reconstruction of the com position of the pre-industrial atm osphere.
N ew m ethods for in situ air sam pling fro m p o la r ice sheets should be developed, as
suggested by Jaw orow ski et al. (1992).
-526. HYDROGEN AND OXYGEN ISOTOPES IN GLACIERS
H ydrogen and oxygen stable isotope determ inations have p ro v id e d useful results in the field of
glac io lo g y (e.g. see H oefs, 1980, and A rnason, 1981, for references). For exam ple, Epstein et al.
(1965) used hydrogen and oxygen isotopes to find an average annual accum ulation rate of 7 cm of
w ater at the South Pole betw een 1958 and 1963, in good agree m ent with stratigraphic and
radioactive dating results. The δD and δ18 O values are used for reconstruction of paleotem peratures
and to date the snow and ice layers as a function of depth.
H ow ever, the isotopic com position of the sno w c a n be changed by num erous factors, including
snow drift (A rnason, 19 8 1 ). T h e isotopic ratios for the light isotopes in snow and ice are changed
considerably every tim e there is a ph ase change (betw een vapor/liquid/solid), especially at low
tem peratures, as a function of m ass ratios involved in the phase change reaction (e.g. Hoefs, 1980).
K inetic isotope effe c ts, com m on at low tem peratures, m ay further change the original isotopic
com positio n of precipitation, and lead to errors in calculations based only on equilibrium isotope
effects.
The short-term fluctuations of δD and δ18 O have been ascribed to annual sum m er/w inter layering
and used for dating the ice and snow sam ples. These short-term peaks are gradually elim inated in
old ice (Langw ay, 1967) and in upper parts of the ice sheets (Grootes et al., 1990). This is due to
hom ogenization processes, such as sublim ation, m elting and refreezing of w ater percolating through
snow , firn and ice, m ovem ent of water vapor along the tem pera tu re g ra d ients, and plastic
deform ations (Faure, 1 9 8 6). Thus, detailed com parisons betw een the isotope records of the ice
sheets in Greenland and A ntarctica are unreliable (H oefs, 1980). A rnason (1981) gives exam ples of
isotope variations in ice cores from the B yrd S tation in A ntarctica w here it w as im possible to interpret
successive sum m er and w inter layers. The isotopic variation w as too irregular to be acceptable for
paleoclim atic use. S till, the hydrogen and oxyg e n is o to p e values have been used as record of
clim atic conditions in term s of calculated m ean air tem peratures, like for the Vostok ice core (Jouzel
et al., 1987; Barnola et al., 1987).
For the Vostok ice core Jouzel et al. (1987) correlated the recent annual averages of the surface
tem peratu re w ith th e recent snow hydrogen isotope ratio, using a correction m odel for the kinetic
isotope effect. This relation was then used to calculate tem perature deviations from the present
surface tem perature (-55.5EC ) for the w hole 2083 m ice core covering 160,000 years. The authors
m ust assum e that the precipitation rate was constant throughout this long tim e span covering several
glacial and interglacial stages. This assum ption is highly questionable. Nor is it likely that the isotopic
record has been unaffected by phase changes in the firn and brine-ice interactions in firn and ice (see
C hapter 5) that alter the original light isotope ratios.
The attenuation and obliteration of the isotopic signature indicate that both firn and ice in the polar
ice sheets do not stand up to the closed system criterion, absolutely essential for using this signature
for paleoclim atic estim ates. D istinct fluctuations of δ18 O on tim e scales of several years to m any
-53decades, observed in m any records, do not have necessarily clim atic significance, and m ay be due
to factors other than air tem perature (W atanabe, 1978; Satow and W atanabe, 1987; Grootes, 1990).
-547. LAG BETWEEN CO 2 LEVELS AND TEMPERATURE CHANGES
T h e slightly increasing m ean tem perature of the Northern H em isphere surface air, w hich w a s
observed from the start of this century to the 1930s, took place prior to the m ajor "greenhouse gas"
em issions (M ichaels, 1990). In the M auna Loa record of CO 2 atm ospheric concentration (K eeling et
al., 1989) the peak values lag behind the global continental tem perature m axim a by abou t five
m onths (K uo et al., 1990). K uo et al. (1990) noted that changing tem peratures lead to changes in the
am ount of C O 2 outgassing or dissolution in the ocean, and to variation in biological activity, and thus
to CO 2 level changes in the atm osphere. The five m onth lag tim e of CO 2 changes behind tem perature
changes, indicates that the causality is: tem perature-to-C O 2 . If heat trapped by C O 2 w as the cause,
the opposite sequence of m axim a should be observed. B arnett (1990) suggested that the interrelation
betw een the tw o variables (i.e. C O 2 and tem perature) m ay be due to tem perature changes arising
from natural clim ate variability, and has nothing to do with an increasing "greenhouse effect". B arnola
et al. (1987) found that in the V ostok ice core, reportedly covering the past 160,000 years, "the C O 2
change [w as] clearly lagging behind the δD decrease", δD being interpreted as a paleo-tem perature
indicator. H ow ever, as discussed in Jaw orow ski et al. (1992) both the CO 2 and stable isotope data
from this core are affected by the processes in the ice sheet and artifacts in the highly contam inated
ice core, and therefore do not represent the original paleo-com position of the atm osphere and of the
precipitation.
O c e a n water is a large sink for C O 2 . B ecause of CO 2 's retrograde solubility in water, increased
tem perature causes degassing of CO 2 from th e o c e a n to the atm osphere (Segalstad, 1990). This
"therm al solubility pum p" accounts for 70% of the CO 2 flux from the ocean to the atm osphere, while
"the biologic pum p" accounts for the rem aining 30% (Volk and Liu, 1988). It has been estim ated that
4000 G T of CO 2 (equivalent to 1000 GT of carbon) is fluxed from the ocean via the atm osphere to the
continental biosphere during the transition from a glacial to an interglacial stage (Faure, 1990), owing
to these tw o "C O 2 pum ps" only. Other natura l c a rb o n fluxes (w eathering, volcanism , hot springs,
carbonate sedim entation, degassing by m etam orphosis of rocks, etc.) are not included in this figure.
-558. GLOBAL TEMPERATURE RECORDS
A s indicated before, it w as assum ed that since the m iddle of the 19th century the concentration of
C O 2 in the atm osphere has increased by about 25% (IP C C , 1990). A ccording to the General
C irculation M odels (GC M s) this should have increased the glo b a l m e a n air tem perature by up to
about 2.5EC today. These m odels have predicted the strongest warm ing, 8 to 10°C , to occur in the
A rctic. The surface tem pera tu re m easurem ents do not confirm these predictions. Stratospheric
tem peratures, w hich according to clim atic m odels should decrease in a "greenhouse-gas" enriched
atm osphere, did not show a statistically significant decline since 1960 (Angell, 1986).
A ttem pts to find a w arm ing trend signal in the long tem perature records are, to a large extent,
based on inspec tio n o f s m oothed curves ("running m eans") and not on m ore reliable tim e series
analyses. The way in which sm oothing procedures m ay totally change the statistical properties of the
original tim e series, and lead to unrealistic conclusions, has been treated by several authors (see e.g.
H isdal, 1956). V ariations with long periods are em phasized in a fascinating m anner, and short period
"noise" is suppressed, the final result depending on the type of sm oothin g filte r applied. E ven
com pletely random series m ay, on the basis of visual inspection, show convincing "clim atic changes"
after having been beautified by such linear operations (cf. Haavelm o, 1951).
H ansen and Le bedeff (1987, 1988) found that the average global land tem perature increased
about 0.7EC during the past one hundred years. Other studies suggested that the global land and
sea-surface tem peratures rose by about 0.5EC (see review in M onastersky, 1989). Hansen (1988)
stated in front of the U .S . H ouse of R epresentatives that he had "99% confidence" in the reality of
the global warm ing trend, and in the cause/effect relationship between a 0.7EC global warm ing and
anthropogenerated "greenhouse" alterations. This statem ent w as criticized on theoretical grounds
(see e.g. review in K err, 1989) and it w as also found that the uncertainty of the estim ated past global
tem peratures is about the size of the w arm ing signal (B arnett, 1989 - after M onastersky, 1989). The
corrections applied on the global data set are of the sam e m agnitude or larger than the warm ing trend
they are claim ed to show , and the results seem very doubtful (Frydendahl, 1989).
A recent study of the N A S A clim ate records in the U nited S tates show ed a w arm ing of 0.4EC in the
tw entieth century. H ow ever, this tem perature increase could be influenced by an urbanization effect
on m easurem ents taken at m eteorological stations s itu a ted in or near the cities (K arl and Jones,
1989). A lso, it w as pointed out that a notable part of the increase w as due to an error of the com puter
program w hen N A S A supplied data to NO A A for analysis (Karl - after M ichaels, 1990).
H anson et al. (1989) found that the records from the 48 contiguous U nited S tates do not indicate
any statis tic ally significant evidence of an overall increase in annual tem peratures or change in
annual precipitation betw een 1895 and 1987.
A new study of the w orldw ide ocean tem peratures since 1850, carried o u t b y a group from M IT
shows little or no global w arm ing over the past century (N ew ell et al., 1989). The authors found that
-56th e a v e ra g e o c e a n surface tem perature is now virtually the sam e as it w as in the 1940s. A s tw o
thirds of the buildup of C O 2 have taken place since 1940, the M IT data do not support the General
C irculation M odel calculation results.
R eynolds et al. (1989), analyzing surface and satellite m easurem ents, concluded that there is no
evidence of any w arm ing trend in the ocean surface w ater betw een January 1982 and June 1988.
This finding refuted an earlier claim by S trong (1989) that the sea surface tem peratures increased
in this period by as m uch as 0.1EC per year.
The variability of natural clim atic fluctuations, and hum an influence on tem peratures (urban and
"heat island" effects) m ake it difficult to o b tain hom ogeneous tim e series of annual global
tem peratures (Landsberg , 1 9 74; Grovem an & Landsberg, 1979). The tem perature variations over
continents are, on the average, 4-6 tim es larger than over oceans during the winter, and 2-3 tim es
larger on a yearly basis. The area ratio continents/oceans betw een 15 and 70E N orth is 0.88, but only
0.14 betw een 0 and 70E S outh. For an area w eighed arithm etic m ean calculated for the Northern and
S outhern Hem isphere, the tem peratures m easured o ver continents w ill therefore dom inate the
N o rthern H em isphere m ean. A m ajor problem lies in m aking the w eighed global annua l m e a n
tem peratures representative w hen m ore tem perature data are available over continents than over
oceans. This problem has still not been satisfactorily solved.
Long, relatively reliable series exist for the Northern Hem isphere from E urope, N orth Am erica,
S oviet Union, and Japan. These data w ere used by B orzenkova et al. (1976), and by G rovem an and
L a n d s berg
(1979)
for calculating w eighed annual average tem peratures for the N orth e rn
H em isphere (Figure 9). In order to extend the tim e series back in tim e (between 300 - 400 years B .P .)
G rovem an and Landsberg used records of tree ring thickness from Finland and A laska, which had
correlated well w ith later instrum ental tem perature records. The older part of their tim e series is the
least reliable one.
A spectral analysis of the about 400 years long tim e series reveals that alm ost half of the variance
can be explained by a periodic variation w ith a w avelength of 99 years (G rovem an and Landsberg,
1979). A plausible explanation of this periodic variation is cataclysm ic volcanic eruptions occurring
at about this tim e interval, e.g. Laki (Iceland) 1783-1785 and Krakatau (Indonesia) 188 3 . These
eruptions released sulfurdioxide explosively to the stratosphere, w here the S O 2 com bined w ith water
to sulfuric acid, effectively blocking a part of the incom ing solar radiation (Segalstad, 1983; Palais
and S igurdsson, 1989). It has been calcula te d fro m geologic evidence that Laki's 1783 eruption
caused the form ation of som e 90 m illion tons of sulfuric acid in the stratosphere, w hile the 1883
K rakatau eruption contributed som e 30 m illion tons. Other volcanic eruptions also contribute steadily
to the atm ospheric input of sulfur. For exam ple, the sm all 1981 e ru p tion of the Krafla volcano
(Iceland) contributed to the form ation of som e 100,000 tons of sulfuric acid (P alais and S igurdsson,
1989).
The irregular influence of volcanoes on the E arth's c lim a te , m akes it necessary to look at tim e
s e ries w hich are several tim es longer than the intervals betw een the eruptions and long e r th a n
about 100 years. For a discussion on tim e spans and resolution of various clim atic records see e.g.
W ebb (1985).
-57It is interesting to com pare the nearly 400 years long series with a shorter portion from the years
1850-1975. Both data sets show nearly the sam e arithm etic m eans and standard deviations (both
norm ally distributed). A dapting a linear trend to the data, the long tim e serie s s h o w s a lm ost no
change at all (+0.076EC per 100 years), while the short series (1850 - 1975) gives a sm all increase
(+0.33EC per 100 years). It is im portant to note, how ever, that the calculated trends "explain" only
about 15% of the variances in the data sets.
A careful statis tical analysis of the long-term series of tem perature (1869-1985) from widely
distributed European stations did not dem ons tra te c o n v in cingly that there has been a C O 2 effect
upon the E uropean clim ate so far (C oops, 1991).
The rescaled range m ethods of M andelbrot and W allis (1968, 1969 a, 1969 b) are appropriate for
exam ination of certain long geophysical records. They were used, for exam ple, for studying the
m any thousand year long record of water level, flooding and drought of the river Nile. S uch records
have in m ost cases been found to be "fractal noise", and statistics based on random series cannot
be used. This m ethod tests the p e rs is te n c e te ndency of the data (i.e. if high or low levels tend to
occur over a long tim e). The m ethod c a n a ls o put constraints on the likelihood of getting extrem e
values in the future.
U sing this procedure, Frøyland (1992) found for the 400 years long tem perature series (Figure 9)
a strong tendency for the tem perature to rem ain within the sam e range for tim es m uch longer than
w hat would be expected if there w as statistical independence in the series. This m eans that there
exists a strong persistence tendency in the series (a warm year is likely to be follow ed by another
w arm year, a w arm decade is likely to be follow ed by another warm decade, etc.).
B ased on exam ination of shorter (100-150 years) series m any researchers have concluded that
a recent rise in tem peratures is caused by only one, anthropogenic, cause and projected this signal
into the future to give 3.5 to 5.0EC higher annual global tem peratures 60 years from now .
B y inspection of the long tim e series (Fig. 7) w e see three other sim ilar tem perature rises (starting
at about 1835, 1675, 1605). If it is true that the anthropogenic influence by deforestation, agriculture
and burning of fossil fuels sta rte d a b o u t 1750, we should be able to see this as a signal in the
tem perature data over the last 200 years, m aking these data noticeably different from those of the
foregoing 200 years. This is not observed. Even a claim ed recent rise in th e 1 9 8 0 s (not show n in
Figure 9) is less dram atic than the rise in tem peratures from about 1810 to about 1830.
U sing the tem perature series com piled by Jones et a l. (cf. Jones, 1988) for both hem ispheres
D em arée (1990) found by statistical tests that the w arm ing occurring around 1920 in the N orthern
H em isphere, and som ew hat later in the S outhern H em isphere, had the character of a com paratively
abrupt change of th e m ean tem perature level. Such "clim atic jum ps" have been pointed out by
several authors also for other periods and for other regions (see e.g. Yam am oto e t a l., 1987). It
seem s unrealistic to ascribe this phenom enon to a gradually increasing CO 2 concentration. It m ay
be m ore plausibly explained on the basis of the non-determ inistic theory of clim atic change proposed
by Lorenz (1976), w ithout assum ing any variation of external o r boundary forcings for the clim atic
system .
-58W e conclude that th e n ature of the tem perature series is such that no signal of hum an-induced
global warm ing is significantly traced in the best available tem p e ra tu re series for the Northern
H em isphere. This agrees with the results of investigation of the currently available data bases for the
global land and sea surface tem peratures carried out by Frydendahl (1989). He concluded that the
uncorrected data indicate rather cooling than w arm ing of the oceanic surface waters and of surface
oceanic air during the last 100 years, and that there exists no unequivocal inform ation which could
prove that the warm ing of the globe occurred during the past 130 years. High recent tem peratures
are not significantly different from earlier sim ilar excursions (Fig. 7), and are not of a m agnitude or
duration (part of a 200 year long increasing trend) th a t is unique or alarm ing com pared to earlier
tem perature fluctuations. R ecent high tem peratures can be explained on the basis of the statistical
properties of the series, according to dynam ic system s theory (Frøyland, 1992).
The m ean 3.3% annual rise in the release of CO 2 from burning of fossil fuels (E lliott et al., 1985)
since about 1750 should give a statistically significant signal in the tem perature, if the current m odels
for the global carbon cycle and the "greenhouse effect" are correct. S uch a signal is not observed.
A s stated before , th e 400 years tem perature record represents a series with a strong persistence
tendency. This record (w hich includes m ore th a n 200 years of increased C O 2 em ission from
m a n -m ade sources) m ay be characterized as resem bling "fractal noise". Its statistical analysis
indicates that processes that w ere operating in this period are unlikely to produce 3.5 to 5.0°C global
heating, regardless of their cause (Frøyland - private com m unication).
A nalysis of the first 10 years (1979 to 1988) of satellite m easure m e n ts o f low er atm ospheric
tem perature changes reveals no heating or cooling tendency for the 10-year period (S pencer an d
C hristy, 1990).
-59-
Figure 9. Lower curve: N orthern Hem isphere annual tem peratures (data from B orzenkova et al., 1976
and G rovem an and Landsberg, 1979). U pper curve: Volcanic D ust V eil Index, D .V .I. (data from Lam b
1970, 1978), w ith increasing values dow nw ards.
-609. AIR TEMPERATURES AND GLACIERS AT HIGH LATITUDES
B ecause of the com bined effect of the "greenhouse gases", w e should, according to M icha e ls
(1990), have effectively gone m ore than half w ay to a doub lin g of C O 2 . A ccording to General
C irculation M odels the strongest greenhouse w arm ing should be expected in high latitudes (in the
A rctic about 2EC in sum m er and 8-14EC in winter w ith CO 2 doubling), causing a decrease in the m ass
of polar ice, and rise in sea level w ith catastrophic consequences. Observations, however, do not
support these predictions.
Tem perature series from the Scandinavian peninsula, D enm ark and Greenland for the last 120-130
years give no evidence for a positive trend during this period (A une, 1989; Frydendahl, 1989). These
records show that in this region in our century there was a general increasing tem perature tendency
up to the 1940s, follow ed by a slight decrease tow ards the end of the series. Long-term tem perature
rec o rd s in Svalbard started in 1912 at Green H arbour. D uring the first years the sm oothed m ean
tem perature rose considerably until the 1920s, particularly during the winter season (B irkeland, 1930;
H esselberg and B irkeland, 1940). S ince then there have been several "nice" waves in the S valbard
tem perature series (Figure 10), as in all sm oothed series, but no definite sign of an in c reasing
"greenhouse" warm ing, that would justify a statistical analysis (Hanssen-B auer et al., 1990).
Five studies of other A rctic tem perature records presente d by M ichaels (1990) show sim ilar
variations as those found in the Norw egian A rctic. O n S em lya Frantsa Iosifa a decrease of average
annual tem perature of m ore than 4°C w as observed between 1950 and 1965 (Frydendahl, 1989), in
agreem ent w ith the trend recorded in Svalbard.
There is no sim ple lin k b e tw e en glacier m ass balance and air tem perature. In m any regions an
increase of glaciers is connected with m ild winters and increased precipitation, and w ith cool
sum m ers. This is opposite to the popular concept on th e re la tionship betw een glaciers and
tem perature. N evertheless glacier variations are an im portant issue in the current discussion on the
"greenhouse" warm ing and sea level changes.
Lefauconnier and Hagen (199 0 ) a n d H agen and Liestøl (1990) found that the glaciers
B røggerbreen and Lovenbreen in S valbard show ed recent deceleration of loss of ice m ass. This is
in agreem ent with the study of K oerner et al. (1989) who have determ ined the ice m ass balance for
the past 10 to 30 years at four ice caps in the C anadian A rctic, and found no indication of increasing
ablation. In the C anadian ice cap K oerner et al. (1989) found that the conditions during the last years
constitute a m arked contrast to the heavy m elting years that characterized the warm period from the
1960s to the early 1980s. D uring the second half of the 1980s, there w as an increase in the ice m ass
at the M elville S outh and M eighen ice caps. S ince about 1968, the advance of sm all glaciers in W est
G reenland coincided w ith a period of decreas in g s u m m e r tem peratures, and could be seen as a
direct response to this clim atic deterioration. S ix out of the nine glaciers studied continued to advance
at least until 1978 (G ordon, 1980). A fter 1976 a change from negative into positive m ass balance w as
observed at W olverine Glacier in Alaska (M ayo and M arch, 1990). In the S w iss A lps about ten tim es
m ore glaciers w ere advancing in the 1980s than around 1900 (Lam b, 1988)
H im alayan glaciers appear to be advancing (Benxing et al., 1984).
Also individual
-61-
Figure 10. Five-year m oving average tem perature for January-M arch for the stations Isfjord Radio
(78E04'N , 13E38'E ) and H open (76E30'N , 25E04'E ).
-62The new satellite s u rv e y s in dicate that both the ice caps of Greenland and A ntarctica are now
increasing, corresponding to a sea level low ering of 0.45 and 0.75 m m per year, respectively (M eier,
1990). A t about 1950 the greater part of the Inland Ice m a rg in in S outh and W est Greenland w as
receding and thinning. By about 1985 m any previously receding parts of the inland ice m argin were
in a state of advance. A t present the zone of advance is spreading from the highland to the low land
regio ns (W eidick, 1991, and references therein). Independent evidence for recent advance and
thickening of the Inland Ice in Greenland c o m e s from geodetic (Seckel, 1977) and from satellite
altim etry (Zw ally, 1989).
Zw ally (1990) and Zwally et al. (1989) found that the southern 40% of the Greenland ice sheet w as
increasing vertically by about 23 cm per year betw een 1978 and 1986. This indicates a 25 to 45%
excess in ice accum ulation over the am ount required to balance the outw ard ice flow . The
corresponding sea-level decrease is 0.2 to 0.4 m m per year. A n increase in snow accum ulation in
A ntarctica, corresponding to a sea level low ering of 1.0 to 1.2 m m per year w as also reported by
M organ et al. (1991). H ence the therm al expansion of the sea water during a w arm ing period can
thus be counterbalanced. Of course, during cooling periods the w ater of the oceans w ill tend to show
therm al contraction.
The above discussion supports the opinion (W igley et al., 1989) that the anthropogenic increasing
"greenhouse" warm ing signal has not yet been detected in a rigorous way.
-6310. CONCLUDING REMARKS
Ú
A s in d icated by the m easurem ents of radiocarbon ( 14 C ) from nuclear test explosions th e
residence tim e of CO 2 in the atm osphere is only 5 years. Assigning a longer atm ospheric
residence tim e (up to 200 years) for m an-m ade C O 2 was m ade to fit the pre-conceived idea
that the 19th century C O 2 atm ospheric level was 25% low er than now. The value of about
280 ppm v, w idely accepted from glacier studies as the pre-industrial atm ospheric C O 2 level
results from invalid assum ptions, processes in ice sheets and artifacts in ice cores.
Ú
The atm ospheric C O 2 is constantly changing and adjusting its concentration according to
the natural changes in the E arth's tem perature. These changes are governed by inorganic
therm odynam ic gaseous, aqueous and m in e ral equilibria, and by biologic processes.
A nthropogenic C O 2 is negligibly sm all com pared to the gigantic natural reservoirs and fluxes
of natural C O 2 .
Ú
C O 2 has a high solubility in w ater. Low er aqueous solubility of CO 2 at hig her tem perature
cause the oceans to degas C O 2 to the atm osphere when the sea and air tem perature rises
as a result of natural clim ate change. This leads to an increase of CO 2 concentration in the
atm osphere, w hich in this case is an effect and not a cause of clim a tic c h a n g e. The
atm ospheric C O 2 increases in the 20th century w ere found to lag behind increases of the
surface air tem pe ra tu re . A w arm ing effect of the increased atm ospheric C O 2 m ay be
counterbalanced by pow erful negative feedback m echanism s, such as increased
cloudiness.
Ú
R adiocarbon (14 C ) studies show that the upper ocean turn over the dissolved atm ospheric
C O 2 in a short tim e (a few decades), thus effectively elim inating the bulk of the m an-m ade
C O 2.
Ú
The exploitable total fossil fuel carbon reservoir is only 11 tim es larger than the atm ospheric
carbon reservoir. A bout fifty parts of CO 2 are dissolved in the oceans for each part released
to the atm osphere. Therefore, a perm anent doubling of the atm ospheric C O 2 cann ot be
realized from burning of fossil fuels only, all other things being held constant.
Ú
There are several m ethodological uncertainties in the analysis and data processing of the
current atm ospheric C O 2 concentration. Further investigations of these p ro b le m s seem
necessary.
-64Ú
The estim ate of "pre-industrial" atm ospheric CO 2 concentration cannot be based on studying
gas inclusions in ice cores, or carbon isotopes in tree ring s , using the current analytical
techniques. The sam pling and analytical m ethods used give results that have an uncertainty
larger than the claim ed variations.
Ú
The presence of liquid water in glacier ice at low tem peratures, and the physical and
chem ical proces s e s in v o lved are likely to m ake ice core results unrepresentative of the
original chem ical and isotopic com position of the ancient atm osphere.
Ú
Tem perature estim ates from hydrogen and oxygen isotopes in glacier ice are highly
uncertain, and in m ost cases m eaningless because ice-brine interactions took place in the
glacier ice, and because of partial transitions betw een solid, liquid, and gas states of H 2 O ,
taking place in the ice sheets and in the ice cores.
Ú
C lear indications of atm ospheric heating caused by anthropogenic releases of CO 2 or other
"greenhouse" gases have not been found in the tem perature records.
Ú
For several arctic and tem perate gla c ie rs a reduction of the negative m ass balance w as
found. B oth in Greenland and A ntarctica an increase of ice m ass was recently observed,
w hich indicates that an im m inent rise in sea level is not likely. A claim ed therm al expansion
of sea w ater during the last w arm ing period m ay thus be counterbalanced.
-6511. REFERENCES
A belson, P.H ., 1990. Uncertainties about global warm ing [editorial]. Science, 247: 1529.
A lder, A ., J. Geis s and A . Renaud, 1969. Gas com position in ice sam ples collected by E .G.I.G . in
G reenland. Medd. Grønland, 177 (2): 100-115.
A ngell, J.K ., 1986. A nnual and seasonal global tem pera tu re c h a n g e s in the troposphere and low
stratosphere. Mon. Weather Rev., 107: 1922-1930.
A nonym ous, 1992. Greenhouse gas tax. Science, 255: 154.
A rnason, B. 1981. Ice and snow hydrology. In: Gat, J.R . and R . Gonfiantini (E ds.): S table isotope
hydrology. Deuterium and oxygen-18 in the w ater cycle. Internat. Atomic Energy Agency, Tech.
Report Series, 210: 143-175.
A une, B., 1989. Lufttem peratur og nedbør i Norge. DNMI Rapport Nr. 26/89: 68 pp.
B acastow , R.B ., C .D . K eeling and T.P . W horp, 1985. S easonal am plitude increase in atm ospheric
C O 2 concentration at M auna Loa, Haw aii, 1959-1982. J. Geophys. Res. 90 (D6): 10529-10540.
B akker, R .J. and J.B .H . Jansen, 1990. P referential water leakage from fluid inclusions by m eans of
m obile dislocations. Nature, 345: 58-60.
B arnett, 1990. Bew are greenhouse confusion. Nature 343, 696-697.
B arnola, J.M ., D . R aynaud, Y .S . K orotkevich and C. Lorius, 1987. V ostok ice core provides 160,000year record of atm ospheric C O 2 . Nature, 329: 408-414.
B arta, L. and D .J. B radley, 1985. E xtension of the specific interaction m odel to include gas solubilities
in high tem perature brine. Geochim. Cosmochim. Acta, 49: 195-203.
B attam , L.J., 1979. Weather (P olish transl.). P anstw ow e W ydawnictw o Naukow e, W arsaw : 182 pp.
B endure, J. and N . Friary, 1990. Hawaii - a travel survival kit. Lonely P lanet P ubl., H authorn, A ustralia,
540 pp.
B enxing, Z., O . W atanabe and D .D . M ulm i, 1984. G lacier features and their variations in the Langtang
H im al Region of Nepal. In: K . Higuchi (E d.): Glacial Studies in Langtang V alley. Data Center for
Glacier Research, Japanese Society of Snow and Ice, Publication No. 2, 121-127.
B erner, W ., P . B ucher, H . O eschger and B . S tauffer, 1977. A nalysis and interpretation of gas content
and com position in n a tu ra l ice. In: Isotopes and Im purities in S now and Ice. Internat. Assoc. of
Hydrol. Sci. Publ. (Washington, D.C.), 118: 272-284.
B erner, W ., B . S tauffer and H . Oeschger, 1978. P ast atm ospheric com position and clim ate, gas
param eters m easured on ice cores. Nature, 276: 53-55.
B erner, W ., H . Oeschger and B . S tauffer, 1980. Inform ation on the CO 2 cycle from ice core studies.
Radiocarbon, 22: 227-235.
B irkeland, B.I., 1930. Tem peraturvariationen auf Spitsbergen. Met. Zeitschr. 47 (6): 234-236.
B ischof, W ., 1960. P eriodical variations of the atm ospheric C O 2 -content in S candinavia. Tellus, 12:
216-226.
B ischof, W ., 1962. Variations in concentration of carbon dioxide in free atm osphere. Tellus, 14: 87-90.
B o d e n , T .A ., P . Kanciruk and M .P . Farrel, 1990. TR E N D S '90 - A com pendium of data on glo b a l
change. Oak Ridge National Laboratory, Oak Ridge, Tennessee, Report No. ORNL/CDIAC-36,
257 pp.
B odnar, R .J., T.J. R eynolds and C.A . K uehn, 1985. Fluid-in c lu s io n system atics in epitherm al
system s. In: B erger, B .R . a n d P .M . B ethke (E ds.): Geology and geochem istry of epitherm al
system s. Soc. Econ. Geol., Rev. in Econ. Geol., 2: 73-97.
B olin, B., 1982. Changing global biogeochem istry. In: P .G. Brew er (E d.): Oceanography - The Present
and Future. S pringer-V erlag, N ew Y ork, 306-326.
B olin, B. 1989. How m uch C O 2 w ill rem ain in the atm osphere. In: B . B olin, B.R . Döös, J. Jäger and
R .A . W arrick (Eds.): The Greenhouse Effect, Climatic Change and Ecosystems (SCOPE 9). John
W iley & S ons, C hichester: 93-155.
B olin, B . and C .D . K eeling, 1963. Large-scale atm ospheric m ixing as deduced from the seasonal and
m eridional variations of carbon dioxide. J. Geophys. Res., 68: 3899-3920.
B orzenkova, I.I., K .Y . V in nikov, L.P . Spirina and D.I. S tekhnovskii, 1976. Variation of Northern
H em isphere air tem perature from 1881 to 1975. Meteorol. i Gidrol., 7: 27-35.
B outron, C .F., 1986. A tm ospheric trace m etals and m etalloids in the snow and ice layers deposited
in Greenland and A ntarctica from prehistoric tim es to present. Advances in Environ. Sci. Technol.,
17: 467-506.
-66B raathen, G.O ., H . D ovland and B . A u ne, 1989. Foreløpig rapport fra NILU og D N M I om
drivhuseffekten og klim autviklingen. Norsk Institutt for Luftforskning, Report NILU OR, 79/89: 76
pp.
B roecker, W .S . and T.-H. Peng, 1974. Gas exchange rates between air and sea. Tellus, 26: 21-35.
B roecker, W .S ., T. Takahashi, H .J . S im pson and T.-H . P eng, 1979. The fate of fossil fuel carbon
dioxide and the global carbon budget. Science, 206: 409-418.
B udyko, M .I., A.B . Ronov and A .L. Yanshin, 1987. History of the Earth's atmosphere. S pringer-V erlag,
139 pp.
B utler, J.N., 1982. Carbon dioxide equilibria and their applications. A ddison-W esley P ubl. Co., 259
pp.
C allendar, G .S ., 1940. Variations in the am ount of carbon dioxide in different air currents. Quart. J.
Roy. Meteorol. Soc., 66: 395-400.
C allendar, G .S ., 1958. On the am ount of carbon dioxide in the atm osphere. Tellus, 10: 243-248.
C ess, R.D., 1989. Presentation to U.S. Department of Energy meeting on research planning for Global
Change, Germantown, MD, April, 1989 - A fter M ichaels, 1990.
C ess, R.D ., G.L. P otter, J.P . B lanchet, G.J. B o e r, S .J . Ghan, J.T. K iehl, H . Le Treut, Z.-X . Liang,
J.F.B . M itchell, J.-J. M orcrette, D .A . R andall, M .R. Riches, E . Roeckner, U. S chlese, A . S lingo,
K .E . Taylor, W .M . W ashington, R.T. W etherald and I. Y agai, 1989. Interpretation of cloud-clim ate
feedback as produced by 14 atm ospheric general circulation m odels. Science, 245: 513-516.
C ess, C .D ., G .L. P otter, M .-H . Zhang, J.-P . B lanchet, S . Chalita, R. Colm an, D.A . Dazlich, A .D . Del
G e n io , V . D im nikov, V . Galin, D . Jerrett, E . K eup, A .A . Lacis, H . Le Treut, X .-Z. Liang, J .-F .
M ahnouf, B .J. M cA vaney, V .P . M eleshko, J.F.B . M itchell, J.-J. M oncrette, P .M . N orris, D .A .
R andall, L. R ikus, E .R oeckner, J.-F. R oyer, U . S chlese, D .A . S heinin, J.M . S lingo, A .P . S okolov,
K .E . Taylor, W .M . W ashington, R .T. W eatherald, and I. Y agai, 1991. Interpretation o f
snow -clim ate feedback as produced by 17 general circulation m odels. Science, 253: 888-892.
C harlson, R .J., J. E . Lovelock, M .O . A ndreae and S.G . W a rre n , 1987. Oceanic phytoplankton,
atm ospheric sulphur, cloud albedo and clim ate. Nature, 326: 655-661.
C oachm an, L.K ., E . Hem m ingsen and P .F. S cholander, 1956. G as enclosures in a tem perate glacier.
Tellus, 8: 415-423.
C oachm an, L.K ., T. Ennis and P .F. Scholander, 1958 a. Gas loss from a tem perate glacier. Tellus,
10: 493-495.
C oachm an, L.K ., E . H e m m ingsen, P.F. Scholander, T. Ennis and H. DeV ries, 1958 b. Gases in
glaciers. Science, 127: 1288-1299.
C oops, A.J., 1991. Analyses of c lim atological series in relation to the detection of the C O 2 effect.
Norsk Geol. Tidsskr., 71: 179-182.
D ecker, R . and R . Koyanagi, 1983. In: A nnual Report of the W orld V olcanic E ruptions. Bull. Volc.
Eruptions 21: 65-66.
D ecker, R .W ., T.L. W right and P .H . S tauffer (E ds.), 1987. V olcanism in Haw aii. US Geol. Surv. Prof.
Pap., 1350: 845-1667.
D eines, P ., 1980. The isotopic com position of reduced organic carbon. In: Fritz, P . and J.C . Fontes
(Eds.): Handbook of environmental isotope geochemistry, Vol. 1. E lsevier: 329-406.
D elm as, R.J., J.-M . A scencio and M . Legrand, 1980. P olar ice evidence that atm ospheric C O 2 20,000
yr BP was 50% of present. Nature, 284: 155-157.
D em arée, G.R ., 1990. D id an a b ru p t w arm ing occur in the 1920s? In: S chietecat G.D. (ed.)
C ontributions á l'E tude des C hangem ents de Clim at. Institut Royal Météorologique de Belgique,
Bruxelles, Publication Série A, 124: 32-37.
D eN iro, M .J. and L.W . C ooper, 1989. P ost-photosynthetic m odification of oxygen isotope ratios of
carbohydrates in the potato: Im plications for paleoclim atic reconstruction based upon isotopic
analysis of wood cellulose. Geochim. Cosmochim. Acta, 53: 2573-2580.
D ruffel, E .R .M . and P .M . W illiam s, 1990. Identification of a deep m arine source of particulate organic
carbon using bom b 14 C . Nature, 347: 172-174.
D rum m ond, S.E ., Jr., 1981. Boiling and mixing of hydrothermal fluids: chemical effects on mineral
precipitation. P h.D . thesis, The Pennsylvania State University: 381 pp.
E ggler, D.H . and B .O . M ysen, 1976. The role of CO 2 in the genesis of olivine m elilitite: Discussion.
Contr. Min. Petrol., 55: 231-236.
-67E lliott, W .P ., 1983. A note on the historical industrial production of carbon dioxide. Climatic Change,
5: 141-144.
E lliott, W .P ., L. M achta and C.D . K eeling, 1985. A n estim ate of the biotic contribution to the
atm ospheric C O 2 increase based on direct m easurem ents at M auna Loa Observatory. J. Geophys.
Res., 90 (D2): 3741-3746.
Encyclopaedia Britannica, 1974. H elen Hem ingw ay Benton Publisher, Chicago.
E pstein, S ., R .P . S harp and A.J. D ow , 1965. S ix -y e a r re c o rd of oxygen and hydrogen isotope
variations in S outh P ole firn. J. Geophys. Res., 70: 1809-1814.
E riksson, E ., 1 9 6 3 . P o s s ible fluctuations in atm ospheric carbon dioxide due to changes in the
properties of the sea. J. Geophys. Res., 68: 3871-3876.
E tc h eto, J. and L. M erlivat, 1988. S atellite determ ination of the carbon dioxide e x c h a n g e
coefficient at the ocean - atm osphere interface: A first step. J. Geophys. Res. 93, 15669-15678.
Faure, H., 1986. Principles of Isotope Geology. John W iley and Sons, New Y ork: 589 pp.
Faure, H., 1990 . C h a n g e s in the global continental reservoir of carbon. Paleont., Paleogeography,
Paleoecol. (Global and Planet. Change Sect.), 82: 47-52.
Fergusson, J.E ., 1985. Inorganic Chemistry and the Earth, Chemical Resources, their Extraction, use
and Environmental Impact. P ergam on Series on Environm ental S ciences, P ergam on Press: 400
pp.
Fonselius, S., F. Koroleff and K .-E. W ärm e, 1956. Carbon dioxide in the atm osphere. Tellus, 8:
176-183.
Frøyland, J., 1992. Introduction to Chaos and Coherence. IO P P ubl. Inc., B ristol: 156 pp.
Frydendahl, K . 1 9 89. Global og regional tem peraturudvikling siden 1850. Danish Meteorological
Institute, Sci. Rep., 89-6: 112 pp.
G arrels, R.M . and C .L. Christ, 1965. Solutions, minerals, and equilibria. Harper & Row : 450 pp.
G arrels, R .M . and M .E . Thom pson, 1962. A chem ical m odel for sea w ater at 25EC a n d o n e
atm osphere total pressure. Am. J. Sci., 260: 57-66.
G audry A ., P . M onfray, G . P olian and G . Lam bert, 1987. The 1982-1983 E l N iño: a 6 billion ton C O 2
release. Tellus, 39 B: 209-313.
G erlach, T.M . and B .E . Taylor, 1990. C arbon isotope constraints on degassing of carbon dioxide from
K ilauea V olcano. Geochim. Cosmochim. Acta, 54: 2051-2058.
G E S A M P , 1983. A n oceanographic m odel for the dispersion of waters disposed off in the deep sea.
IM O /FA O /U N E S C O /W H O /IA E A /U N E P Jo int Group of E xperts on the Scientific A spects of the
M arine P ollution - GE S A M P . International Atomic Energy Agency, Vienna, Reports and Studies No.
19, 1-182.
G onfiantini, R ., 1981. The δ-notation and the m ass-spectrom etric m easurem ent techniques. Internat.
Atomic Energy Agency Tech. Rep. Ser., 210: 35-84.
G ordon, J.E ., 1980. R ecent clim atic trends and local glacier m argin fluctuations in W est Greenland.
Nature, 284: 157-159.
G orshkov, V.G., 1982. The possible global budget of carbon dioxide. Nuovo Cimento, Ser. 1, 5 C (2):
204-222.
G ow , A .J., 1968. D eep core studies of accum ulation and densification of snow at B yrd Station and
Little Am erica V, A ntarctica. U .S . A rm y Cold Regions Research and Engineering Laboratory,
H anover, N ew H am pshire, Research Report, CRREL-197: 45 pp.
G ow , A .J. and T. W illiam son, 1975. G as inclusions in the Antarctic ice sheet and their glaciological
significance. J. Geophys. Res., 80: 5101-5108.
G rootes, P .M ., M . S tiver, T.L. S aling, P .A . M ayew ski, M .J. S pencer, R .B . A lley and D. Jenssen, 1990.
A 1400-year oxygen isotope history from Ross S ea area, Antarctica. Ann. Glaciol., 14: 94-98.
G rovem an, B .S . and H .E . Landsberg, 1979. Reconstruction of N orthern Hem isphere tem p erature:
1579-1880. Meteorol. Prog., Univ. Maryland, Publ. 79-181: 1-46.
H aavelm o, T., 1951. Innføring i teoretisk statistikk. Del II. M em orandum fra S osialøkonom isk Institutt,
U niversity of Oslo: 123 pp.
H agen, J.O ., 1990. P ersonal com m unication.
H agen, J.O . and O . Liestøl, 1990. Long-term m ass balance investigations in Svalbard 1950-1988.
Ann. Glaciol., 14: 102-106.
-68H ansen, J.E ., A . Lacis, D. R ind, G . Russell, P . Stone, I. Fung, R. Ruedy and J. Lerner, 1984. Clim ate
sensitivity: A nalysis of feedback m echanism s. In: J.E . H ansen and T. Takahashi (E ds.) C lim ate
P rocesses and C lim ate S ensitivity, Am. Geophys. Union, Geophys. Monogr., 29.
H ansen, J., 1988. Testimony to U.S. House of Representatives, 6/88. - A fter M ichaels, 1990
H ansen, J. and S . Lebedeff, 1987. G lobal trends of m easured surface air tem perature. J. Geophys.
Res., 92: 13345-13372.
H ansen, J. and S . Lebedeff, 1988. G lobal surface tem peratures: update through 1987. Geophys. Res.
Lett., 15: 323-326.
H anson, K ., G .A . M aul and T.R . K arl, 1989. A re atm ospheric "greenhouse" effects apparent in the
clim atic record of the contiguous U .S . (1895-1987)? Geophys. Res. Lett., 16: 49-52.
H anssen-B auer, I., M . K ristensen S olås and E .L. S teffensen, 1990. The clim ate of S pitsbergen. Det
N orske M eteorologiske Institutt, Oslo, DNMI Rapport Nr. 39/90: 1-40.
H arris, D .M . and A .T. A nderson, Jr., 1983. Concentrations, sources, and losses of H 2 O , CO 2 , and S
in K ilauea basalt. Geochim. Cosmochim. Acta, 47: 1139-1150.
H em m ingsen, E., 1959. Perm eation of gases through ice. Tellus, 11: 355-359.
H enley, R.W ., A .H . Truesdell and P .B . Barton, Jr. (E ds.), 1984. Fluid-m ineral equilibria in
hydrotherm al system s. Soc. Econ. Geol., Rev. in Econ. Geol., 1: 267 pp.
H esselberg, T . and B .J. Birkeland, 1940. Säkulare Scw ankungen des K lim as von N orwegen. Die
Lufttem peratur. Geofys. Publ. (Oslo), 14: 106 pp.
H isdal, V ., 1956. Statistical descriptio n of pressure. In: Norwegian-British-Swedish Antarctic
Expedition, 1949-52. Scientific Results. Norsk P olarinst., Oslo: 7-34.
H oefs, J. 1980. Stable isotope geochemistry, 2nd. ed. M inerals and Rocks, V ol. 9. S pringer-V erlag:
208 pp.
H odgm an, C.D ., R .C . W east, R .S . S hankland and S .M . S elby, 1962 . Handbook of Chemistry and
Physics, 44th ed. C hem . R ubber P ubl. C o., Cleveland, Ohio: 3604 pp.
H olland, H.D ., 1984. The chemical evolution of the atmosphere and oceans. P rinceton University
P ress: 582 pp.
H ollister, L .S . and M .L. C raw ford, 1981. S hort course in fluid inclusions: applications to petrology.
Mineral. Assoc. of Canada, Short Course Handbook, 6: 304 pp.
H orita, J., 1989. Analytical aspects of stable isotopes in brines. Chem. Geol. (Isotope Geosci. Sect.),
79: 107-112.
Iansiti, E . a nd L. N iehaus, 1989. Im pact of energy production on atm ospheric concentration of
greenhouse gases. IAEA Bull., 31 (2): 12-20.
IP C C , 1990. Intergovernm ental Panel on C lim atic Change: The IPCC Scientific Assessment. Report
prepared for IPCC by Working Group 1. Cam bridge University P ress, Cam bridge, 365 pp.
Jaw orow ski, Z., 1989. Pollution of Norwegian A rctic. Norsk P olarinstitutt, Oslo, Rapportserie Nr. 55,
1-115.
Jaw orow ski, Z., T.V . S egalstad and N. Ono, 1992. D o glaciers tell a true atm ospheric C O 2 story? Sci.
Tot. Environ., 114: 227-284.
Jouzel, J., C. Lorius, J.R . P etit, C . Genthon, N .I. B arkov, V .M . K otlyakov and V.M . P etrov, 1987.
V ostok ice core: a continuous isotope tem perature record over the last clim ate cycle (1 6 0,000
years). Nature, 329: 403-408.
Jones, P.D., 1988. Hem ispheric surface air tem perature variations: R ecent trends and an update to
1987. J. Clim., 1: 654-660.
K arl, T.R. and P .D . Jones, 1989. U rban bias in area-averaged surface air tem perature trends. Am.
Met. Soc. Bull., 70: 265-270.
K atsura, T. and E . Ito, 1990. M elting and subsolidus phase relations in the M gS iO 3 -M gC O 3 system
of high pressures: im plications to evolution of the Earth's atm osphere. Earth Planet. Sci. Lett., 99:
110-117.
K eeling, C.D ., R.B . B acastow , A .E . B ainbridge, C.A . Ekdahl, Jr., P .R . G uenther and L.S . W aterm an,
1976. A tm ospheric carbon dioxide variations at M auna Loa Observatory, Haw aii. Tellus, 28: 538551.
K eeling, C .D ., R .B . B acastow , A .F. C arter, S .C . P iper, T.P . W horf, M . H eim ann, W .G . M ook and H .
R oeloffzen, 1989. A three-dim ensional m odel of atm ospheric C O 2 transport based on observed
w inds: 1. Analysis of observational data. In: P eterson, D .H. (E d.): A spects of clim ate variability
in the P acific and the W estern A m ericas. Am. Geophys. Union, Geophys. Mono., 55: 165-236.
K err, R.A ., 1989. Greenhouse skeptics out in the cold. Science, 244: 1041-1043.
-69K oerner, R ., R . D ubey and M . P arnandi, 1989. Scientist m onitor clim ate and pollution from the ice
caps and glaciers. Geos, 18: 33-38.
K ondratyev, K.Y a., 1988. Climate shocks: natural and anthropogenic. John W iley & S ons, New Y ork:
296 pp.
K orotkevich, Y .S ., V .N . P etrov, N .I. B arkov, L.I. S ukhonosova, D .N . D m itriy e v and V .G . P ortnov,
1978. R esults of studies of the vertical structure of the A ntarctic Ice Sheet near V ostok Station.
Informatsionnyy Byulleten' Sovetskoy Antarkticheskoy Espeditsii, 97: 135-148.
K o tlyakov, B .M ., 1961, Indications of seasonal snow deposits in the central regions of A nta rc tic a .
Byulleten' Sovietskoy Antarcticheskoy Ekspeditsii, 26: 1-17.
K oyanagi R. and T. W right, 1987. Bull. Volcanol., 49: 42-43.
K oyanagi, R ., G . Ulrich and T. W right, 1987. Bull. Volcanol., 49: 44-45.
K ram er, J.R ., 1965. History of sea water. Constant tem perature-pressure equilibrium m odels
com pared to liquid inclusion analyses. Geochim. Cosmochim. Acta, 29: 921-945.
K rauskopf, K .B ., 1979. Introduction to geochemistry, 2nd ed. M cG raw -H ill, Inc.: 617 pp.
K udryashov, B .B ., V .K . C histyakov, E .A . Zagrivny and V.Y . Lipenkov, 1984 a. P relim inary results of
deep drilling at V ostok S tation, A ntarctica, 1981-82. P roceedings of the S econd Intern a tio n a l
S ym posium on Ice D rilling Technology, Calgary, Canada, 30-31 A ug u s t, 1982. US Army Cold
Region Res. and Eng. Spec. Rep. (Hanover, N.H.), 86-340: 123-124.
K udryashov, B .B ., V .K . C histyak o v, V .H . P ashkevich and V.N . P etrov, 1984 b. S election of a low
tem perature filler for deep holes in the A n ta rc tic ice sheet. P roceedings of the Second
International Sym posium on Ice D rilling Technology, Calgary, Canada, 30-31 A ugust, 1982. US
Army Cold Region Res. and Eng. Spec. Rep. (Hanover, N.H.), 86-340: 137-138.
K uo, C ., C . Lindberg and D .J. Thom son, 1990. C oherence established betw een atm ospheric carbon
dioxide and global tem perature. Nature, 343: 709-714.
Lam b, H.H ., 1970. Volcanic dust in the atm osphe re , w ith a chronology and assessm ent of its
m eteorological significance. Phil. Trans. Roy. Soc. London, Ser. A, 266: 425-533.
Lam b, H.H ., 1978. Introduction to the volcanic Dust Veil Index. Climate Monitor, 6: 57-67.
Lam b, H.H ., 1988. Weather, Climate and Human Affairs. Routledge, London: 364 pp.
Landsberg, H.E ., 1974. M an-m ade clim atic changes. In: Physical and dynamic climatology.
Proceedings of the symposium on physical and dynamic climatology. World Meteorol. Org., 347:
262-303.
Langway, C .C ., Jr., 1958. B ubble pressures in G reenland glacier ice. P hysics on the m ovem ent of
ice, Sym posium of Cham onix 16-24 S ept. Internat. Assoc. Sci. Hydrol. Publ., 47: 359-369.
Langw ay, C.C ., Jr., 1967. S tratigraphic analysis of a deep ice core from Greenland. CRREL Research
Report, 77.
Latham , J. and M .H . S m ith, 1990. E ffect on global w arm ing of wind-dependent aerosol generation
at the ocean surface. Nature, 347: 372-373.
Lefauconnier, B . and J.O . H agen, 1990. Glaciers and c lim a te in Svalbard, statistical analysis and
reconstruction of the B røgger Glacier m ass balance for the last 77 years. Ann. Glaciol., 14: 148152.
Loosli, H .H ., 1983. A dating m ethod with 39 A r. Earth Planet. Sci. Lett., 63: 51-62.
Lorenz, E.N ., 1976. Nondeterm inistic theories of clim atic change. Quat. Res., 6: 495-506.
Lorius, C., J. Jouzel, D . Raynaud, J. Han s e n a n d H . Le Treut, 1990. The ice-core record: Clim ate
sensitivity and future w arm ing. Nature, 347: 139-145.
M accagnan, M ., 1981. C ontribution a l'etude des proprietes dielectriques de la glace antarctique;
application geoch im ique. (Third cycle thesis). Laboratoire de Glaciologie et Gephysique de
l'Environment du CRNS Publ., 373.
M acIntyre, R., 1970. W hy the sea is salt. Sci. Am., 223 (5): 104-115.
M aeno, N ., 1967. A ir bubble form ation in ice crystals. In: H. Oura (Ed.): Physics of Snow and Ice. The
Institute of Low Tem perature Science, Hokkaido University: 207-218.
M aeno, N . and D . K uroiw a, 1967. M etam orphism of air bubbles in a snow crystal. J. Glaciol., 6: 561564.
M aier-R eim er, E . and K . Hasselm ann, 1987. Transport and storage of CO 2 in the ocean - an inorganic
ocean-circulation carbon cycle m odel. Climate Dynamics, 2: 63-90.
M anabe, S . and R .T. W etherald, 1980. O n the distribution of clim atic change resultin g fro m a n
increase in C O 2 content of the atm osphere. J. Atmosph. Sci., 37: 99-118.
-70M andelbrot, B .B . and J.R . W allis, 1968. N oah, Joseph, and operational hydrology. Water Resources
Res., 4: 909-918.
M andelbrot, B .B . and J.R . W allis, 1969 a. S om e long-run properties of geophysical records. Water
Resources Res., 5: 321-340.
M andelbrot, B .B . and J.R . W allis, 1969 b. R obu s tn e s s of the Rescaled Range R/S in the
m easurem ent of noncyclic long run statistical dependence. Water Resources Res., 5: 967-988.
M atsuo, S. and Y . M iyake, 1966. Gas com position in ice sam ples from A ntarctica. J. Geophys. Res.,
71 (22): 5235-5241.
M ayo, L.R . and R .S . M arch. A ir tem perature and precipitation at W olverine G lacier, A laska; Glacier
grow th in a w arm er, w etter clim ate. Ann. Glaciol., 14: 191-194.
M eier, M .F., 1990. Reduced rise in sea level. Nature, 343: 115-116.
M ichaels, P .J., 1990. The greenhouse effect: beyond the popular vision of catastrophe. Liberty, 3: 2733.
M iljøverndepartem entet, 1988-89. Miljø og utvikling. Norges oppfølging av Verdenskommisjonens
rapport. Stortingsmelding, 46 (1988-89). M iljøverndepartem entet: 180 pp.
M iller, S .L., 1969. Clathrate hydrates of air in A ntarctic ice. Science, 165: 489-490.
M iller, S .L., 1973. The clathrates - their nature and occurrence. In: W halley, E ., S .J. Jones and L.W .
G old (Eds.): Physics and chemistry of ice. Roy. S oc. Can. (O ttaw a): 42-50.
M itchell, Jr., J.M ., 1975. A reassessm ent of atm ospheric pollution as a cause of long-term changes
of global tem perature. In: S .F. Singer (E d.): The Changing Global Environment. Reidel, D ordrecht:
149-173.
M oene, A., 1991. O n the oceanic control of the global surface air tem perature. Norsk Geol. Tidsskr.,
71 (3): 195-197.
M onastersky, R., 1989. Looking for M r. G reenhouse. Sci. News, 135: 216-221.
M organ, V .I., I.D . Goodw in, D .M . E th e ridge and C .W . W ookey, 1991. E vidence from A ntarctic ice
cores for recent increases in snow accum ulation. Nature, 354: 58-60.
M ozeto, A. A ., P . Fritz and E .J. R eardon, 1984. E xperim ental observations on ca rb o n isotope
exchange in carbonate - w ater system s. Geochim. Cosmochim. Acta 48, 495-504.
M ulvaney, R., E .W . W olf and K . Oates, 1988. S ulphuric acid at grain b o u n d a ries in Antarctic ice.
Nature, 331: 247-249.
M urozu m i, M ., T.J. C how and C .C . P atterson, 1969. Chem ical concentrations of pollutant lead
aerosols, terrestria l d u st and sea salts in Greenland and A ntarctic snow strata. Geochim.
Cosmochim. Acta, 33: 1247-1294.
N eftel, A ., H . Oeschger, J. S c h w a n d e r, B . Stauffer and R . Zum brunn, 1982. Ice core sam ple
m easurem ents give atm ospheric C O 2 content during the past 40,000 years. Nature, 295: 220-223.
N eftel, A ., H . Oeschger, J. Schw ander and B . Sta uffer, 1983. Carbon dioxide concentration in
bubbles of natural cold ice. J. Phys. Chem., 87: 4116-4120.
N e ftel, A ., E . M oore, H . Oeschger and B . S tauffer, 1985. E vidence from polar ice cores for th e
increase in atm ospheric C O 2 in the past tw o centuries. Nature, 315: 45-47.
N ew ell, N.E ., R.E . N ew ell, J. H siung and W . Zhogxiang, 1989. Global m arine tem perature variation
and the solar m agnetic cycle. Geophys. Res. Lett., 16: 311-314.
N g, A . and C .C . P atterson, 1981. N atural concentrations of lead in ancient A rctic and A ntarctic ice.
Geochim. Cosmochim. Acta, 45: 2109-2121.
N ozaki, Y ., D .M . R ye, K .K . Turekian and R .E . D odge, 1978. A 200 year record of c a rb o n -1 3 a nd
carbon-14 variations in a B erm uda coral. Geophys. Res. Lett., 5: 825-828.
N riagu, J.O., 1978. P roperties and the geochem ical cycle o f lead. In: N riagu, J.O. (E d.): The
Biogeochemistry of lead in the Environment. P art A . E lsevier, N orth Holland B iom edical P ress,
A m sterdam : 1-14.
N utt, D., 1959. Recent studies of gases in glacier ice - a sum m ary. Polar Notes, Occasional
Publication of The Stephansson Collection, Dartmouth College Library (Hannover, NH), 1: 57-66.
O akes, C.S ., R.J. B odnar and J.M . S im onson, 1990. The system N aC l-C aC l 2 -H 2 O : I. The ice liquidus
at 1 atm total pressure. Geochim. Cosmochim. Acta, 54: 603-610.
O eschger, H. U. S iegenthaler, U . S chotterer and A . Gugelm ann, 1975. A box diffusion m odel to study
the carbon dioxide exchange in nature. Tellus, 27: 168-192.
-71O eschger, H ., B . S tauffer, R . Finkel and C .C . Langway Jr., 1985. V ariations of the CO 2 concentration
of occluded air and of anions and dust in polar ice cores. In: E .T. S undquist and W .S . B roecker
(Eds.): The carbon cycle and atm ospheric C O 2 : natu ra l variations A rchean to present. Am.
Geophys. Union Geophys. Mono., 32: 132-142.
O eschger, H . and U . S iegenthaler, 1988. H ow has the atm ospheric concentration of CO 2 changed?
In: Row land, F.S . and I.S .A . Isaksen (E ds.): The Changing Atmosphere. John W iley & S ons, Ltd.,
C hichester: 5-23.
O eschger, H ., 1989. V ariations in com position of the atm osphere (sum m ary). S ym posium on "H va
skjer m ed klim aet i P olarom rådene?" Oslo, 25-26 A pril, 1989. Norsk Polarinstitutt Rapportserie,
53.
O hm oto, H. and R .O. Rye, 1979. Isotopes of sulfur and carbon. In: B arnes, H.L. (E d.): Geochemistry
of hydrothermal ore deposits, 2nd. ed. John W iley & S ons: 509-567.
O 'R eilly, S ., W .L. G riffin and T.V . S egalstad, 1990: The nature and role of fluids in the upper m antle:
E vidence in xenoliths from V ictoria, Australia. In: Herbert, H. K. and S .E . Ho (E d s .): S table
isotopes and fluids in m ineralization. Geol. Dept. & Univ. Extension, Univ. of West. Australia Publ.,
23: 315-323.
P alais, J. M . and H . S igurdsson, 1989. P etrologic evidence of volatile em issions from m ajor historic
and pre-historic vo lc a n ic e ruptions. In: Berger, A ., R .E . Dickinson and J. W . Kidson (Eds.):
U nderstanding C lim ate C hange. Am. Geophys. Union, Geophys. Mono., 52; and Internat. Union of
Geodesy and Geophysics, 7 (Washington, D.C.): 31-53.
P ales, J.C . and C .D . K eeling, 1965. The concentration of atm ospheric carbon dioxide in H aw aii. J.
Geophys. Res., 70: 6053-6076.
P a s s e l, P ., 1989. Curing the greenhouse effect could run into the trillions. The New York Times,
November 19: 1.
P earm an, G.I., D . E theridge, F. de Silva and P.J. Fraser, 1986. E vidence of changing concentrations
of atm ospheric C O 2 , N 2 O and C H 4 from air bubbles in A ntarctic ice. Nature, 320: 248-250.
P etit, J.R ., L. M ounier, J., Y .S . K orotkevich, V .I. K otlyakov and C . Lorius, 1990. P alaeoclim atological
and chronological im plications of the V ostok core dust record. Nature, 343: 56-58.
P lum m er, L.N ., T.M .L. W igley and D .L. P ankhurst, 1978. The kinetics of calcite dissolution in C O 2 w ater system s at 5E to 60EC and 0.0 to 1.0 atm CO 2 . Am. J. Sci., 278: 179-216.
P lum m er, L.N . and E . Busenberg, 1982. The solubilities of calcite, aragonite and vaterite in C O 2 -H 2 O
solutions betw een 0 and 90EC , and an evaluation of the aqueous m odel for the system C aC O 3 C O 2 -H 2 O . Geochim. Cosmochim. Acta, 46: 1011-1040.
P rentice, K .C . and I.Y . Fung, 1990. The sensitivity of terrestrial carbon storage to the clim ate change.
Nature, 346: 48-51.
P ytkow icz, R.M ., 1967. Carbonate cycle and the buffer m ec h anism of recent oceans. Geochim.
Cosmochim. Acta, 31: 63-73.
R am anathan, V., M .S . Lian, and R .D . C ess, 1979. Increased atm ospheric C O 2 : zonal and seasonal
estim ates of the effect on the radiation balance and surface tem perature. J. Geophys. Res., 84:
4949-4958.
R am anathan, V., R .J. C icerone, H .B . S ingh, and J.T. K iehl, 1985. Trace gas trends and their potential
role. J. Geophys. Res., 90: 5547-5565.
R am anathan, V., R .D . C ess, E .F. H arrison, P . M innis, B .R . B arkstrom , A . A hm ad, and D. Hartm ann,
1989 a. Cloud-radiative forcing and clim ate: results from the E arth radiation budget experim ent.
Science, 243: 57-63.
R am anathan, V., B .R . B arkstrom and E . F . H a rris on, 1989 b. Clim ate and the Earth's radiation
budget. Physics Today, 42 (5): 22-32.
R aval, A . and V . R am anathan, 1989. O bservational determ ination of the greenhouse effect. Nature,
342: 758-761.
R au, G.H ., 1991. Another recipe for 14 C dilution. Nature, 350: 116.
R aynaud D ., and R . D elm as, 1977. C om position des gaz contenus dans la glace polaire. In: Isotopes
and Im purities in S now and Ice. International Association of Hydrological Sciences (Washington,
D.C.) Publ., 118: 371-381.
R aynaud, D . and J.M . B arnola, 1985. A n Antarctic ice core reveals atm ospheric C O 2 variations over
the past few centuries. Nature, 315: 309-311.
-72R epp, K ., 1978. S now accum ulation stratigraphy on R iiser-Larsenisen, D ronning M aud Land,
A ntarctica. Norsk Polarinstitutt Skrifter, 169: 81-92.
R evelle, R. and H .E . S uess, 1957. C arbon dioxide exchange betw een atm osphere and ocean and
the question of an increase of atm ospheric CO 2 during the past decades. Tellus, 9: 18-27.
R eynolds, R .W ., C .K . Folland and D .E . P arker, 1989. B iases in s a te llite-derived sea-surfacetem perature data. Nature, 341: 728-731.
R oedder, E ., 1984. Fluid inclusions. A n introduction to studie s o f a ll types of fluid inclusions, gas,
liquid, or m elt, trapped in m aterials from earth and space, and their applic a tio n to the
understanding of geologic processes. Mineral. Soc. Am., Rev. in Mineral., 12: 644 pp.
R ose, D.J., M .M . M iller and C . Agnew , 1983. Global energy futures and CO 2 -induced climate change.
R eport M ITE L 83-015, C am bridge M A , M IT Energy Laboratory.
R ow land, F.S . and I.S .A . Isaksen, 1988. Introduction. In: R ow land, F.S . and I.S .A . Isaksen (Eds.): The
Changing Atmosphere. John W iley & S ons, Ltd.: 1-4.
R ubey, W .W ., 1951. Geologic history of sea water. Geol. Soc. Am. Bull., 62: 1111-1148.
S akshaug, E ., 1990. M ot en ny istid? UnitNytt 20 (4): 3-4 & 18-19.
S atow , K . and O. W atanabe, 1985. N et accum ulation and oxygen isotope com position of snow on
M izuho P lateau, Antarctica. Ann. Glaciol., 6: 300-302.
S chlesinger, M .E . and J.F.B . M itchell, 1987. Clim ate m odel sim u la tions of the equilibrium clim atic
response to increased carbon dioxide. Rev. Geophys., 25: 760-798.
S ch litz e r, R ., W . R oether, W . W eidm ann, P . K alt and H .H . Loosli, 1985. A m eridional 14 C and 39 A r
section in northeast Atlantic deep w ater. J. Geophys. Res., 90 (C4): 6945-6952.
S chneider, S .H., 1975. On the carbon dioxide - clim ate confusion. J. Atmosph. Sci., 32: 2060-2066.
S chneider, S .H., 1989. The greenhouse effect: S cience and policy. Science, 243: 771-779.
S cholander, P .F. and D .C . N utt, 1960. B ubble pressure in Greenland icebergs. J. Glaciol., 3: 671-678.
S cholander, P .F., E .A . H em m ingsen, L.K . Coach m a n a n d D .C . Nutt, 1961. Com position of gas
bubbles in Greenland icebergs. J. Glaciol., 3: 813-822.
S chw ander, J. and B . Stauffer, 1984. Age difference between polar ic e and the air trapped in its
bubbles. Nature, 311: 45-47.
S chw artzm an, D .W . and T. V olk, 1989. B iotic e n h ancem ent of w eathering and the habitability of
E arth. Nature, 340: 457-460.
S eckel, H ., 1977. Höhenänderungen im grönländischen Inlandeis zw ischen 1959 and 1968. Medd.
Grønland, 187 (4): 1-58.
S egalstad, T.V ., 1983. Eksplosive vulkanutbrudd - 100 år siden K rakatau-utbru ddet. Naturen, 107:
163-169.
S egalstad, T.V ., 1990. Tem peratur og C O 2 . Teknisk Ukeblad, 137 (17): (4)-(5).
S egalstad, T.V . and Z. Jaw orow ski, 1991. CO 2 og globalt klim a. Kjemi, 2 (10): 13-15.
S hchennikova, M .K ., G.G . D em vyam ikh and I.A . K orshunov, 1957. Determ ination of solubility of
carbon dioxide in aqueous solutions of s u lfuric acid by the m ethod of isotopic dilution. Zhurnal
Prikladnoi Khimii, 30: 833-838.
S hepherd, T.J., A .H . R ankin and D .H .M . A lderton, 1985. A practical guide to fluid inclusion studies.
B lackie & S ons, Ltd.: 239 pp.
S iegenthaler, U ., 1990. El Niño and atm ospheric CO 2. Nature, 345: 295-296.
S ie g e n th aler, U . and H . Oeschger, 1987. Biospheric CO 2 em issions during the past 200 years
reconstructed by deconvolution of ice core data. Tellus, 39B: 140-154.
S im kin, T., L. Siebert, L. M cC lelland, D. Bridge, C. New hall and J. H. Latter, 1981. Volcanoes of the
World. S m ithsonian Institution, H utchinson R oss Publishing Co., P A : 232 pp.
S kirrow , G., 1975. The dissolved gases - c a rb o n d io x ide. In: Riley, J.P . and G. Skirrow (Eds.):
Chemical Oceanography, Vol. 2, 2nd ed. A cadem ic P ress: 1-192.
S lingo T., 1989. W etter clouds dam pen global greenhouse warm ing. Nature, 341: 104.
S locum , G., 1955. H as the am ount of carbon dioxide in the atm osphere changed significantly since
the beginning of the tw entieth century? Monthly Weather Review, October: 225-231.
S m ith, S .D . and E .P . Jones, 1985. E vidence of w ind-pum p in g o f air-sea gas exchange based on
direct m easurem ents of CO 2 fluxes. J. Geophys. Res., 90 (C1): 869-875.
-73S pencer, R .W . and C hristy, J.R ., 1990. P recise m onitoring of global tem perature trends from
satellites. Science, 247: 1558-1562.
S pittlehouse, D.L. and A .E . Ripley, 1977. Carbon dioxide concentrations over a native grassland in
S askatchew an. Tellus, 29: 54-65.
S ta n h ill, G., 1983. C om m ents on "The M ontsouris series of carbon dioxide concentratio n
m easurem ents. 1877-1910". Climatic Change, 4: 221-237.
S tauffer, B ., W . Berner, H . Oeschger, J. Schwander, 1981. Atm ospheric C O 2 history from ice core
studies. Zeitschr. für Gletcherkunde und Glaziol., 17: 1-15.
S tauffer, B ., H . H ofer, H . O eschger, J. S chw ander and S iegenthaler, 198 4 . A tm ospheric C O 2
concentration during the last glaciation. Ann. Glaciol., 5: 160-164.
S tauffer, B . and H . Oeschger, 1985. Gaseous com ponents in the atm osphere and the historic record
revealed by ice cores. Ann. Glaciol., 6: 54-59.
S tauffer, B ., A . N eftel, H . O eschger and J. S chwander, 1985. C O 2 concentration in air extracted from
G reenland ice sam ples. In: Langway, C .C ., Jr., H . Oeschger and W . D ansgaard (E ds.): Greenland
ice core: geophysics, geochem istry, and the environm ent. Am. Geophys. Union, Geophys. Mono.,
33: 85-89.
S tehle, N .S ., 1967. M igration of bubbles in ice under a tem perature g ra d ie n t. In: Oura, H. (E d.):
Physics of Snow and Ice. The Institute of Low Tem perature S cience, H okkaido University: 119232.
S treet-Perrott, F.A ., 19 9 1 . General circulation (GC M ) m odelling of paleoclim ates: A critique. The
Holocene, 1 (1): 74-80.
S trong, A. E., 1989. Greater global warm ing revealed by satellite-derived sea-surface-tem perature
trends. Nature, 338: 642-645.
S tuiver, M ., R .L. Burk and P .D . Q uay, 1984. 13 C/ 12 C ratios in tree rings and the transfer of biospheric
carbon to the atm osphere. J. Geophys. Res., 89: 11731-11748.
S tum m , W . and J.J. M organ, 1970. Aquatic chemistry. An Introduction Emphasizing Chemical
Equilibria in natural waters. John W iley & Sons: 583 pp.
S undquist, E .T., 1 9 8 5 . G e o logical perspectives on carbon dioxide and the carbon cycle. In:
S undquist, E .T. and W .S . B roecker (E ds.): The carbon cy c le a nd atm ospheric C O 2 : natural
variations A rchean to present. Am. Geophys. Union, Geophys. Mono., 32: 5-59.
Takahashi, T. 1961. C arbon dioxide in the atm osphere and in A tlantic ocean water. J. Geophys. Res.,
66 (2): 477-494.
Takenouchi, S . and G .C . K ennedy, 1965. D issociation pressures of the phase C O 2 ! 5 3 / 4 H 2 O . J. Geol.,
73: 383-390.
Talsm an, S .J., B . W iw char, W .D . Gunther and C.M . S carfe, 1990. D issolution kinetics of calcite in the
H 2 O -C O 2 system along the steam saturation curve to 210EC . In: S pencer, R .J. and I.-M . Chou
(E ds.): Fluid-m ineral interactions: A tribute to H.P . Eugster. The Geochem. Soc., Spec. Publ., 2: 4155.
Tans, P .P . and W .G. M ook, 1980. P ast atm ospheric levels and the 13 C / 12 C ratios in tree rings. Tellus,
32: 268-283.
Thoning, K .W ., P .P . Tans and W .D . K om hyr, 1989 . A tm ospheric carbon dioxide at M auna Loa
O bservatory. 2. Analysis of the N O A A GM CC Data, 1974-1985. J. Geophys. Res., 94 (D6): 85498565.
Toggw eiler, J.R ., 1990. Bom bs and ocean carbon cycles. Nature, 347: 122-123.
Tw ickler, M .S ., M .J. S pencer, W .B . Lyons and P.A . M ayewski, 1986. M easurem ent of organic carbon
in polar snow sam ples. Nature, 320: 156-158.
V olk, T. and Z. Liu, 1 9 8 8 . C o n trols of CO 2 sources and sinks in the Earth scale surface ocean:
tem perature and nutrients. Glob. Biogeochem. Cycles, 2: 73-89.
W ahlen, M ., D . A llen, B . D eck and A . Herc h e n roder, 1991. Initial m easurem ents of C O 2
concentrations (1530 to 1940 AD ) in air occluded in the G IS P 2 ice core from Central Greenland.
Geophysical Res. Lett., 18: 1457-1460.
W alder, I.F. and T.V . S egalstad, 1987. Hydrotherm al vein m olybdenite deposits associated with the
D ram m en granite in the P erm ian Oslo Rift, Norway. American Current Research on Fluid
Inclusions. New Mexico Inst. of Mining and Technol., 60.
-74W alker, J.C.G. and J.I. D rever, 1988. Geochem ical cycles of atm ospheric gases. In: Gregor, C .B .,
R .M . G arrels, F.T. M ackenzie and J.B . M aynard (E ds.): Chemical Cycles in the Evolution of the
Earth. John W iley & S ons, Ltd.: 55-76.
W alker, J.C .G., P .B . Hays and J.F. Kasting, 1981. A negative feedback m echanism for the long-term
stabilization of Earth's surface tem perature. J. Geophys. Res., 86 (C10): 9776-9782.
W ashington, W .M . and G.A . M eehl, 1984. S easonal cycle experim ent on the clim ate sensitivity due
to doubling C O 2 w ith an atm ospheric general circulation m odel coupled to a sim ple m ixed-layer
ocean m odel. J. Geophys. Res., 89: 9475-9503.
W atanabe, 1977. Stratigraphic observations of surfa c e s n o w cover. In: O. W atanabe (Ed.)
G laciological Research P rogram in M izuho P lateau, W est E nderby Land, E ast Antarctica, Part
4, 1974-1975, National Institute of Polar Research (Tokyo). JARE Data Reports, 36: 61-125.
W atanabe, O ., 1978. S tratigraphic studies of the snow cores in M izuho P lateau. Memoirs of National
Institute of Polar Research (Tokyo), Special Issue, 7: 154-181.
W aterm an, L.S ., 1983. C om m ent on "The M o n ts o u ris series of carbon dioxide concentration
m easurem ents, 1877-1910 by S tanhill". Climatic Change, 5: 413-415.
W aterm an, L.S ., D .W . N elson, W .D . K om hyr, and T.B. H arris, 1989. A tm ospheric c a rb o n d ioxide
m easurem ents at Cape M atatula, Am e ric an S am oa, 1976-1987. J. Geophys. Res., 94: 1481714829.
W east, R .C., D .R . Lide, M .J. Astle and W .H . Beyer (Eds.), 1989. CRC Handbook of Chemistry and
Physics, 70th ed. C R C P ress, B oca R aton, Florida.
W ebb, T., III, 1985. H olocene palynology and clim ate. In: H echt, A .D. (E d.): Paleoclimate analysis and
modeling. W iley & S ons, Ltd.: 163-195.
W eeks, W .F. and S .F. Ackley, 1982. The grow th, structure, and properties of sea ice. Cold Regions
Research and Eng. Lab. Mono., 82-1. - A fter Roedder, 1984.
W eidick , A ., 1 9 91. A change in the status of the Greenland inland ice. In: A .K . Higgins and M .
S ønderh o lm (E ds.) Current Research including R eport of Activities, 1990. Grønlands Geol.
Unders., 152: 39-41.
W etherald, R.T. and S . M anabe, 1988. C loud feedback processes in a general circulation m odel. J.
Atmos. Sci., 45: 1397-1415.
W igley, T.M .L., 1983. The pre-industrial carbon dioxide level. Climatic Change, 5: 315-320.
W igley, T.M .L., 1989. Possible clim ate change due to SO 2 -derived cloud condensation nuclei. Nature,
339: 365-367.
W igley, T.M .L., P .D . Jones and P .M . K elly, 1989. E m pirical clim atic studies. In: B olin, B . and J. D öös
(Eds.): The Greenhouse Effect Climatic Change and Ecosystems. J. W iley & S ons: 271-322.
W ilson, C.A . a nd J.F.B . M itchell, 1987. A double CO 2 clim ate sensitivity experim ent with global
clim ate m odel. J. Geophys. Res., 92: 13315-13343.
W olff, E.W ., R . M ulvaney and K . Oates, 1988. The location of im purities in A ntarctic ice. Ann. Glaciol.,
11: 194-197.
W yllie, P.J. and W .L. Huang, 1976. Carbonation and m elting reactions in the system C aO -M gO -S iO 2 C O 2 at m antle pressures with geophysical and petrological applications. Contr. Mineral. Petrol.,
54: 79-107.
Y akir, D ., M .J. D eN iro and P .W . R undel, 1989. Isotopic inhom ogeneity of leaf water: E vidence and
im plications for the use of isotopic signals transduced by plants. Geochim. Cosmochim. Acta, 53:
2769-2773.
Y am am oto, R., T. Iw ashim a and M . Hoshiai, 1987. Clim atic jum p in the polar region (I). Proc. National
Institute of Polar Research, Tokyo Symp. on Polar Meteorology and Glaciology, 1: 91-102.
Zw ally, H.J., 1989. Grow th of Greenland ice sheet: interpretation. Science, 246: 1589-1591.
Zw ally, H.J., A .C . B renner, J.A. M ajor, R .A . B indschadler and J.G . M arsh, 1989. G row th of Greenland
ice sheet: m easurem ent, Science, 246: 1587-1589.
-7512. SAMMENDRAG (NORWEGIAN SUMMARY)
Forutsigelsene om en m enneske-skapt (antropogen) klim aendring gjennom brenning av fossile
karbon-brennstoffer (kull, gass, olje) til C O 2 -gass, er hovedsakelig basert på fortolkninger av m ålte
konsentrasjoner av C O 2 i atm osfæ ren og i bre-is.
V iktig for Jordens C O 2 -syklus er havet og m in e ralenes kjem iske likevekter. P å grunn av deres
enorm e buffer-kapasitet, stabiliserer de den geokjem iske likevekt for CO 2 -gass m ellom hydro-, atm o-,
litho- og biosfæ ren. N yere studier av radiokarbon ( 14 C ) viser at sirkulasjonstiden for oppløst organisk
karbon i de øverste lag i havet bare er noen få tiår. D ette betyr at C O 2 dannet ved brenning av hele
Jordens forråd av fossile karbon-brennstoffer vil bli oppløst i havet før atm osfæ rens konsentrasjon
av C O 2 kan dobles i forhold til det nåvæ rende nivå.
M ålinger av C O 2 i atm osfæ ren i det nittende århundre ble foretatt m ed en unøyaktighet på opptil
100% . E n verdi på 290 ppm (volum deler av en m illion) ble valgt som en gjennom snittsverdi for det
nittende århundre, ved å forkaste "ikke-representative" verdier, som var 10% eller m er forskjellige fra
"det generelle gjennom snitt for denne tiden". En avgjørende subjektiv faktor inngår derfor i anslaget
for det før-industrielle nivå for C O 2 i atm osfæ ren.
M auna Loa-observatoriet på H aw aii har væ rt regnet som et ideelt sted for m åling av det globale
C O 2 -nivå. Im idlertid er observatoriet lagt til toppen av en aktiv vulk a n , s o m gjennom snittlig har ett
utbrudd hve rt 3,5 år. D et er uavbrutt utstrøm m ing av C O 2 fra en sprekkesone bare 4 km fra
observatoriet, o g v e rd ens største aktive vulkankrater ligger bare 27 km fra observatoriet. Disse
spesielle forhold gjorde det nødvendig å innføre "redigering" av m ålingene som fast prosedyre. C O 2 verdiene har derfor væ rt gjenstand for en viss skjønnsm essig vurdering før publisering. Lignende
prosedyrer blir også brukt ved andre CO 2 -observatorier. Den teknikk, som brukes for m åling av
atm osfæ risk C O 2 , kan gi ikke-representative resultater for CO 2 -konsentrasjonen.
C O 2 -konsentrasjonene i luftbobler innesluttet i bre-is blir ofte tatt som representative for tidligere
atm osfæ riske konsentrasjoner. D et ble antatt at sam m ensetningen av luften i boblene forble uendret.
D ette var igjen basert på den antagelse at det ikke kunne væ re væ ske i isen ved en gjennom snittlig
årstem peratur på rundt -25EC , o g a t det derfor ikke kunne ventes noen forandringer som skyldtes
diffusjon. Im idlertid ble det nylig funnet årer fylt m ed væ ske i isen i A ntarktika ved tem peraturer helt
ned til -73EC . Tallrike studier viser at flere kjem iske og fysiske prosesser kan gjøre at CO 2 -innholdet
i isen enten øker eller avtar kraftig i forhold til innholdet i den op p rin n e lige atm osfæ re. Ved flere
undersøkelser ble det funnet at C O 2 -nivået i før-industriell is var det doble av nivået i d agens
atm osfæ re.
M etoder som benytter tørr ekstraksjon a v C O 2 fra knust is, frigjør bare om trent halvparten av
denne gassen fra isen. C O 2 i luftboblene kan trenge gjennom isen ved diffu s jo n eller oppløsning i
væ ske som er til stede langs korngrensene, m ed e n h a s tighet forskjellig fra luftens andre gasser.
D annelse av faste C O 2 klatrater (hydrater) er et problem for bestem m elsen av C O 2 -m e n g d en i
gassinneslutninger i is. A ndre gasser danner også klatrater, m en ved andre tem peraturer og trykk.
D ette kan lede til betydelige forandringer i gass-sam m ensetningen i inneslutninger ved forskjellige
dyp. Derfor er C O 2 -m ålinger av luftbobler innesluttet i is ikke re p re s e n tative for den opprinnelige
atm osfæ re . D ette innebæ rer at bre-is ikke kan betraktes som et bestandig m ateriale. Iskjerner vil
-76derfor ikke væ re et tilfredsstillende m ateriale for observasjon av atm osfæ rens utvikling gjennom lange
tidsrom .
D et kan også væ re vanskelig å fastsette alderen til luftbobler innfanget i is. Dette m edfører nok
en usikkerhet i m åleresultatene, til og m ed fra de øvre ~100 m is avsatt på G rønland og i A ntarktika.
D et opprinnelige N 2 /O 2 /A r-forhold i luften er ikke bevart. F o rh oldene er m er lik de m an finner for
oppløselighet av gasser i vann. G assblæ rene m å i hovedsak ha blitt dannet ved utskillelse da vannet
frøs, og ikke fra gjenlukkede kanaler, som opprinnelig var åpne til luften.
B eregninger av tidligere tiders tem peraturer basert på lette stabile isotoper (D /H og
18
O / 16 O ) i is
er også beheftet m ed stor usikkerhet. E tter oppdagelsen av væ ske m ellom iskrystaller i dypfrosset
is i A ntarktika vil faseforandringer m åtte ventes. D et kan regnes m ed stor isotop-utbytting og stor
forandring i isotop-forholdene, noe som m edfører at beregnete paleo-tem pera turer fra isotopforholdene vil væ re m eningsløse når faseforandringer oppstår i næ rvæ r av en m obil væ skefase.
D et har væ rt gjort forsøk på å beregne den tidligere konsentrasjon av C O 2 i atm osfæ ren fra stabilisotopforholdet for k a rb on ( 13 C / 12 C ) i træ rs årringer. D et konkluderes her m ed at CO 2 -innholdet i
atm osfæ ren beregnet fra slike analyser ikke kan bli brukt som en gangbar m etode i paleoklim atologi,
og ikke som bevis for forandring i atm osfæ rens CO 2 -innhold.
D et m est om talte "drivhuseffekt-signalet", en antropogen økning av den globale lufttem peratur,
påstått å ha funnet sted gje n n o m d e siste tiår, er ikke blitt bekreftet gjennom studier av lange
tem peratur-serier.
D et har væ rt hevdet, i henhold til m odellberegninger, at denne oppvarm ingen skulle væ re m est
m erkbar i A rktis. Im idlertid er det ikke påvist noen tem peraturøkning i dette om rådet i løpet av de siste
to tiår. O v e r fle re s tø rre om råder i A rktis og A ntarktis viser breene positiv eller avtagende negativ
m assebalanse. D ette tyder ikke på at den påståtte økning av havnivået er sæ rlig sannsynlig.