T ellus (2002), 54B, 696–712 Copyright © Blackwell Munksgaard, 2002
Printed in UK. All rights reserved
TELLUS
ISSN 0280–6509
Three years of trace gas observations over the EuroSiberian
domain derived from aircraft sampling —
a concerted action
By I. LEVIN1*, P. CIAIS2, R. LANGENFELDS3, M. SCHMIDT1, M. RAMONET2, K. SIDOROV4,
N. TCHEBAKOVA5, M. GLOOR6, M. HEIMANN6, E.-D. SCHULZE6, N. N. VYGODSKAYA4,
O. SHIBISTOVA5 and J. LLOYD6, 1Institut fu
¨
r Umweltphysik, Universita
¨
t Heidelberg (UHEI-IUP),
Im Neuenheimer Feld 229, 69120 Heidelberg, Germany; 2L aboratoire des Sciences du Climat et de
l’Environnement (L SCE), CE Saclay, Orme des Merisiers, Ba
ˆ
t. 709, 91191 Gif-sur-Y vette Cedex, France;
3CSIRO Atmospheric Research (CAR), Private Bag 1, Aspendale, V ictoria 3195, Australia; 4Svertsov
Institute for Evolutionary and Ecological Problems (IPEE), L eninskii pr. 33, 117071 Moscow, Russia;
5Institute of Forest, Siberian Branch, Russian Academy of Sciences (IF SB RAS), Akademgorodok, 660036
Krasnoyarsk, Russia; 6Max-Planck-Institut fu
¨
r Biogeochemie (MPI-BGC), Postfach 100 164, 07701 Jena,
Germany
(Manuscript received 2 July 2001; in final form 22 April 2002)
ABSTRACT
A three-year trace gas climatology of CO
2
and its stable isotopic ratios, as well as CH
4
,N
2
O
and SF
6
, derived from regular vertical aircraft sampling over the Eurasian continent is presented.
The four sampling sites range from about 1°Eto89°E in the latitude belt from 48°Nto62°N.
The most prominent features of the CO
2
observations are an increase of the seasonal cycle
amplitudes of CO
2
and d13 C–CO
2
in the free troposphere (at 3000 m a.s.l.) by more than 60%
from Western Europe to Western and Central Siberia. d18O–CO
2
shows an even larger increase
of the seasonal cycle amplitude by a factor of two from Western Europe towards the Ural
mountains, which decreases again towards the most eastern site, Zotino. These data reflect a
strong influence of carbon exchange fluxes with the continental biosphere. In particular, during
autumn and winter d18O–CO
2
shows a decrease by more than 0.5‰ from Orle
´
ans (Western
Europe) to Syktyvkar (Ural mountains) and Zotino (West Siberia), mainly caused by soil
respiration fluxes depleted in d18O with respect to atmospheric CO
2
.CH
4
mixing ratios in the
free troposphere at 3000 m over Western Siberia are higher by about 20–30 ppb if compared
to Western Europe. Wetland emissions seem to be particularly visible in July–September, with
largest signals at Zotino in 1998. Annual mean CH
4
mixing ratios decrease slightly from 1998
to 1999 at all Russian sites. In contrast to CO
2
and CH
4
, which show significant vertical
gradients between 2000 and 3000 m a.s.l., N
2
O mixing ratios are vertically very homogeneous
and show no significant logitudinal gradient between the Ural mountains and Western Siberia,
indicating insignificant emissions of this trace gas from boreal forest ecosystems in Western
Siberia. The growth rate of N
2
O (1.2–1.3 ppb yr−1 ) and the seasonal amplitude (0.5–1.1 ppb)
are similar at both aircraft sites, Syktyvkar and Zotino. For SF
6
an annual increase of 5% is
observed, together with a small seasonal cycle which is in phase with the N
2
O cycle, indicating
that the seasonality of both trace gases are most probably caused by atmospheric transport
processes with a possible contribution from stratosphere–troposphere exchange.
* Corresponding author.
e-mail: ingeborg.levin@iup.uni-heidelberg.de
Tellus 54B (2002), 5
697
1. Introduction bution over the extended land masses of European
Russia and Siberia is still very limited. Up to date,
only one systematic study of vertical aircraft pro-The inexorable rise of CO
2
in the atmosphere
primarily reflects anthropogenic emissions from filing over the Eurasian continent extending from
Moscow to Yakutsk was reported by Nakazawafossil fuel and land use changes. The annual
average rate of atmospheric CO
2
increase is, how- and co-workers (1997a). They made regular obser-
vations of CO
2
and its stable isotope ratios asever, less than half the anthropogenic emissions,
which implies that sinks, probably both in the well as CO, CH
4
and N
2
O; however, these were
limited to the summer periods of 1992, 1993 andoceans and on land, currently absorb the other
fraction. More than 10 years ago, Keeling et al. 1994. Within the European project EuroSiberian
CarbonFlux three new sites for regular aircraft(1989), in a modelling study based on observed
latitudinal gradients of CO
2
and 13CO
2
, located a sampling over Russia have been established:
Fyodorovskoye, Syktyvkar and Zotino. Togetherlarge fraction of these sinks in the northern hemi-
sphere, mainly in the North Atlantic ocean, while with new data from a parallel aircraft program
over Western Europe, at Orle
´
ans, these measure-Tans et al. (1990), in a study based on sea-to-air
fluxes inferred from the CO
2
saturation state of ments provide important new information on the
trace gas climatology over the Eurasian continentsurface sea waters and on the latitudinal distribu-
tion of atmospheric CO
2
, pointed out that a from about 1°Eto89°E in the latitude belt of
48°Nto62°N. In this paper, only the results fromsignificant sink of carbon must be attributed to
the Northern Hemisphere continental biosphere. flask sampling at and above 2000 m a.s.l. are
discussed. Also, we will only compare the measure-Since then, atmospheric studies (Fan et al., 1998;
Bousquet et al., 1999; Kaminski et al., 1999; ments of greenhouse gases, i.e. CO
2
concentration
and stable isotope ratios in CO
2
as well as CH
4
,Rayner et al., 1999) and ecological measurements
(Valentini et al., 2000) tend to support this latter N
2
O and SF
6
mixing ratios. For more information
on the complete set of trace substances and verticalfinding by Tans et al. (1990), but the partitioning
of the northern land sink between longitudes is profiles measured at these sites, see detailed reports
by Lloyd et al. (2002), Ramonet et al. (2002a),still a point of debate.
In the last decade, the global network of atmo- and Sidorov et al. (2002).
spheric stations significantly increased at mid to
high northern latitudes. However, most of the new
2. Sampling sites and methods
stations are located in the marine boundary layer,
or on the fringe of continents. As a consequence,
2.1. Flight locations
the land fluxes remain poorly constrained.
Although the atmospheric transport is a superb Figure 1 shows the map of the EuroSiberian
study region with the sampling locations.integrator of the surface fluxes, it is nevertheless
difficult to make measurements in the continental Extending from Western Europe to the Siberian
Highlands, the four sampling sites represent con-atmosphere that are representative of large-scale
sources and sinks, say at the regional to contin- siderably different catchment areas which are illus-
trated by trajectories that extend three daysental scale (>106 km2 ). This is because the sources
are spatially heterogeneous and highly variable in backwards in time starting from the stations at
a height of 2500 m a.s.l. The back-trajectoriestime, and because the atmospheric transport is
also more variable over vegetated areas than over are calculated with the Hybrid Single-Particle
Lagrangian Integrated Trajectory (HYSPLIT4)the oceans (Denning et al., 1996). Therefore, stud-
ies aiming to deduce the fluxes from atmospheric model of Draxler and Hess (1997) using wind
fields from the National Center for Environmentalmeasurements are primarily limited by the lack of
adequate data. It has been hypothesised (Tans Prediction/National Oceanic and Atmospheric
Administration (NCEP/NOAA) medium-rangeet al., 1996) and further quantified (Gloor et al.,
2000) that repeated vertical profiles in the middle weather forecast model. While the Orle
´
ans site
largely experiences North Atlantic air masses, theof continents will improve the diagnostic power
for quantifying the land carbon fluxes. Fyodorovskoye and Syktyvkar sites, although still
generally influenced by westerly winds, duringOur knowledge of the atmospheric CO
2
distri-
Tellus 54B (2002), 5
. .698
Fig. 1. Map of Eurasia with the sampling sites for vertical aircraft soundings and 3 d back trajectories calculated
for the height level of 2500 m a.s.l. for the individual sampling dates and sites at Orle
´
ans, Fyodorovskoye, Syktyvkar
and Zotino.
aircraft sampling are more influenced by north ated by the Svertsov Institute for Evolutionary
and Ecological Problems (IPEE) in close co-western air masses which spent considerable
periods of time over north eastern Europe and operation with Max-Planck-Institut fu
¨
r Biogeo-
chemie (MPI-BGC). The flight location forScandinavia. By contrast, the most continental
Zotino sampling site on the margin of the west Fyodorovskoye is at 33°E, 56°N about 300 km
north-west of Moscow. Vertical profiles of continu-wind belt also occasionally observes air masses
from north eastern Siberia and from the high ous CO
2
NDIR, temperature and relative humid-
ity measurements as well as flask sampling fromArctic.
The Orle
´
ans aircraft program is operated by 100 to 3000 m are performed every 2–4 wk. The
sampling location is over the Central Forestthe Laboratoire des Sciences du Climat et de
l’Environnement, CE Saclay (LSCE), in co- Reserve at Fyodorovskoye in Central European
Russia, the southern taiga, where continuousoperation with Meteo France since 1996. The
flight location is at 1°E, 48°N, about 300 km south ground measurement sites in a forest and a
bog have been established within EuroSiberianof Paris over an area of agricultural land and
forests. Regular vertical aircraft profiles for flask CarbonFlux (Schulze et al., 2002). Flask samples
have been analysed by LSCE for CO
2
concentra-sampling are performed every 2–3 wk from 100 to
3000 m a.s.l. Flasks are analysed at LSCE for CO
2
tion and stable isotope ratios in CO
2
, as well as
for CO and CH
4
mixing ratios. Due to severalconcentration and stable isotope ratios (13C/12C
and 18O/16O) in CO
2
, as well as for CO and CH
4
logistical and sampling problems, the records at
this site contain large gaps, and complete seasonalmixing ratios.
The Fyodorovskoye aircraft program is oper- cycles of trace gases from flask sampling are not
Tellus 54B (2002), 5
699
yet available. For this first summary of the climato- ARROW up to June 1999 and Piper AZTEC from
logy of greenhouse gases over the EuroSiberian
Meteo France afterwards, whereas at the Russian
region, we will only compare individual 3000 m
sites, local Antonov-AN2 bi-plane aircraft were
profile data points from Fyodorovskoye with
used. Separate air intake lines (6 mm dekabon
the other aircraft data. A more detailed report
tubing) for continuous NDIR CO
2
measurements
summarising results from flasks and in situ profile
(LiCor 6152) and for flask sampling systems were
measurements is given by Ramonet et al. (2002a).
installed in the wings of the respective aircraft.
The Syktyvkar aircraft program is operated by
Whole air samples were collected into pre-condi-
IPEE in close cooperation with MPI-BGC. Flights
tioned 1-L cylindrical flasks made of Pyrex glass
are performed 400 km west of the Ural mountains
with PFA O-ring valves (Glass Expansion,
over the Vychegodsko-Mezenskoj plain, about
Australia) at both ends. Drying of the air was
100 km south east (52°E, 61°N) of the city of
performed via magnesium perchlorate. Flasks
Syktyvkar (160 000 inhabitants), the capital of the
were flushed for more than 5 min at a flow rate
Republic of Komi. The area belongs to the northern
of ca. 4 L min−1, and pressurised to 1 atmosphere
European taiga, dominating vegetation in this area
above ambient pressure at final filling (pump:
is pine forest (Pinus sylvestris). Vertical profiles of
KNF-Neuberger, Germany, N86KNDC with
continuous CO
2
NDIR soundings as well as temper-
EPDM membrane).
ature and relative humidity are measured every
Drying of the air with magnesium perchlorate
2–4 wk. Duplicate flasks are collected here only
under conditions of changing pressure may influ-
at 2000, 2500 and 3000 m a.s.l. Flask sam-
ence the CO
2
mixing ratio ending up in the flask.
ples have been analysed by the Institut fu
¨
r
In our sampling system, the drying unit is located
Umweltphysik, Universita
¨
t Heidelberg (UHEI-
upstream of the pump unit, therewith experiencing
IUP) for CO
2
concentration and stable isotope
only pressure changes according to the vertical
ratios in CO
2
, as well as for CH
4
,N
2
O and SF
6
atmospheric pressure change. Two different experi-
mixing ratios.
ments have been performed to test possible system-
The Zotino aircraft program is operated by the
atic biases caused by the drying agent, one in the
Institute of Forest, Siberian Branch, Russian
laboratory, simulating inlet pressure drops similar
Academy of Sciences (IF SB RAS) and MPI-BGC.
to those experienced during aircraft flights up to
The flight location for Zotino is at 89°E, 61°N,
8.5 km (Langenfelds et al., 1996), and two tests
about 600 km north of the city of Krasnoyarsk close
during real aircraft flights (Ramonet et al., 2002b).
to the small village Zotino located on the west bank
The laboratory tests resulted in no detectable
of the Yenisei river. It is situated at the eastern edge
changes associated to pressure changes while the
of the West Siberian Lowland, an extended
(3×106 km2 ) poorly drained area covered by 55%
tests performed during aircraft flights when com-
bogs and about 40% forests (Schulze et al., 2002).
paring samples collected in parallel, with and
The whole region belongs to the Siberian taiga
without magnesium perchlorate showed differ-
(boreal coniferous forest). Vertical profiles of CO
2
,
ences between pairs of samples collected dry
measured with continuous NDIR, temperature and
respectively wet of ‘‘dry–wet’’=−0.23±0.3
relative humidity, are obtained every 2–4 wk, and
(n=4) during the first flight and of ‘‘dry–wet’’
air samples are collected in glass flasks between 100
=−0.10±0.3 (n=4) during the second flight.
and 3000 m a.s.l. Flask samples collected between
Within the standard deviation of this comparison,
June 1998 and July 2000 have been analysed by
no significant offset could be observed. Any pos-
CSIRO Atmospheric Research (CAR) for CO
2
con-
sible bias in the mixing ratios from aircraft samples
centration and stable isotope ratios in CO
2
, as well
reported here are thus probably smaller than
as for CO, CH
4
,N
2
O and H
2
mixing ratios. Flask
−0.25 ppm.
samples collected from July 2000 onwards were
analysed by MPI-BGC in a newly commissioned
2.3. Flask analysis
facility.
Laboratories responsible for flask analysis were
2.2. Flask sampling
LSCE at Orle
´
ans and Fyodorovskoye, UHEI-
IUP at Syktyvkar and CAR and MPI-BGC at
At the Orle
´
ans site, vertical profiling for flask
sampling was performed with light aircraft Piper Zotino. At LSCE, CO
2
concentration analysis was
Tellus 54B (2002), 5
. .700
performed by NDIR (Hartmann & Braun, ory working standards (pure CO
2
) and finally to
the whole-air standards used to check the extrac-Germany, URAS-3G), whereas at UHEI-IUP,
CAR, and at MPI-BGC gas chromatographic tion procedures in the individual laboratory. This
is particularly true for d18O. However, whole-air(GC) systems with a nickel catalyst for conversion
of CO
2
into CH
4
, and flame ionisation detectors standards such as those for CO
2
mixing ratios are
not yet available for stable isotope ratios in CO
2
;(FID) were used. Stable isotope ratios in CO
2
were measured in all four laboratories by isotope therefore, using carbonate or water standards
transferred to CO
2
is the only ‘‘absolute’’ way toratio mass spectrometry (IRMS, Finnigan-252,
Bremen, Germany), after cryogenic extraction of provide calibration of isotope ratio measurements
in atmospheric CO
2
. To date, MPI-BGC d18OCO
2
from the whole air samples. For description
of the CO
2
concentration and isotopic measure- data are calibrated against a whole air standard
using the value assigned to it by CAR. All isotopement procedures see Francey et al. (1996), Bourg
and Ciais (1998, 1999), Neubert (1998), Ramonet ratio data are corrected for interference with N
2
O.
The absolute calibration procedures used in theet al. (1999), and Jordan and Brand (2001). In all
laboratories, CH
4
was analysed by GC-FID individual labarotories are reported in the detailed
descriptions of the data from the individual sites(Levin et al., 1999; Jordan and Brand, 2001;
Langenfelds et al., 2001; Werner et al., 2001). The (Lloyd et al., 2002; Ramonet et al., 2002a; Sidorov
et al., 2002).N
2
O mixing ratio was analysed by the GC elec-
tron capture detector technique (ECD) in all For CH
4
all laboratories relate their mixing
ratios to the NOAA/CCGG scale. For N
2
Onoflasks from Syktyvkar and Zotino by UHEI-IUP,
CAR and MPI-BGC (Jordan and Brand, 2001; internationally agreed calibration scale is avail-
able. CAR and MPI-BGC relate their standardsLangenfelds et al., 2001; Schmidt et al., 2001). SF
6
was only measured in flasks from Syktyvkar by to the scale that is maintained at CAR and derived
from a suite of mixtures gravimetrically preparedUHEI-IUP also using GC-ECD technique (Maiss
et al., 1994). by NOAA/CMDL, while UHEI-IUP is prelimin-
arily linked to the SIO93 scale maintained at
Scripps Institution of Oceanography for the
2.4. Calibrations and drift corrections
ALE/GAGE and AGAGE programs (Weiss et al.,
1981; Schmidt et al., 2001). The scale factor linkingFor the CO
2
mixing ratio, all data are reported
in the WMO CO
2
mole fraction scale maintained these scales has been precisely determined through
regular exchange by CAR and SIO of high-pres-at NOAA/CCGG, Boulder, CO, USA. Laboratory
standards (CO
2
in natural air) were obtained from sure cylinders used as calibration standards in the
AGAGE program to 0.992 52, equivalent to aNOAA/CCGG. Stable isotope ratios in CO
2
are
reported on the Vienna-PDB-CO
2
scale. Respect- difference of 2.4 ppb (CAR-SIO93) at a N
2
O
mixing ratio of 315 ppb. The Zotino data pre-ive carbonate reference standard material
(NBS-19) is provided by the International Atomic sented here have been adjusted to the SIO93 scale.
Preliminary results from a flask intercomparisonEnergy Agency (IAEA), Vienna. These standards
have to be chemically processed to yield carbon between CAR and UHEI-IUP during the period
1998–2000 performed with stainless steel con-dioxide gas for calibration of the respective
working gases (pure CO
2
) at the mass spectro- tainers (2.5 L Sirocans, n=16) and with glass
flasks identical to those used in this program (n=meters. For d18O, calibration of the mass spectro-
meter can also be obtained through water 10) showed a mean difference of 0.7 and 0.5 ppb,
respectively, with UHEI-IUP data being higher.standards ( V-SMOW) which are equilibrated with
CO
2
. The V-SMOW scale is then mathematically This not yet finally confirmed scale difference must
be kept in mind when comparing Zotino andconverted to the V-PDB-CO
2
scale. In this kind
of calibration, chemical or physical processing of Syktyvkar N
2
O flask data. SF
6
data from UHEI-
IUP are reported relative to a diluted gravimetricprimary reference material is necessary for the
transfer of the V-PDB (and the V-SMOW) scale standard gas provided by Messer Griesheim,
Mannheim, Germany. The dilution procedure isto atmospheric CO
2
samples. Through these pro-
cesses, laboratory biases may be introduced which described by Maiss et al. (1996). Its absolute
accuracy is better than 1%.may cause systematic calibration errors of laborat-
Tellus 54B (2002), 5
![Fig. 4. Comparison of CO2 mixing ratio (upper panel ), ively, Dleaves is the leaf isotope discrimination andd13C–CO2 (middle panel ) and d18O–CO2 ( lower panel ) d18Osoil and esoil , respectively, the isotopic com-at the four aircraft sites for the 3000 m level. The lines position of soil-respired CO2 and the diffusiveare harmonic fit curves through the respective data soil–atmosphere fractionation [7.2‰ after Miller(Orléans: dashed line; Syktyvkar: solid line; Zotino: et al. (1999)]. Assuming A to be zero in latedashed–dot–dot line). autumn and early winter, and a d18Osoil of−12‰ from a global modelling study (Cuntz et al., 2002),](/figures/fig-4-comparison-of-co2-mixing-ratio-upper-panel-ively-22sha8lx.webp)
