Subsea permafrost and hydrates in the East Siberian Arctic Shelf (ESAS) constitute a substantial carbon pool, and a potentially large
source of methane to the atmosphere. Previous studies based on interpolated
oceanographic campaigns estimated atmospheric emissions from this area at
8–17 TgCH
Most long-range global climate projections forecast a warming in the Arctic
of 2–8
Methane emissions from the Arctic Ocean are lower than land emissions, but
more uncertain relatively, as synthesized by
Observations of atmospheric methane mole fractions and of methane isotopes in
the Arctic can improve our understanding of the ESAS emissions
In this paper, atmospheric methane observations and high-resolution
simulations of atmospheric transport in the Arctic are combined to evaluate
the potential of an 8 TgCH
This study is based on the statistical analysis of 1 year of synoptic
atmospheric signal (days to weeks) reaching atmospheric observation sites.
Therefore, continuous observations are needed as weekly or biweekly sampling
does not allow us to capture synoptic changes. As the focus here is on
emissions from ESAS, continuous observations which are sensitive to these
emissions are needed. The year 2012 was chosen as the year with the largest
number of available observations at the time the paper was written. The
double constraint of data availability and of data sensitivity to the ESAS
emissions leaves 4 relevant sites for our analysis (see detailed
characteristics in Table
The methane mole fractions at the observation sites are analysed with
instruments maintained by Environment Canada (EC; ALT), NOAA/Earth System
Research Laboratory (NOAA/ESRL; BRW), the Norwegian Institute for Air
Research (NILU; ZEP), and the Finnish Meteorological Institute (FMI; PAL and
TIK). They are calibrated with standards traceable to the WMO X2004 CH
The continuous observations are hereafter compared to simulated mixing
ratios. Atmospheric transport models have a known bias at nighttime when the
vertical mixing close to the surface is very small
For enhancing atmospheric insights on the ESAS emissions, especially about
the underlying physical processes causing emissions, we also analyse isotope
measurements from ZEP with clear identified origin from East Siberia
Map of the domain of CHIMERE simulations (see
Sect.
Atmospheric transport is simulated with the Eulerian mesoscale
nonhydrostatic chemistry transport model CHIMERE
Methane has a lifetime of 8–9 years regarding oxidation by the OH radicals
Seasonal cycle of prior emissions as used in the model CHIMERE.
The regional transport model CHIMERE requires boundary conditions to its
limited-area domain: (i) surface emissions within the domain and (ii) lateral
and top 3-D concentration fields accounting for transport and emissions
outside the domain to force its open sides (lateral and top sides). Lateral
boundary 3-D fields of mole fractions are interpolated from global analyses
obtained by assimilating surface mole fraction measurements in the global
circulation model LMDz
Surface emissions for the CHIMERE domain are deduced from state-of-the-art
models and inventories: (1) EDGAR v4.2 FT2010 inventory for anthropogenic
emissions (
EDGAR v4.2 FT2010 reports emissions for the year 2010, and not 2012.
Anthropogenic emissions are reported on an annual basis in this inventory and
have been found to only change slightly for the Arctic in the fast track
recent release for 2012 (
The first three types of emissions are projected from their original grids of
Figure
The four types of emissions are run as separate passive tracers in polar
CHIMERE for 2012, which allows the analysis of the contribution of each
source separately at observation sites. The combination of the contributions
from the four types of emissions and from the transported lateral boundary
mole fractions provides the modelled methane mole fractions including the
ESAS contribution. The emission scenario not including ESAS emissions is hereafter
referred to as the basic scenario; the scenario with ESAS emissions is called the reference
scenario. The basic and reference scenarios are compared to
observed time series in Sect.
The magnitude of the ESAS emissions can be derived by adding scaled ESAS
emissions to the basic scenario (see Sect.
The Monte Carlo ensemble (20 000 samples hereafter) is generated by randomly
scaling the anthropogenic emissions, the wetland emissions, and the lateral
boundary conditions, for each month of the year. The distributions used for
these perturbations are Gaussian distributions of, respectively, 50, 75, and
50 % of errors. In addition to the emission scaling, we also add a random
Gaussian noise on the simulated mixing ratios of 60 ppb of standard
deviation. This noise on the simulations is expected to represent the errors
on the transport and from the imperfect distribution of the emissions. The
scaling factors applied on emissions and the random noise in the Monte Carlo
sampling have been chosen in the upper range of known uncertainties in the
used data sets (e.g.
Taylor diagram representation of the statistical analysis of the
ESAS emissions (see Sect.
For each Monte Carlo sample
With this definition of the scores, varying the ESAS emissions results in
trajectories in the Taylor diagram, as illustrated in
Fig.
This statistical analysis is not performed on the whole available data set,
but on afternoon averaged mixing ratios. This processing protocol is widely
used in atmospheric quantitative studies and reduces the impact of local
emissions not-well mixed in the mesoscale transport model (see
Sect.
In the following, simulated mole fractions for the four source contributions described above are compared with methane continuous observations. Then, the Monte Carlo statistical analysis is applied to estimate the methane emissions from ESAS which best fit the atmospheric methane observations. Finally, isotopic remote observations are used to confirm the origin of the ESAS methane emissions.
Time series of observed and simulated methane mole fractions at
five Arctic sites in 2012. The filled-in areas depict the daily afternoon
contributions from wetlands (W, green), fossil fuels and other anthropogenic
emissions (FF, red) and from ESAS (LS, blue; 8 TgCH
At PAL, BRW, ZEP, and ALT, as shown in Fig.
At PAL, a site scarcely influenced by the ESAS emissions, most of the
atmospheric signal is explained by the lateral boundary conditions (i.e. by
air masses coming from outside the CHIMERE domain), especially the large
synoptic variations during winter months. Polar CHIMERE computed with the basic emission
scenario demonstrates a very good skill in winter in representing the atmospheric methane
mole fraction variability at high-latitude sites; the performance is slightly less good in summer.
As shown in Fig.
Observation site characteristics. The site location is displayed in
Fig.
At ZEP, ALT, and BRW (Fig
In summer, at ALT and ZEP, the fit of the reference scenario to the
observations is less favourable than in winter (
Getting closer to ESAS, TIK methane observations compared to simulations
confirm that the simulated contribution of ESAS emissions from January to
April and from October to December is overestimated
(Fig.
However, as confirmed by the footprint analysis at TIK (Fig. S1), observations from July to September are mostly influenced
by regional emissions (closer than 200 km), including ESAS. Within this
radius of influence, wetland emissions from north Yakutia (mainly along
Laptev Sea shores between the Lena and Indigirka rivers) could also significantly
contribute to observed methane mole fractions at TIK. If such wetlands are
poorly represented in the LPJ model at
In summary, the emission scenario from
As seen in Sect.
Figure
The score analysis points at high methane emissions (up to 1.6 TgCH
The estimate computed here is to be considered as an upper bound for the ESAS
emissions for the two following reasons. First, the monthly flat temporal
emission profile from ESAS in our emission scenario underestimates the impact
of the ESAS region on synoptic methane variations at observation sites. In
the real world, concentration peaks due to shorter and more intense methane
release from ESAS would be larger, thus reducing further the estimated
emissions in order to match atmospheric observations. Second, the local and
regional influence of wetland emissions may be systematically underestimated
in the global LPJ model at high latitudes
Monthly fluxes in TgCH
Keeling plot for observations carried out at ZEP observatory in
September–October 2009. Only the observations with a dominant origin from ESAS and
Siberia or from the Arctic Ocean are kept here. The
The isotopic composition in
During the observation campaigns, episodes with identified air origin from
Ob River and eastern Siberia exhibited a mean signature of
Our simple methodology does not allow us to propose a partition of this
biogenic contribution between degrading thawing marine permafrost, degassing
of marine hydrates, and continental biogenic emissions, which are mostly
related to wetlands and freshwaters, but it is possible to eliminate a
dominant thermogenic and pyrogenic contribution. To go further, a full
atmospheric inversion assimilating both
We suggest some insights on methane emissions from the East Siberian
Arctic Shelf using atmospheric methane observations, to complement the
intensive in situ oceanographic measurement campaigns carried out mostly in
summer in the region. We test the consistency of a methane emission scenario
including an 8 TgCH
A multi-year analysis with more observation sites and an improved
representation of the regional wetland area should be carried out in order
to reduce the uncertainties in ESAS emission estimates and to properly
identify the sensitivity of the emissions to the ice cover or to other
meteorological conditions and the distribution and short-scale variability of
the fluxes. The use of another transport model would also be important to
address biases in the representation of transport, not addressed by our
statistical analysis based on centred perturbations. The development of
continuous
We thank the principal investigator from the observation sites we used for maintaining methane measurements at high latitudes and sharing their data. We are grateful to the anonymous referees for their valuable comments which led to a substantial improvement in the presentation of the paper. We also thank Patrick Crill (Department of Geological Sciences, Stockholm University) for his kind advice and remarks on the manuscript and F. Marabelle (LSCE) and the LSCE IT team for the computer resources. This study was supported by the CEA, ANR-CLIMSLIP project, and iZomet Franco-Swedish project. Edited by: M. Heimann