Page 1 of 16
Project:
GEOCARBON
Project full title:
Operational Global Carbon Observing System
European Commission - FP7
Collaborative Project (large scale integrating project) - for specific
cooperation actions (SICA) dedicated to international cooperation partner countries
Grant agreement no.: 283080
Del. no: 13.1
Deliverable name: Synthesis of the global to regional land carbon fluxes and stocks from multiple
CCDAS
Version: 1.0
WP no: 13
Lead beneficiary: LSCE-UVSQ
Delivery date from Annex I (project month): 32
Actual delivery date (project month): 36
1. Introduction
1.1
Short summary
The results of different Carbon Cycle Data Assimilation Systems (CCDASs) have been synthesized and
compared to other independent approaches (i.e., an ensemble of atmospheric inversions and Dynamic
Ecosystem Model simulations (DGVMs)). The results corroborate a global net land sink of 1.3±0.5
PgC/yr for the 2001-2009 period (natural exchanges plus biomass burning). Additionally, we observe a
significant trend of around 0.2 PgC/yr increase during the last decade (2000s). We note that the
optimized natural land flux depends on the biomass burning flux that is imposed to each system. The
spatial distribution of the carbon sink is in favour of i) a large sink in the North (from 0.8 to more than
2. PgC/yr) except for one system that has a positive flux (DALEC), ii) nearly neutral tropical CO2
exchanges (except for a large sink in DALEC) and iii) a small sink or source in the South (except for
one system with a large source (BETHY)). At the regional scale, the model differences become too
large to draw robust conclusions. Overall, the CCDAS results fall in between the flux estimates from
standard DGVMs and from atmospheric inversions. The assimilation of atmospheric CO2
concentrations into these process-based models (through model parameters optimization) provides a
strong and unique constraint on the seasonal cycle of the CO2 fluxes, reconciling the initial
discrepancies seen in the DGVMs as well as on the inter-annual flux variations (IAV). CCDASs further
Page 2 of 16
offer the possibilities to diagnose the optimal contribution from Gross Primary Production (GPP) and
Respiration. The results tend to indicate that the GPP year-to-year variations are relatively similar
between the different systems but that the respiration fluxes explain most of the differences.
Overall, such analysis represents the first comparison of its kind; it highlights the potential of CCDASs
and pave the road of future directions to investigate.
1.2
Rationale for this deliverable
Importance:
Deliverable 13.1 represents a first summary of the results of several CCDASs that describe what
can be learned on the land carbon cycle though the optimization of land surface models. The
comparison of CCDAS results to other approaches (pure atmospheric inversions, ecosystem model
simulations, data-oriented models, …) is important as it highlights i) the benefit of assimilating
various data streams into current process-based land ecosystem models and ii) the critical issues that
remain to be resolved to improve future model data fusion schemes.
Issue addressed: We address the current level of agreement between different data assimilation
schemes based on different land ecosystem models. We specifically discuss the impact of data
assimilation to constrain i) the seasonal cycle of the net CO2 exchanges, ii) the inter-annual
variability and the spatial distribution of the net flux and iii) the gross carbon fluxes (in particular
the GPP). Indications about key differences between the systems that can explain the obtained
systematic differences in the net carbon fluxes are provided.
Fitting overall frame of project: This task provides a comprehensive analysis of the potential of
model parameter optimization using various data streams. Although not mature enough, these data
assimilation scheme open the road for new insight on the carbon cycle and how to reduce future
uncertainty in the prediction of the land carbon budgets and thus the future climate change.
1.3
Problems encountered and envisaged solution
No particular problems attached to the completion of this deliverable have been encountered. Few
difficulties, mentioned in a previous deliverable (D9.1), concerned the completion of the
assimilation run for several CCDAS and a late delivery of the results. These delays, however, only
marginally impacted the proposed land carbon synthesis, given the three months extension of the
overall project.
Page 3 of 16
2 Full description
We discuss here first the overall land carbon sink results from the different CCDAS approaches
used in GEOCARBON, at the global and continental scales:

BETHY-CCDAS: Optimization of BETHY model parameters using NDVI and atmospheric
CO2 data (Rayner et al. 2005).

LSCE-CCCDAS: Optimization of ORCHIDEE model parameters (Krinner et al. 2005)
using NDVI, FluxNEt and atmospheric CO2 data (http://orchidas.lsce.ipsl.fr/).

JSBACH-CCDAS: Optimization of JSBACH model parameters using fAPAR and
atmospheric CO2 data (Kaminsky et al. 2013).

DALEC-CCDAS: Optimization of DALEC model parameters using forest biomass, soil
carbon and satellite LAI data (Bloom et al. 2014).

CTDAS: Carbon Tracker Europe data assimilation system optimizing fluxes with
atmospheric data (Van der Velde et al. 2014)
We then put these results into the context of other partly independent studies to investigate the
contribution of data assimilation and model parameters optimization. Finally we briefly highlight
some remaining challenges.
2.1
Overall carbon balance from the different CCDAS
Figure 1 displays the global and latitudinal CO2 budget of the 5 CCDASs used in GEOCARBON,
with a focus on the land budget. The results from these graphs have already been partially discussed
in the deliverable D12.1. Overall for the mean global land carbon fluxes over the 2001-2009 period,
we find that:
 Large differences occur in terms of natural land-ecosystem carbon uptake, with values from
1.6 PgC/yr (DALEC) to 3.3 PgC/yr (JSBACH). Note first that the JSBACH model results
only cover three year of that decade.
 The differences are partly due to differences in biomass burning emissions, with low
emissions in the LSCE and BETHY systems (prescribed from deforestation fluxes and
around 0.9/0.6 PgC/yr for LSCE/BETHY) to much higher emission in CTDAS (1.9 PgC/yr,
including both deforestation fluxes and savanna’s burning that also regrow in the following
wet season).
 When the differences in biomass burning are accounted for, the net ecosystem fluxes
become more comparable between the different systems, with a mean carbon uptake around
1.3 PgC/yr, except for DALEC.
 Few particular cases appear. DALEC system that does not use the atmospheric constraint
obtains the smallest natural land uptake (around 1,6 PgC/yr) and only used similar biomass
burning fluxes than CTDAS with thus not accounting for the regrowth. In this particular
case it is thus more relevant to evaluate the natural flux only. JSBACH on the other hand,
has only assimilated two years of atmospheric CO2 data and thus obtain a slightly different
global budget with larger natural and net land uptake than the other systems. Such feature
will likely change with upcoming results that will have assimilated several years of
atmospheric CO2 data.
If we now consider the break down of the net land fluxes between North (above 30°N), Tropic and
South (<30°S) (Figure 1) we obtain even larger differences between the CCDASs. While CTDAS
and LSCE systems have a relatively similar uptake in the North, between 0.8 and 1.1 PgC/yr,
Page 4 of 16
BETHY and JSBACH systems have a much larger uptake, between 1.8 and 2.2 PgC/yr. Note that
DALEC provides a large positive flux to the atmosphere, using as a constraint, satellite NDVI,
FluxNet, forest biomass and soil organic carbon observations, but not atmospheric CO2 data. Over
the Tropics, few systems have more similar fluxes (except DALEC) with a small uptake in LSCE
and BETHY (around 0.3 PgC/yr) and nearly neutral flux in CTDAS. JSBACH provides on the other
hand a significant emission to the atmosphere (0.5 PgC/yr) while DALEC has the largest tropical
land uptake, above 1. PgC/yr. The southern land is nearly neutral for most systems, i.e. between 0.3 and +0.3 PgC/yr, except for BETHY that has a large positive source (around 1.4 PgC/yr) that
compensate for the large northern land sink.
It is thus difficult from these results to draw strong conclusions about the tropical versus extratropical carbon fluxes.
Figure 1: Global mean carbon budget for the period 2001-2009 for the five different “CCDAS”. A) Split of the global carbon
balance between biomass burning, natural land as well as the total land flu. B) Split of the net land carbon fluxes for the
North, Tropic, and South regions. Negative indicate a sink for the atmosphere.
If we now consider the inter-annual variations (IAV) of the net terrestrial flux (natural plus fire
fluxes, Figure 2), a few common features across the different systems are visible:
 First, we notice a significant trend over the last decade (after 2002) with increasing global
land sink (around 0.2 PgC/yr) in all systems, including the DALEC system that does not use
the integrated constraint from atmospheric CO2 observations.
 Such trend is primarily driven by the northern fluxes while over the tropics no coherent
trend between the systems is obtained (not shown).
 Second, the year-to-year variations of the CO2 fluxes are in relatively good phase between
the different CCDASs, although with significant differences in the amplitude (DALEC
having much small year to year flux variation than the LSCE and BETHY systems).
 Prior to 2000, the LSCE and BETHY systems show a broad agreement in the flux IAV with
still noticeable differences. For instance, the 1997-1998 El-nino anomaly is associated to a
positive flux anomaly that reach a maximum in 1998 in the LSCE case but in 1997 in the
BETHY case. The so called “years of the biosphere” after the 1991 Pinatubo volcanic
eruption, associated to a maximum carbon uptake by the vegetation, are also different in the
two models due to some differences in the timing of the respiration flux anomalies (not
shown).
Page 5 of 16
1 .5
Flux ( Pg C/ye a r )
1 .0
0 .5
0 .0
-0 .5
-1 .0
-1 .5
-2 .0
-2 .5
-3 .0
1982
1984
1986
1988
1990
1992
1994
1996
1998
2000
2002
2004
2006
2008
2010
Dat e
BETH Y CCD AS / Te rre st ria l f
lux / 0 5 Globa l La nd / Ye a rly m e a n
CTEU CCD AS V 2 / Te rre st ria l f
lux / 0 5 Globa l La nd / Ye a rly m e a n
D ALEC CCD AS / Te rre st ria l f
lux / 0 5 Globa l La nd / Ye a rly m e a n
JSBACH CCD AS V 2 / Te rre st ria l f
lux / 0 5 Globa l La nd / Ye a rly m e a n
LSCE CCD AS V 3 0 / Te rre st ria l f
lux / 0 5 Globa l La nd / Ye a rly m e a n
Figure 2: Global annual net land terrestrial ecosystem CO2 fluxes for the 5 CCDASs used in GEOCARBON.
If we now consider the gross carbon fluxes, i.e. the Gross Primary Productivity (GPP) and
Ecosystem respiration (Reco), we can further investigate the sources of the net flux differences
between the CCDASs. Figure 3 shows the GPP annual fluxes and annual flux anomalies for each
system. We clearly see large differences in terms of mean annual GPP, with the LSCE system
having the highest GPP (around 150 PgC/yr) and the BETHY system having twice lower GPP
(around 75 GtC/yr). CTDAS, JSBACH and DALEC CCDASs provide values around 120 PgC/yr, a
level that corresponds to the estimates from the data oriented statistical model of Jung et al. (2011).
Despite large differences in terms of mean values, the year-to-year variations of the GPP are
relatively similar between the different systems (figure 3, lower panel) with multi-annual variations.
Figure 3: Global GPP for the five CCDASs used in GEOCARBON. a) annual mean values; b) annual anomalies in PgC/year.
Finally, we investigated the spatial distribution of the net land carbon fluxes (Figure 4) averaged
over the period 2001-2009 (we do not consider JSBACH that only covers 3 years). Large
differences appear in the regional distribution of the net ecosystem fluxes, with the existence of
both source and sink regions. The BETHY system provides the largest spatial contrasts with strong
carbon emission (up to 500 gC/m2/yr) nearby large carbon uptake regions (up to 400 gC/m2/yr).
These patterns are associated to the distribution of the Plant Functional Types (PFTs) used in
Page 6 of 16
BETHY. The other systems have net sources and sinks per square meter that are less intense with
CTDAS having the lowest flux intensity and more heterogeneous flux patterns. Despite these large
differences, a few broad scale features may be drawn:
 Over tropical America, BETHY, DALEC and LSCE systems have a strong uptake over the
Amazon basin while only CTDAS suggests a small source.
 Over North America, there is a tendency to have carbon uptake over the eastern part of
United States.
 Over North Asia, most systems (except DALEC) tend to provide carbon uptake in eastern
China, the northern forest and in particular northeast Siberia for BETHY.
 Over Africa, we observe a dipole with uptake in the equatorial forest and release in the
tropical savanna regions for all systems, except CTDAS that provides a reverse pattern.
Figure 4: Spatial distribution of the mean net ecosystem CO2 exchanges for 4 CCDAS approaches of GEOCARBON (mean
flux over the 2001-2009 period). Positive values indicate emission to the atmosphere.
2.2
Comparison of CCDAS results with other approaches: continental land sink
2.2.1 Different approaches
In order to provide a comprehensive analysis of the carbon fluxes over land and to estimate the
contribution of the GEOCARBON data assimilation approaches, we have compared the results of
our 5 Carbon Cycle data Assimilation Systems (CCDASs) to other approaches, namely:


The results of 8 land dynamic global vegetation model run with a common forcing to
estimate trends in the land carbon fluxes (TRENDY experiment); the simulations that were
used (S2 scenario) correspond to fix land cover but changing climate and CO2 forcing (Sitch
et al. 2014).
The results of 10 land dynamic global vegetation models used within Earth System Models
(ESM); we took the results of the historical runs made for the CMIP5 inter-comparison.
Page 7 of 16

The estimated net land surface fluxes from 11 atmospheric inversions that were compared in
Peylin et al. 2013.
As revealed in Figure 7, there is a gradient of data-streams assimilated in these different
approaches. While the DGVM rely directly on the processed embedded in the model, the CCDAS
have further used several data streams to optimize the original model parameters and the inversions
rely deeply on one particular data stream (atmospheric data) with a flux correction approach. The
comparison of these different approaches should help to understand the strengths and weaknesses of
data assimilation for estimation of the land carbon fluxes and stocks.
We should first notice that the comparison is statistically biased given that we don't have the same
number of model for each approach; in particular the CCDAS approach is only represented by 4
independent realizations.
Figure 5: Different modeling approaches that will be compared in terms of net carbon exchange with the atmosphere: i) pure
land ecosystem model (DGVM) with two sets of models (TRENDY and CMIP5 ensemble), ii) atmospheric inversions that
represent a modification of DGVM prior fluxes to fit the atmospheric CO2 gradients and iii) the CCDAS which have
assimilated the largest number of data streams.
2.2.2 Seasonal CO2 fluxes
Figure 6 and Figure 7 display the mean seasonal cycle for of the natural land CO2 fluxes for two
regions, Northern land (> 30°N) and Tropical land (30°S<->30°N) for the 4 different approaches
described above.
For the northern land, we first notice a very large spread for the DGVMs both in terms of amplitude
and phase of the seasonal cycle. The amplitudes vary by a factor of roughly three, while the phase
of the maximum carbon uptake varies by nearly two months between the different models. These
differences are not due to the climate forcing as they are similar for the TRENDY models that have
used the same climate forcing and for the CMIP5 model results. On the other hand, the atmospheric
inversions present a very coherent seasonal cycle in the northern extra-tropics, with an amplitude of
roughly 3 PgC/month and a maximum uptake always in July. The CCDAS results also present the
same phase and nearly the same amplitude as the atmospheric inversions. Note that we have only 5
models for this approach, which complicate the statistical comparison with the other approaches.
This analysis illustrates the strong constraint brought by the atmospheric data to constraint the
amplitude and phase of the seasonal cycle. It also suggests that there are enough degrees of freedom
with the optimization of a few set of parameters to obtain the same fluxes than when correcting the
DGVM fluxes themselves (atmospheric inversions). This is an important result that reveals the
Page 8 of 16
importance of model calibration and the use of atmospheric CO2 data to constrain the seasonal
cycle of land fluxes.
Figure 6: Mean seasonal cycle of the natural carbon fluxes exchanged between the northern land and the atmosphere. In each
panel the results from 4 different modeling approaches are displayed (8 TRENDY DGVMs, 10 CMIP5 DGVMs, 11
atmospheric inversions and 5 CCDAS). The shaded area represents the envelope of all models for each approach, while the
grey lines represent the individual models.
For the tropical land, the situation is different, with a seasonal cycle that is much smaller than for
the North and that is controlled by the successions of dry and wet seasons. The model spread for
each approach is relatively similar both in terms of amplitude and phase of the seasonal cycle. It is
thus difficult to draw specific conclusions as for the strength of the data assimilation approaches.
The atmospheric data are indeed relatively sparse over the Tropic, which partly explains the poor
level of constrain brought with the atmospheric inversions. A more in-depth analysis of the
inversion results could be made, given that part of the spread is due to two models only. The
CCDAS results tend to have the smallest seasonal amplitude with no clear difference between the
wet and the dry seasons. Such result should be taken with care, given the small number of CCDASs
and needs also further investigations (beyond the time frame of GEOCARBON).
Page 9 of 16
Figure 7: Mean seasonal cycle of the natural carbon fluxes exchanged between the Tropical land and the atmosphere. In each
panel the results from 4 different modeling approaches are displayed (8 TRENDY DGVMs, 10 CMIP5 DGVMs, 11
atmospheric inversions and 5 CCDAS). The shaded area represents the envelope of all models for each approach, while the
grey lines represent the individual models.
2.2.2 Trend and inter-annual variations of CO2 land fluxes
Figure 8, Figure 9 and Figure 10 display the anomalies of the yearly mean natural land CO2 fluxes
(yearly mean minus the mean over the whole period) for three regions, Northern land (> 30°N),
Tropical land (30°S<->30°N) and Southern land (<30°S) for 3 different approaches described above
(TRENDY, Inversions and CCDASs). Note that we did not consider the CMIP5 results given that
the climate simulated by the earth system model are very different which preclude from a direct
comparison of individual yearly anomalies. For the CCDAS, given the small number of models, we
present both a time series with each individual models (lower left graph) and the same graphic but
with a shaded areas to represent the model ensemble.
For the Northern land, the TRENDY models show a large spread in the year to year net carbon
fluxes with poor coherence in the inter-annual variations (IAV), except for a few periods where
most model agree (i.e., larger uptake in 2004, smaller uptake in 2006, for example). Although each
TRENDY model show a small trend with increasing sink over the 2000s, when we account for the
overall model spread such trend is not clearly visible over the short 1990-2010 period. On the other
hand, the atmospheric inversions show more coherence in the estimated northern land flux IAV,
with for instance a positive flux anomaly in 1994 and 2003 not observed in the TRENDY models.
The inversions show also a coherent and significant increase of the land carbon uptake during the
last decade (2000s), compared to the 90s. Such trend of the atmospheric inversion results has been
discussed more extensively in Peylin et al. (2013). If we now consider the CCDAS approaches, we
do not find similar coherence between the 5 systems than for the inversions. If we exclude the
JSBACH-CCDAS that only provides 3 years of fluxes (and that has used only two years of
atmospheric CO2 data) we obtain also a trend during the 2000s with increasing land carbon sink.
Like for the inversions the magnitude of such trend results from the fit to the atmospheric CO2
concentrations (their temporal evolution). We also note that the coherence of the CCDAS results
and in particular the LSCE and BETHY systems is lower for the northern land fluxes than for the
global fluxes as previously discussed (Figure 2). Indeed the correlation of the northern land flux
anomalies between these two systems is close to zero over the 2 decades that are considered. Such
Page 10 of 16
results is primarily due to differences in the respiration fluxes while the GPP IAVs tend to have a
much stronger agreement than the net CO2 flux (not shown). Additional work is thus required to
better optimize the model respiration flux, possibly using a larger set of observations related to soil
carbon decomposition.
Figure 8: Interannual variations of the yearly mean natural carbon fluxes exchanged between the Northern land and the
atmosphere. In each panel the results from 3 different modeling approaches are displayed (8 TRENDY DGVMs, 11
atmospheric inversions and 5 CCDAS). The shaded area represents the envelope of all models for each approach, while the
grey lines represent the individual models. For the CCDAS, one graph with the individual models (lower left) as well as the
shaded version (lower right) are displayed.
For the Tropical land, the IAV the carbon fluxes appear to be large and relatively coherent between
the different models of each approach. These anomalies are controlled primarily by the El-nino
events with large positive anomalies during these periods. The 1998 positive anomalies is visible in
all DGVMs; the inversion results provide a similar anomaly but with few inversions indicating a
bigger anomaly in 1997 than in 1998. The two CCDAS approaches, LSCE and BETHY provide
also a stronger anomalous flux in 1997 than in 1998. Note that for this tropical region, LSCE and
BETHY show a much bigger agreement in terms of phase and amplitude of the flux IAV compared
to the North (see above). Over the 2000s, we notice some differences between the TRENDYDGVMs and the atmospheric inversions, especially in 2002 where the inversion indicate a strong
positive anomaly not observed in the DGVMs. No clear signal is found over that period from the
CCDAS, which cannot favor one or the other result. DALEC CCDAS provide little flux IAV over
the tropics compared to the other CCDAS, which suggest that part of the IAV is dictated by the use
of atmospheric inversions. If we consider the GPP fluxes (not shown), like for the North, we also
find a larger agreement between LSCE and BETHY CCDAS (than for the net flux), which suggests
that the respiration over the tropics needs to be further constrain to draw more clear and coherent
messages for the flux IAV.
Page 11 of 16
Figure 9: same as Figure 8 but for tropical land.
For the Southern land (< 30°S), the carbon flux IAVs appear to be relatively coherent between the
different models of each approach, like for the Tropics. For the DGVMs the phase of the anomalies
are coherent but the amplitudes significantly differ between the models. The atmospheric inversions
provide similar variations except in 1997-1998, where they show a large positive anomaly as in the
Tropics. Either such anomalies correspond to difficulties of the atmospheric inversions to separate
the tropical versus southern land fluxes, given the low density of station over the tropics or it
reveals that most DGVMs do not capture the large fires associated to the El-nino event in 19971998 in the southern part of the tropics. The two CCDASs, BETHY and LSCE do not help to
resolve this issue as they provide opposite flux anomaly for that period. During the 2000s, the
different CCDAS provides similar flux IAV, except DALEC that has much small year-to-year flux
variations.
Figure 10: same as Figure 8 but for southern land.
Page 12 of 16
2.3
Trend in the amplitude of the net CO2 fluxes
Several papers have highlighted a recent trend in atmospheric CO2 concentrations, namely an
increase of the seasonal cycle in the northern hemisphere by up to 50% over the past 50 years
(Graven et al., 2013). Several factors have been proposed to explain this increase, including changes
in oceanic fluxes and atmospheric transport of CO2, increased fossil-fuel emissions, the response of
the terrestrial biosphere to climate change, and recently the intensification of agriculture (Zen et al.
2014 and Gray et al. 2014). However, the relative magnitude and latitudinal contribution of each are
still debated.
Figure 11 displays in the upper panel the signal observed in the atmosphere with increasing CO2
seasonal amplitude at Barrow and Mauna Loa sites (following Graven et al., 2013). In the lower
panel it shows the normalized seasonal CO2 flux amplitude (with respect to the start of the
simulation) for the Northern hemisphere from the different CCDASs. The main outcome is that
current CCDASs do not provide any clear trend in the land ecosystem seasonal flux amplitude to
explain the atmospheric signal. Given that the other possible contributors (ocean and fossil fuel
fluxes) are unlikely to be the main driver, the current CCDASs still fail to represent such trend in
the biosphere exchanges, although they have assimilated atmospheric CO2 observations. This
suggests that current optimization schemes capture first order spatial and temporal gradients of
atmospheric concentrations (trend in the mean concentration and mean spatial gradients) but not the
changes in amplitude and the associated processes. Drivers of the seasonal amplitude flux changes
for the northern ecosystems are either still missing or parameters associated to the existing drivers
are not completely optimized with the current systems. Further investigations are needed to capture
this important feature of atmospheric CO2 observations.
Page 13 of 16
Figure 11: Temporal evolution of the seasonal amplitude of the atmospheric CO2 concentration at Point Barrow and Mauna
Loa stations (upper graphics) and temporal evolution of the seasonal amplitude of the natural land CO2 fluxes of the
northern hemisphere from the different CCDAS model, expressed relatively to the amplitude at t0 (lower graph).
2.4
Remaining challenges
The synthesis of the CCDAS results described above highlighted particular strengths of a “Data
Assimilation procedure” to better constrain the land carbon cycle and its drivers but also specific
modeling challenges and issues to work on in order to fully exploit the approach:


Atmospheric CO2 data act as global constraint to the overall land and ocean carbon budget;
using them to optimize model parameters is promising but we should make sure the land
modeling framework accounts for all surface sources and sinks of CO2 and in particular the
disturbances due to biomass burning and possibly lateral flow of carbon to the ocean via
rivers.
More specifically the differences between CCDAS estimates of net ecosystem exchange
depend on the treatment of biomass burning. Comparison of the natural fluxes should
always account properly for growth and net CO2 flux due to biomass burning.
Page 14 of 16



Most of the current CCDAS (except DALEC) do not use observations of carbon pools (i.e.,
either forest above ground biomass or soil carbon content), that may provide strong
constraints on the slow component of the land carbon cycle and may change the current
estimates of key parameters linked to respirations and “turn-over time”.
A natural follow up of GEOCARBON CCDAS results is to investigate the impact of the
land model parameter optimization on the future carbon cycle and see whether it decreases
the current model spread. Such effort will follow from the dynamic initiated in
GEOCARBON.
Overall, the GEOCARBON project helped to pave the road of multiple data streams
assimilation for the carbon cycle; it substantially pushed several groups to build the next
generation data assimilation systems that will benefit from “ecosystem model
developments” on one hand and on the other hand from increasing of both atmospheric CO2
data and more direct C-cycle observations (fluxes, biomass, fluorescence,..).
4 References
Bloom A. A., Williams, M., 2014 (in prep.): Estimating terrestrial carbon dynamics in a datalimited world: assimilating ecological “common sense” in an ecosystem carbon model-data-fusion
framework.
Graven H. D. et al. Science 341, 1085–1089 (2013).
Gray, J. M. et al. Nature 515, 398–401 doi:10.1038/nature13957 (2014).
Jung, M., Reichstein, M., Margolis, H. a., Cescatti, A., Richardson, A. D., Arain, M. A., Arneth, A.,
Bernhofer, C., Bonal, D., Chen, J., Gianelle, D., Gobron, N., Kiely, G., Kutsch, W., Lasslop, G.,
Law, B. E., Lindroth, A., Merbold, L., Montagnani, L., et al.: Global patterns of land-atmosphere
fluxes of carbon dioxide, latent heat, and sensible heat derived from eddy covariance, satellite, and
meteorological observations, J. Geophys. Res. Biogeosciences, 116, 2011.
T. Kaminski, W. Knorr, G. Schürmann, M. Scholze, P. J. Rayner, S. Zaehle, S. Blessing,
W. Dorigo, V. Gayler, R. Giering, N. Gobron, J. P. Grant, M. Heimann, A. Hooker-Strout,
S. Houweling, T. Kato, J. Kattge, D. Kelley, S. Kemp, E. N. Koffi, C. Köstler, P.P. Mathieu,
B. Pinty, C. H. Reick, C. Rödenbeck, R. Schnur, K. Scipal, C. Sebald, T. Stacke, A. Terwisscha van
Scheltinga, M. Vossbeck, H. Widmann, and T. Ziehn. The BETHY/JSBACH Carbon Cycle Data
Assimilation System: experiences and challenges. J. Geophys. Res., 118:doi:10.1002/jgrg.20118,
2013.
Krinner, G., N. Viovy, N. de Noblet-Ducoudre, J. Ogee, J. Polcher, P. Friedlingstein, P. Ciais, S.
Sitch, and I. C. Prentice (2005), A dynamic global vegetation model for studies of the coupled
atmosphere-biosphere system, Global Biogeochemical Cycles, 19(1).
Le Quéré, C., G. P. Peters, R. J. Andres, R. M. Andrew, T. A. Boden, P. Ciais, P. Friedlingstein, R.
A. Houghton, G. Marland, R. Moriarty, S. Sitch, P. Tans, A. Arneth, A. Arvanitis, D. C. E. Bakker,
L. Bopp, J. G. Canadell, L. P. Chini, S. C. Doney, A. Harper, I. Harris, J. I. House, A. K. Jain, S. D.
Jones, E. Kato, R. F. Keeling, K. Klein Goldewijk, A. Körtzinger, C. Koven, N. Lefèvre, F.
Maignan, A. Omar, T. Ono, G.-H. Park, B. Pfeil, B. Poulter, M. R. Raupach, P. Regnier, C.
Rödenbeck, S. Saito, J. Schwinger, J. Segschneider, B. D. Stocker, T. Takahashi, B. Tilbrook, S.
Page 15 of 16
van Heuven, N. Viovy, R. Wanninkhof, A. Wiltshire, and S. Zaehle, 2014, Global carbon budget
2013, Earth Syst. Sci. Data, 6, 235-263, www.earth-syst-sci-data.net/6/235/2014/ doi:10.5194/essd6-235-2014
Peylin, P., R.M. Law, K.R. Gurney, F. Chevallier, A. R. Jacobson, T. Maki, Y. Niwa, P. K. Patra,
W. Peters, P. J. Rayner, C. Rödenbeck, I.T. van der Laan-Luijkx, and X. Zhang, 2013, Global
atmospheric carbon budget: results from an ensemble of atmospheric CO2 inversions,
Biogeosciences, 10, 6699–6720, www.biogeosciences.net/10/6699/2013/ doi:10.5194/bg-10-6699201
Rayner, P., M. Scholze, W. Knorr, T. Kaminski, R. Giering, and H. Widmann (2005) Two decades
of terrestrial Carbon fluxes from a Carbon Cycle Data Assimilation System (CCDAS). Global
Biogeochemical Cycles, 19:doi:10.1029/2004GB002254.
Sitch et al., revised version under review, see also originally submitted version: S. Sitch, P.
Friedlingstein, N. Gruber, S. D. Jones, G. Murray-Tortarolo, A. Ahlström, S. C. Doney, H. Graven,
C. Heinze, C. Huntingford, S. Levis, P. E. Levy, M. Lomas, B. Poulter, N. Viovy, S. Zaehle, N.
Zeng, A. Arneth, G. Bonan, L. Bopp, J. G. Canadell, F. Chevallier, P. Ciais, R. Ellis, M. Gloor, P.
Peylin, S. Piao, C. Le Quéré, B. Smith, Z. Zhu, and R. Myneni, Trends and drivers of regional
sources and sinks of carbon dioxide over the past two decades, Biogeosciences Discuss., 10, 2011320177, 2013.
van der Velde, I. R., J B Miller, Schaefer, K., Van Der Werf, G. R., Krol, M. C., & Peters, W.
(2014). Towards multi-tracer data-assimilation: biomass burning and carbon isotope exchange in
SiBCASA. Biogeosciences Discussions, 11, 107–149.
Zeng, N. et al. Nature 515, 394–397, doi:10.1038/nature13893 (2014).
Page 16 of 16