ICOS Science Conference 2020: Day 3, Plenary

1. Changes in Net Ecosystem
Exchange over Europe
During the 2018 Drought
R. L. Thompson, G. Broquet, C. Gerbig, T. Koch, M. Lang,
G. Monteil, S. Munassar, A. Nickless, M. Scholze, M. Ramonet,
U. Karstens, E. van Schaik, Z. Wu and C. Rödenbeck

2. Introduction
u In 2018 Europe experienced a severe heatwave and drought
u Expect extreme meteorological conditions to impact NEE
u Used atmospheric CO2 observations with models of
atmospheric transport to estimate NEE over Europe
u Compared NEE in 2018 with the past 10 years
Study for the ICOS special issue in Phil Trans Roy Soc
(see: Thompson et al., Phil Trans Roy Soc, 2020, in press)

3. Background
NOAA-20 satellite's VIIRS instrument on 30 June and 6 August 2018 show
the browning of western Europe after several weeks of hot, dry weather

4. Background
u Anomalies in 2m temperature and shortwave radiation
compared to 2009-2018 mean (ERA5)
MAM JJA SON

5. Background
u Anomalies in precipitation and soil water content
compared to 2009-2018 mean (ERA5)
MAM JJA SON

6. Methodology overview
u Use atmospheric inversions of CO2 to estimate NEE
posterior estimate
certain
criteria met
end iteration
ACTM model conc.
observed conc.model-obs. error
best guess estimate
optimization method
run iteratively
updating “best
guess” each time

7. Inversion frameworks
Carboscope
Regional
FLEXINVERT LUMIA NAME-HB PyVAR-
CHIMERE
Transport
model
STILT FLEXPART FLEXPART NAME CHIMERE
Meteo ECMWF IFS ECMWF IFS ECMWF EI UK Met ECMWF IFS
Optimization Variational Variational Variational Metropolis
Hastings
Variational
Resolution 0.5°×0.5° 0.5°×0.5° 0.5°×0.5° 0.23°×0.35° 0.5°×0.5°
Temporal 3 hours 12 hours 1 month 6 hours 6 hours

8. Atmospheric network
10°W 0° 10°E 20°E 30°E
40°N
50°N
60°N
70°N
80°N
mountain
lowland
tall tower
u Total of 48 in-situ
measurement sites
u Selected 16 sites with quasi-
continuous records for 2009-
2018 for reference inversion
(black points)
u Europe split into Northern,
Temperate and Mediterranean
regions (based on Koeppen-
Geiger climate regions)

9. Inversion cases
Case Description Purpose Inversion frameworks
SELECT Selected 16 sites
with quasi-continuous
observations
Same uncertainty
reduction in
posterior fluxes
every year
Carboscope,
FLEXINVERT, LUMIA,
NAME
ALL No selection - can
use up to 48 sites
Maximum
observational
constraint
Carboscope, LUMIA,
PyVAR
CLIM As in “SELECT” but
used a climatological
prior NEE estimate
Inter-annual
variability in NEE
driven solely by
observations
Carboscope,
FLEXINVERT, LUMIA,
NAME

10. Prior fluxes
Carboscope
Regional
FLEXINVERT LUMIA NAME-HB PyVAR
CHIMERE
NEE VPRM SiBCASA LPJ-Guess LPJ-Guess VPRM
Fossil fuel* EDGARv4.32 EDGARv4.32 EDGARv4.32 EDGARv4.32 EDGARv4.32
Ocean Mikaloff-
Fletcher et
al. 2007
SOCATv1.6 SOCATv1.6 Takahashi et
al. 2009
zero
(optimized)
Biomass
burning
None GFEDv4.1s None None None
*Fossil fuel data extrapolated to 2019 using BP Statistical Review of World Energy
2019 and temporally disaggregated to hourly using sector-specific activity factors

11. Eddy covariance sites
u Used for independent
comparison (not in inversions)
u Hourly NEE estimates from 17 EC
sites in the FLUXNET network
u Sites selected for quasi-
continuous records for 2009-2018
10°W 0° 10°E 20°E 30°E
40°N
50°N
60°N
70°N
80°N
flux site

12. Comparison CO2 mixing ratios
u All inversion frameworks and cases similar agreement with observations
0.0 0.5 1.0 1.5
0.0
0.5
1.0
1.5
0.1 0.2 0.3
0.4
0.5
0.6
0.7
0.8
0.9
0.95
0.99
prior CSR
LU
PYV
FI
NA
select
all
clim
0.0 0.5 1.0 1.5
0.0
0.5
1.0
1.5
0.1 0.2 0.3
0.4
0.5
0.6
0.7
0.8
0.9
0.95
0.99
posterior

13. North Europe
Annual NEE
u Cases “select” and
“clim” close to 2SD
above mean in 2018
u Negative anomaly in
soil water (SW) in
2018
u Positive anomaly in
temperature in 2018
−15
−10
−5
0
5
NEE(µgCm−2
s−1
)
prior
EC
select
all
clim
−4
−2
0
2
4
∆NEE(µgCm−2
s−1
)
prior
EC
select
all
clim
−0.3
−0.2
−0.1
0.0
0.1
0.2
0.3
∆SW(10xm3
m−3
)
wetter
drier
−2
−1
0
1
2
∆T(C)
2009 2011 2013 2015 2017
cooler
warmer

14. North Europe
Monthly NEE
u All cases show positive
NEE anomaly (decreased
uptake) for July-August
u Prior models VPRM and
LPJ-Guess also positive
anomaly (but not SiBCASA)
u Summer SW 14% less than
2009-2018 mean
−40
−30
−20
−10
0
10
20
NEE(µgCm−2
s−1
)
Spring Summer Autumn Annual
prior
posterior
EC
mean
2018
−15
−10
−5
0
5
10
15
∆NEE(µgCm−2
s−1
)
0±0.05 0.04±0.09 0.03±0.03 0.02±0.02
prior
posterior
EC
select
all
clim
−0.6
−0.4
−0.2
0.0
0.2
0.4
0.6
∆SW(10xm3
m−3
)
−5% −14% 0% −5%
wetter
drier
−4
−2
0
2
4
∆T(C)
J F M A M J J A S O N D
−13% 11% 36% 25%
cooler
warmer

15. Temperate Europe
Annual NEE
u Cases “select” and “clim”
2SD above mean in 2018,
while “all” only slightly
elevated (owing to result
of one inversion)
u Anomaly also seen in EC
data
u Negative SW and positive
temperature anomalies in
2018
−15
−10
−5
0
5
NEE(µgCm−2
s−1
)
prior
EC
select
all
clim
−4
−2
0
2
4
∆NEE(µgCm−2
s−1
)
prior
EC
select
all
clim
−0.3
−0.2
−0.1
0.0
0.1
0.2
0.3
∆SW(10xm3
m−3
)
wetter
drier
−2
−1
0
1
2
∆T(C) 2009 2011 2013 2015 2017
cooler
warmer

16. Temperate Europe
Monthly NEE
u All cases found positive
NEE anomaly in summer
u Cases “select” and “all”
also found positive
anomaly in spring
u Generally consistent with
EC data
u Summer SW 14% less than
2009-2018 mean
−40
−30
−20
−10
0
10
20
NEE(µgCm−2
s−1
)
Spring Summer Autumn Annual
prior
posterior
EC
mean
2018
−15
−10
−5
0
5
10
15
∆NEE(µgCm−2
s−1
)
0.11±0.28 0.23±0.23 0.03±0.16 0.09±0.06
prior
posterior
EC
select
all
clim
−0.6
−0.4
−0.2
0.0
0.2
0.4
0.6
∆SW(10xm3
m−3
)
0% −14% −11% −6%
wetter
drier
−4
−2
0
2
4
∆T(C) J F M A M J J A S O N D
10% 8% 8% 9%
cooler
warmer

17. Convergence of inversion results
u Histograms of summer and
annual NEE anomalies from
all 11 inversions
u For Temperate Europe see
convergence a posteriori
and centre of ensemble
closer to mean anomalies
from EC data
−0.4 0.0 0.2 0.4
0
2
4
6
8
10
−0.4 0.0 0.2 0.4
0
2
4
6
8
10
ΔNEE (PgCy−1
)
−0.4 0.0 0.2 0.4
0
2
4
6
8
10
ΔNEE (PgCy−1
)
−0.4 0.0 0.2 0.4
0
2
4
6
8
10 summer
annual
North Europe Temperate Europe
PosteriorPrior
−0.2 −0.2

18. Spatial distribution of anomaly
u Mean summer and annual
anomalies from 11
inversions
u Spatially posterior summer
anomalies correspond
closely to the prior but
are larger
u Posterior positive annual
NEE anomaly over
Scandinavia not see in
prior

19. Comparison to NDVI anomalies
u Summer NEE anomaly
corresponds closely to
negative NDVI anomaly
u Discrepancy over Scandinavia:
positive annual NEE and NDVI
anomalies
u NDVI not perfect indicator for
NEE (not all absorbed PAR
used for photosynthesis)

20. Summary of main results
North Temperate
Summer Mean
2009-2018
-0.33 ± 0.19 -1.11 ± 0.69
anomaly 0.04 ± 0.09 0.23 ± 0.25
Annual Mean
2009-2018
-0.04 ± 0.05 -0.08 ± 0.17
anomaly 0.02 ± 0.02 0.09 ± 0.06
NEE ensemble mean and standard deviation (PgC y-1)

21. Comparison to 2003 drought
2003 2018
Affected area 20-28 Mha 24-38 Mha
Annual NEE 0.5 PgC/y (Ciais et al. 2005) 0.01 PgC/y (this study)
Warm dry spring Yes Yes
Water-stress effect Yes Yes
Anomalies in net
CO2-uptake from
11 DGVMs
(Bastos et al. 2019)

22. Future perspective
u The 2018 drought was due to a
blocking high pressure system
u Climate models predict such
blocking highs will become more
frequent in summer leading to
more frequent droughts
u Increased temperature in spring
leads to increased plant growth but
depleted soil moisture making
summers more susceptible to soil
water deficits
u Spring increases in C-uptake may
be offset by summer decreases

23. Conclusions
u European drought of 2018 was exceptional for the last 10 years
u Annual NEE anomaly in 2018 for Temperate Europe of
0.09±0.06 PgC/y making the region close to carbon neutral
u Annual NEE anomaly in North Europe of 0.02±0.02 PgC/y making
it also close to carbon neutral
u Decrease in C-uptake in summer likely driven by a soil water
deficit

24. Acknowledgements
u Atmospheric observations (ICOS Carbon Portal)
u Eddy covariance measurements (ICOS Carbon Portal)
u Meteorological data (ECMWF)
u G. Jannsens-Maenhout (EDGARv4.3 anthropogenic CO2 emissions)
u Funding: RCN (ICOS-Norway), EC (VERIFY), SRC (EUROCOM, MERGE,
eSSENCE, BECC), DKRZ, NERC (DARE-UK)