Estimating Fossil Fuel CO2 Emissions Using Continental CO2 and 14CO2 Observational Networks
1. 3rd ICOS Science Conference, Prague, 11-13 September 2018
Potential of continental CO2 and 14CO2
observational networks to estimate fossil fuel
CO2 emissions
Yilong Wang, Grégoire Broquet, Philippe Ciais,
Frédéric Chevallier, Felix Vogel, Dan Zhu and other colleagues
LSCE – CEA-CNRS-UVSQ- UniversitéParis Saclay, Gif-sur-Yvette, France
Laboratory for Earth Surface Processes, Peking University, China
2. 3rd ICOS Science Conference, Prague, 11-13 September 2018
2/13Need to monitor anthropogenic CO2 emissions
• Fossil fuels CO2 emission is the largest net CO2 source (~ 10 Pg year-1)
• The 2ºC goal of the Paris Agreement on Climate requires a trend reversal in
the short term and large + sustained emissions reductions in the long term
• But large uncertainty in the self-reported inventories make it difficult to
validate these efforts
(Image: Glen Peters via twitter) (Liu et al., 2015)
3. 3rd ICOS Science Conference, Prague, 11-13 September 2018
3/13Radiocarbon (14C) in the atmosphere as an indicator of fossil CO2
(Turnbull et al., 2006)
• Fossil fuels is devoid of 14C
• Adding FFCO2, the 14CO2/12CO2 ratio (~Δ14C) close to the sources decreases
CO2(ppm) Time series of Niwot Ridge ground-based measurements
background value
Wind from Denver
metropolitan area
4. 3rd ICOS Science Conference, Prague, 11-13 September 2018
4/13Model simulation of atmospheric 14CO2
𝒅𝑪 𝒂
𝒅𝒕
= 𝑯 𝒕𝒓𝒂𝒏𝒔𝒑
𝑳𝑴𝑫𝒁
𝑭 𝑭𝑭 + 𝑭 𝑵𝑷𝑷 + 𝑭 𝒓𝒆𝒔𝒑 + 𝑭 𝒂𝒐 + 𝑭 𝒐𝒂
Fxx: CO2 fluxes
δxx: 14CO2 signature in CO2
fluxes/reservoirs
Rstd: 14C:12C standard ratio
CO2
14CO2
𝒅𝑪 𝒂 𝜹 𝒂
𝒅𝒕
= 𝑯 𝒕𝒓𝒂𝒏𝒔𝒑
𝑳𝑴𝑫𝒁
𝜹 𝑭𝑭 𝑭 𝑭𝑭 + 𝜹 𝑵𝑷𝑷 𝑭 𝑵𝑷𝑷 + 𝜹 𝑯𝑹 𝑭 𝑯𝑹 + 𝜹 𝒂𝒐 𝑭 𝒂𝒐 + 𝜹 𝒐𝒂 𝑭 𝒐𝒂 +
𝟏
𝑹 𝒔𝒕𝒅
𝑭 𝒏𝒖𝒄 +
𝟏
𝑹 𝒔𝒕𝒅
𝑭 𝒄𝒐𝒔𝒎
• A full carbon cycle of
all the CO2 and 14CO2
fluxes is simulated
over 10 years
(Naegler, 2005)
5. 3rd ICOS Science Conference, Prague, 11-13 September 2018
5/13Evaluation at stations over the globe
Vertical distributionTime series of 14C
• Modeled time series and vertical distribution of atmospheric 14C match with
actual measurements from the stations across the globe
6. 3rd ICOS Science Conference, Prague, 11-13 September 2018
6/13
• We generate a set of observations yo, from assumed “true” fluxes:
gradients of CO2 and CO2·δ between a set of ground-based sites and a
reference site
daily/2-week afternoon mean CO2 and/or 2-week afternoon mean 14CO2
measurements from 43 (ICOS) sites (~1000 14CO2 samples per year)
Color: Fossil CO2 emission (IER-Carbones)
green: Low -> red: High
Circles : station near populated area
Triangles : remote stations
Observing System Simulation Experiments (OSSE)
7. 3rd ICOS Science Conference, Prague, 11-13 September 2018
7/13
Bayesian inversion: xa=Argmin J: 𝐽 𝐱 =
1
2
𝐱 − 𝐱b 𝑇
𝐁−1
𝐱 − 𝐱b
+
1
2
𝐻 𝐱 − 𝐲 𝑜
𝑇
𝐑−1
𝐻 𝐱 − 𝐲 𝑜
• We solve for x, start from a prior knowledge xb (different from the true fluxes):
• monthly budgets of fossil, biogenic, ocean CO2 fluxes, nuclear and
cosmogenic 14CO2 fluxes and monthly scaling factors of 14C signature in
corresponding CO2 fluxes for a set of regions
fossil fuel (×55), biosphere fluxes (2×54) nuclear power plant(×10)
ocean fluxes(2×9)
cosmogenic production:
13 latitudinal bands• Metric 1: uncertainty reduction in x (posterior vs prior)
• Metric 2: error in x (posterior xa vs prior xb)
Observing System Simulation Experiments (OSSE)
8. 3rd ICOS Science Conference, Prague, 11-13 September 2018
8/13CO2 and 14CO2 signals of prior uncertainties
CO2(ppm)
days
• CO2 Signals of fossil emissions are much smaller than the CO2 signals
of biogenic fluxes
CO2·δ(ppm‰)
days
• But, 14CO2 signals are dominated by fossil fuel emissions
Signals of prior uncertainties in different fluxes at Heidelberg site (HEI)
9. 3rd ICOS Science Conference, Prague, 11-13 September 2018
9/13Results
CO2 daily 2-week daily
14CO2 - 2-week 2-week
• Uncertainty reductions (URs) in first inversion is smaller than the other two
• Posterior fossil fuel emissions assimilating 14CO2 data are more close to the truth than that
assimilating only CO2 data
• Atmospheric 14CO2 data is more efficient and accurate to constrain fossil fuel emissions
Uncertainty
reduction of
fossil CO2
emissions
prior
truth
daily CO2+2-week 14CO2
daily CO2
10. 3rd ICOS Science Conference, Prague, 11-13 September 2018
10/13Isotopic inversion configured with high-resolution transport model
• x: monthly budgets of CO2 fluxes and
monthly scaling factors of 14CO2 signature
for a set of regions
• y: gradients of CO2 and CO2·δ
43 stations
• Transport model: CHIMERE
0.5°×0.5°(latitude×longitude)
11. 3rd ICOS Science Conference, Prague, 11-13 September 2018
11/13Simulation of atmospheric 14CO2
LMDZ
(2.5°×3.75°)
CHIMERE
(0.5°×0.5°)
Signals of prior uncertainties in different fluxes at Heidelberg site (HEI)
• Fossil CO2 signals simulated with high-resolution morel are comparable with those of
biogenic CO2 at HEI
CO2·δ(ppm‰)CO2·δ(ppm‰)
12. 3rd ICOS Science Conference, Prague, 11-13 September 2018
12/13Results of regional inversion
CO2 daily 2-week
14CO2 - 2-week
Uncertainty reduction
of global inversions
Uncertainty reduction
of regional inversions
• Regional inversion show higher potential of constraining the fossil fuel emissions than
global inversion
13. 3rd ICOS Science Conference, Prague, 11-13 September 2018
13/13Conclusion
• Atmospheric 14CO2 data provide additional constrains to
separate the fossil fuel CO2 emissions and biogenic CO2 fluxes
• 14CO2 observations will be a key component in the observing
system to monitor anthropogenic emissions
Thank you!