Observation-based estimates of non-CO greenhouse gases · 2019. 3. 19. · HIMMELI...
Transcript of Observation-based estimates of non-CO greenhouse gases · 2019. 3. 19. · HIMMELI...
Observation-based estimates of
non-CO2 greenhouse gasesRona Thompson (NILU), Tuula Aalto (FMI), Peter Bergamaschi (JRC),
Philippe Bousquet (LSCE), Dominik Brunner (EMPA), Sander Houweling
(VUA), Adrian Leip (JRC), Marilena Muntean (JRC), Pete Smith (UABDN),
Timo Vesala (UH), Pierre Regnier (ULB) & Jean-Matthieu Haussaire (EMPA)
General Objectives
➢ Deliver estimates of CH4 and N2O fluxes, including
anthropogenic as well as natural sources, and build this
capacity into a pre-operational system
➢ Improve the understanding of the processes driving
fluxes of CH4 and N2O, and reduce the uncertainties in
their budgets and trends at national, regional and
continental scales
13 March 2019 VERIFY-CHE GA, Reading 2
Sources of methane
Total global emission of 563 Tg/y
Wetlands
Termites
Ocean
Geological
Coal
Oil & Gas
Livestock
Rice
Waste
Biomass & Biofuel
13 March 2019 VERIFY-CHE GA, Reading 3
Data from the Global Carbon Project
naturalanthropogenic
Sources of nitrous oxide
13 March 2019 VERIFY-CHE GA, Reading 4
Total global emission of 17 Tg/y
Ocean
Natural soil
Agricultural soil
Industrial processes
Biomass burning
Manure management
Waste
Data from the Global Carbon Project
naturalanthropogenic
Uncertainties in BU estimates
Methane
• Microbial emissions which
depend strongly on local
conditions (temperature, soil
moisture etc.)
• Uncertainties on some sources
>100%
• Challenging to represent with
process-based models
Nitrous oxide
• Microbial emissions which
depend strongly on local
conditions (temperature, soil
moisture etc.)
• Emission factor uncertainty
range 30 to 300%
• Challenging to represent with
process-based models
13 March 2019 VERIFY-CHE GA, Reading 5
Large uncertainties in BU methods, therefore, likely
that TD methods can contribute to improved estimates
Overview of the workflow
13 March 2019 VERIFY-CHE GA, Reading 6
Bo
tto
m-u
p Agricultural
emissions of CH4
and N2O
Natural emissions
of CH4
Non-agricultural
anthrop. emissions
of CH4 and N2O
To
p-d
ow
nR
esearc
h &
develo
pm
en
t
Phase 1: Flux estimates using selected inversions
Phase 2: Flux estimates using CIF
DAS for
wetland
fluxes CH4
High-Res
satellite
data in
inversions
Suppl.
tracers in
inversions
High-Res
inversions
Pro
cess-b
ased a
nd
sta
tistical m
odels
Atm
ospheric
invers
ions
13 March 2019 VERIFY-CHE GA, Reading 7
Bottom-up methods
for estimating CH4 and N2O emissions
Agricultural emissions: methods
• Process-based model: ECOSSE
• Statistical model: CAPRI
13 March 2019 VERIFY-CHE GA, Reading 8
Sources Method/Model Notable Inputs
enteric fermentation (CH4) CAPRI
based on IPCC
Tier 1 approach
(emission factors)
crop areas, yield,
livestock densities,
nutrient inputsmanure management (CH4 + N2O)
direct + indirect emissions (CH4 + N2O)
soil emissions (N2O)
(cropland, grassland, forests)
ECOSSE
process-based
land surface
model
climate, land-use,
nutrient inputs, &
soil data
Agricultural emissions: N2O
13 March 2019 VERIFY-CHE GA, Reading 9
avg.
N2O
[kg/h
a/y
r]
N2O emissions for cropland (annual average 2005-2015)
N2O emissions for grassland (annual average 2005-2015)
Preliminary results
Model approach is still under
development
Problem:
Emissions in Central (e.g. NL)
and Western Europe too low
Solution:
Fertilizer application needs to
be changed from fertilizer
demand to actual application
rate
For grasslands:
In the actual assumptions
management is not considered.
2.5
0.0
1.4
0.0
avg.
N2O
[kg/h
a/y
r]
Other anthropogenic emissions
• EDGAR – Emission Database for Greenhouse Gas Research: uses activity
data and IPCC Tier 1 approach. Estimates emissions for both agricultural and
non-agricultural anthropogenic sources
• Used for both CH4 and N2O for all countries
• EDGAR final products: emission time series and global maps at 0.1°×0.1°
13 March 2019 VERIFY-CHE GA, Reading 10
Sources (for VERIFY) Method/Model Notable Inputs
industrial Based on EDGAR.v4.3.2
(1970-2012) and
extended forward in time
to 2015 (results
expected 2019)
Activity data from
international statistics
and satellite remote
sensing
residential
transport
energy
Other anthropogenic emissions
13 March 2019 VERIFY-CHE GA, Reading 11
EDGARv4.3.2 CH4
Other anthropogenic emissions
13 March 2019 VERIFY-CHE GA, Reading 12
EDGARv4.3.2 N2O
Other anthropogenic emissions
13 March 2019 VERIFY-CHE GA, Reading 13
Uncertainties calculated by error propagation
Error bars reflect 95% confidence level
1B1: fugitive emissions from solid fuel
1B2: fugitive emissions from oil/gas
3A1: enteric fermentation
3A2: manure management
3C7: rice cultivation
4D: waste treatment
Soil and peatland emissions
13 March 2019 VERIFY-CHE GA, Reading 14
water tabledepth
CATOTELM
ACROTELM
microbe processesproduction: CH4 CO2
aerobic respiration: CO2 O2
anaerobic respiration: CH4 CO2 O2
oxidation: CH4 CO2 O2
transport processesdiffusionplant transportebullition
Fluxes of CH4, CO2 and O2
Temp,
Prec, Rad
CRU-JRA
climate
data
0.5°×0.5°
resolution
JSBACH land
surface
model
Peat YASSO
soil carbon
model
WTD, LAI,
Resp, Temp
HIMMELI
JSBACH-YASSO-HIMMELI models
Soil and peatland emissions
13 March 2019 VERIFY-CHE GA, Reading 15
Mineral soils both
emit and take up
methane
Peatland emissions
are predominantly
in northern Europe
Inland water body emissions
13 March 2019 VERIFY-CHE GA, Reading 16
CH4 and N2O fluxes from inland water-systems• Computes fate of terrestrial-derived & in-situ produced C along the flow path
• Relies on estimates of river (Lauerwald et al., 2015), reservoir (GranD) and
lake (HydroLAKES) volumes and surface areas available for gas exchange
• Model inputs: Estimates of terrestrial C and nutrient inputs
13 March 2019 VERIFY-CHE GA, Reading 17
Top-down methods
for estimating CH4 and N2O emissions
Inversion method
13 March 2019 VERIFY-CHE GA, Reading 18
x0 xi
yobs
ymod = H(xi)
∆y = ymod–yobs
J′(xi+1) = B-1(xi+1 – x0) +
HadjR-1(∆y) = 0
prior
fluxes
best guess
estimate
modelled mixing
ratio
model errorobserved
mixing ratioiterate
xa
posterior
fluxes
end iteration once
reach certain criteria
H = transport and chemistry model
Hadj = adjoint transport model equiv. to (H′(x))T
B = prior error covariance matrix
R = obs. error covariance matrix
estimate xi+1 by solving:
Based on Bayesian statistics
Atmospheric network
Sites measuring CH4 and N2O in Europe
13 March 2019 VERIFY-CHE GA, Reading 19
●
●
●
●
●
●
●●
●
●
●
●●
●
●
●
●
●
●●● ●
●●
●●
●
●
●
●
●
●
●
40
50
60
−10 0 10 20 30
Type●
●
in−situ
flask
Emission estimates: CH4Results 1st year inversions
13 March 2019 VERIFY-CHE GA, Reading 20
Preliminary results
▪ Model: TM5-4DVAR
▪ Global 6°×4°, Europe1°×1°
▪ Period: 2005-2016
▪ Priori estimate: GCP-CH4
13 March 2019 VERIFY-CHE GA, Reading 21
Results 1st year inversions
Emission estimates: CH4
Preliminary results
▪ Model: FLEXPART-ExtKF
▪ Resolution: 0.5°×0.5°
▪ Period: 2005-2016 (shown for 2010)
▪ Priori estimate: based on EDGAR
15°W 5°W 5°E 15°E 25°E 35°E
33°N
43°N
53°N
63°N
73°N
0
20
40
60
80
100
15°W 5°W 5°E 15°E 25°E 35°E
33°N
43°N
53°N
63°N
73°N
0
20
40
60
80
100
A priori flux (mg m-2 day-1)
A posteriori flux (mg m-2 day-1) A posteriori – a priori flux (mg m-2 day-1)
13 March 2019 VERIFY-CHE GA, Reading 22
Emission trends: CH4Preliminary results
▪ Model: TM5-4DVAR
▪ Global 6°×4°, Europe1°×1°
▪ Period: 2005-2016
▪ Priori estimate: GCP-CH4
All observations assimilated as available
(i.e. not continuous for all stations over
2005-2016 period)
Emission estimates: N2O
Results 1st year inversions
13 March 2019 VERIFY-CHE GA, Reading 23
▪ Model: FLEXINVERT+
▪ Resolution: 0.5°×0.5°
▪ Period: 2005-2015
▪ Priori estimate: EDGAR-v4.32
15°W 5°W 5°E 15°E 25°E 35°E
30°N
40°N
50°N
60°N
70°N
0.000
0.005
0.010
0.015
0.020
0.025
0.030
15°W 5°W 5°E 15°E 25°E 35°E
30°N
40°N
50°N
60°N
70°N
0.000
0.005
0.010
0.015
0.020
0.025
0.030
Mean prior emissions (gN m-2 d-1) Mean posterior emissions (gN m-2 d-1)
National emissions: N2O
13 March 2019 VERIFY-CHE GA, Reading 24
N2O
(G
gN
/y)
0
50
100
150
Iberia P
enn.
Fra
nce
Germ
any
Benelu
x
UK
Fennosc
andia
Italy
Aust
ria+S
wis
sPola
nd
Hung+C
zech
+S
lova
k
N2O
(G
gN
/y)
0
50
100
150
Iberia P
enn.
Fra
nce
Germ
any
Benelu
x
UK
Fennosc
andia
Italy
Aust
ria+S
wis
sPola
nd
Hung+C
zech
+S
lova
k
Mean annual emission
2005-2009 (GgN y-1)
Mean annual emission 2010-
2015 (GgN y-1)
prior
posterior
Detection of hotspots: N2O
• Only <1% land area responsible for 20% of
emissions
• Hotspots defined as >99.5 percentile emissions
• Hotspots largely from nitric and adipic acid
production (sector 2B)
13 March 2019 VERIFY-CHE GA, Reading 25
0
500
1000
1500
2000
N2O
(G
gN
y-1) Total
prior
posterior
0
100
200
300
400
500
N2O
(G
gN
y-1)
2006 2008 2010 2012 2014
Hotspots
15°W 5°W 5°E 15°E 25°E 35°E
30°N
40°N
50°N
60°N
70°N
0.000
0.005
0.010
0.015
0.020
0.025
0.030
Total Land Emissions (GgN y-1)
Posterior total
15°W 5°W 5°E 15°E 25°E 35°E
30°N
40°N
50°N
60°N
70°N
0.000
0.005
0.010
0.015
0.020
0.025
0.030EDGAR 2B
Research & Development
• Using CH4 retrievals from satellite instrument TROPOMI aboard
Sentinel 5P (launched Oct-2017)
• High-resolution 7×7 km
13 March 2019 VERIFY-CHE GA, Reading 26
Summary and conclusions
➢ Bottom-up methods have very large uncertainties on
some sources of CH4 and N2O leading to very large
overall uncertainties
➢ Potential for top-down methods to contribute to improved
emission estimate for these species
➢ Preliminary inversions show slightly higher CH4 and N2O
emissions for EU28 compared to UNFCCC reports
➢ Both CH4 and N2O show a small decreasing trend over
2005-2016
13 March 2019 VERIFY-CHE GA, Reading 27