Von der Messung zum Produkt:
Bes3mmung von Quellen und Senken mi9els Inverser Modellierung
Christoph Gerbig, Panagiotis Kountouris, Christian Rödenbeck (MPI-BGC, Jena),
Thomas Koch (DWD), Ute Karstens (ICOS-CP, Lund)
11. Klimatagung, DWD, Offenbach, 5. Okt. 2017
WRF-GHG: mesoscale simula3ons of CO 2
WRF-GHG simula3on Aug. 2006, 70 m level VPRM
Biosphere model
using satellite imagery + solar radia3on
Anthropogenic emissions
Lateral boundary
condi3ons from global model
• CO
2highly variable in atmospheric boundary layer
• Anthropogenic and biospheric signatures hard to separate
CarboScope regional inversion system
anthropogenic emissions
regional atmospheric transport
• STILT (Stochas3c Time Inverted Lagrangian Transport)
• Stochas3c turbulence
• Driven by meWields from ECMWF or WRF
• “Footprints”: sensi3vity of
observa3ons to upstream fluxes
CarboScope regional inversion system
prior biosph. fluxes + error cov.
regional atmospheric transport
• STILT (Stochas3c Time Inverted Lagrangian Transport)
• Stochas3c turbulence
• Driven by meWields from ECMWF or WRF
• “Footprints”: sensi3vity of
observa3ons to upstream fluxes
• VPRM (Vegeta3on Photosynthesis and Respira3on Model)
• MODIS reflectances (EVI + LSWI)
• Driven by T2m and radia3on from ECMWF or WRF
• Parameter op3miza3on using Ecosystem flux measurements
CarboScope regional inversion system
prior biosph. fluxes + error cov.
lateral boundary condi3on
anthropogenic emissions
regional atmospheric transport
observed concentra3ons
• EDGARv4.3 (Emissions Database for Global Atmospheric Research),
0.1 deg resolu3on, temporal profiles
• Combined with BP report to update to recent year
• Fuel type and IPCC category specific
Minimiza3on of differences by flux op3miza3on
CarboScope regional inversion system
prior biosph. fluxes + error cov.
lateral boundary condi3on
anthropogenic emissions
regional atmospheric transport
simulated concentra3ons
observed concentra3ons
S C M T UP 1.5 2.5 1.5 1.5 4
S: Near shore
C: Con3nental (surface) M: Mountain
T: Tall tower
UP: Urban polluted
CO2 Model-data mismatch error in ppm (for weekly 3me scales)
• 16 atmospheric sta3ons (2007) (Con3nuous measurements and analysed flask samples)
• Day3me 11-16 local (mountain: 23-04)
Inversion setup
Atmospheric observa3ons:
Prior error structure (derived from differences prior fluxes – flux observa3ons):
• Diagonal: 2.3 μmoles/m2/s (daily fluxes, 0.5x0.5 ° lat-lon)
• error correla3ons: 30 days, 100 km
=> error infla3on needed to obtain consistency with global inversions
0.3 GtC/yr for annual and domain wide aggregated prior error
• B1 case: Error infla3on (scaling of covariance matrix)
• S1 case: Error infla3on by adding a bias term (constant in 3me, respira3on
shape) Kountouris et al., 2016a ACPD
Pseudo data inversion – Country-scale C budget
• Successfully retrieved fluxes at monthly and na3onal scales
• Reduc3on in
Uncertain3es (prior ->
posterior) larger for countries with more observa3ons
Kountouris et al., 2016a ACPD
• Successfully retrieved fluxes at monthly and na3onal scales
• Reduc3on in
Uncertain3es (prior ->
posterior) larger for countries with more observa3ons
Kountouris et al., 2016a ACPD
Pseudo data inversion – Country-scale C budget
Small scale structure:
vegeta3on coverage, radia3on, temperature (a priori)
Larger scale
correc3ons from
atmospheric constrain
Innova3on:
posterior – prior
Real data inversion 2007
Daily averaged flux es3mates in gC d
-1m
-2Kountouris et al., 2016b ACPD
Real data inversion 2007: Valida3on
Extrac3ng posterior fluxes at Eddy Covariance Flux sites comparison to independent flux observa3ons
(case B2, BIOME-BGC prior fluxes not dependent on flux observa3ons)
Kountouris et al., 2016b ACPD
CH 4 inversion – no prior
• The inversion system
• Synthe3c data inversion
• Real data inversion
• Uncertainty in atmospheric transport
• Summary
Bergamaschi, Karstens, et al., 2017 ACPD Con3nuous CH4 measurements
Flask sample CH4 measurements
• The inversion system
• Synthe3c data inversion
• Real data inversion
• Uncertainty in atmospheric transport
• Summary
CH 4 inversion – prior (EDGAR + E-PRTR)
Bergamaschi, Karstens, et al., 2017 ACPD
CH 4 inversions – mult. models & setups
Bergamaschi, Karstens, et al., 2017 ACPD
Observa(on based GHG budgets - Year n
WP 3 Climate forcing at 10 km
Official inventories GHG budgets - Year n
WP 2-4 BoDom-up GHG flux maps from
Ecosystem models Emission models
WP 2-4 Atmospheric transport fields
WP2-4 Top-down GHG flux maps from
Atm. inversions WP 2-4 Atm. GHG
concentra(on data
WP1 First official GHG na(onal inventories
WP5 synthesis of observa(on based
es(mates
WP6 Reconcilia(on Na(onal and EU
GHG budget
fact sheets COP
Regular GHG Monitoring within VERIFY (H2020)
Herausforderungen im Hinblick auf die quan3ta3ve Verwendung in den THG Berichten
• Trennung von biogenem und anthropogenem CO2 - Verwendung hochaufgelöster Satellitendaten - Verwendung von 14C in CO2 nahe “Hotspots”
- Gemeinsame Inversion von CO2, CO und CH4
• Emissionsmodellierung auf na3onaler Ebene - Räumliche Verteilung der Emissionen
- Sektoren- und Brennstoff- spezifisch
- Jährliche Updates, aktuelle zeitliche Profile für Ak3vitäten - Fehlerquan3fizierung für Modellparameter
- Einbindung in Inversion, Op3mieren der Parameter anstelle der Flüsse
• Unsicherheiten in der Transportmodellierung – Ver3kaltransport - Verwendung nächtlicher Ver3kalprofile an hohen Messtürmen - tropospherische Profildaten (IAGOS)
- Nutzung der Mischungsschichthöhen (PBLH) von Ceilometer
• Unsicherheiten in der Transportmodellierung – Repräsenta3onsfehler - Sta3onen in der Nähe starker Punktquellen
- Höhere Auflösung der Modelle
- Anbindung der Inversionsmodelle an ICON als atmosphärisches Modell