A Carbon Cycle Data Assimilation System at LSCE
A Carbon Cycle Data Assimilation System at LSCE using multiple data streams (CARBONES / GEOCARBON EU-project ) Philippe Peylin, Natasha Mac. Bean, Cédric Bacour, Sébastien Leonard, Vladislav Bastrikov, Fabienne Maignan, Sylvain Kuppel, Diego Santaren, Frédéric Chevallier, Patricia Cadule, Philippe Ciais, Jonathan Barichivich, Catherine Ottle Laboratoire des Sciences du Climat et de l’Environnement, Paris, France. & CARBONES / GEOCARBON projects & DATA providers 1
The LSCE - CCDAS The need for Carbon Cycle Data Assimilation Systems 2
Fate of Anthropogenic CO 2 Emissions (20002008) 1. 4 Pg. C y-1 4. 1 Pg. C y-1 45% + 7. 7 Pg. C y-1 3. 0 Pg. C y-1 29% 26% 2. 3 Pg. C y-1 Le Quéré et al. 2009, Nature-geoscience; Canadell et al. 2007, PNAS, updated 3
Atmospheric inversions: IAV Land Ocean Comparison of 11 inversions (RECCAP) JENA_s 96 LSCE_var LSCE_an CTrac_US CTrac_EU C 13 CCAM C 13 MATCH TRCOM RIGC JMA NCAM Peylin et al. 2013 BG 4
Atmospheric inversions: IAV Comparison of 11 inversions (RECCAP) Strengths: Ocean • Include all processes • Acurate at large scale JENA_s 96 • Land inter-annual variability appears robust Land LSCE_var LSCE_an CTrac_US Weaknesses: CTrac_EU • No insight on the processes C 13 CCAM • Poor regional constraint C 13 MATCH • Land / ocean partition is not robust TRCOM RIGC • No prediction capabilities JMA NCAM Peylin et al. 2013 BG 5
Role of land surface models IPSL Land surface models and dynamic global vegetation models are used to: • Monitor long-term trends in carbon, water, energy, vegetation • Predict changes into the future under new climate and greenhouse gas regimes Change in global biomass historical future <-uncertainty-> • Attribute the causes of trends and variability LPJ Jones 2013 6
Needs for C-Cycle Data Assimilation System Data streams Atm. data C flux to atmosphere (Gt. C/yr) Large uncertainty from land to predict global C-balance (C 4 MIP) Data Assimilation Optimized ecosystem models reduce the spread ? Improve: Ø Process understanding Ø Uncertainty estimates Ø Future climate predictions 7
The LSCE - CCDAS Description of the ORCHIDEE LAND SURFACE MODEL 8
Dynamic Global Vegetation Model ORCHIDEE Simulates the Energy, Water and Carbon balance Land component of the IPSL Earth System Model 9
Main processes Temperature Precipitation Air Humidity CO 2 Concentration Radiation Wind Speed Solar and infra-red Evapotranspiration Air Turbulence Net Photosynthesis CO 2 Flux Convection of dry heat Interception by the canopy Growth & Maintenance Respiration Allocation of the assimilates Surface runoff litter Surface Temperature Infiltration, storage, drainage Carbon Budget & nutriments 10
Surface description : a tile approach p A mosaïc of vegetation Land cover map • 13 different Plant functional types 11
Example of dominant PFT map 12
Plant Functional Types p The same set of equations governs C/W/E dynamics p But parameter values differ among PFTs 13
“Slow biogeochemical” Processes 14
“Slow biogeochemical” Processes • Phenology - Budburst based on GDD, soil water. . . • Senescence: Based on Leaf age, Temp. . . • Carbon Allocation: • • 8 pools of living biomass 4 litter pools and 3 soil carbon pools (CENTURY) • Autotrophic respiration: Maintenance & Growth • Heterotrophic Respiration • Fire module (SPITFIRE) • Turnover : death of plants, etc. 15
Biomass allocation 16
Hydrological Processes in ORCHIDEE 17
Hydrological Processes in ORCHIDEE • Partition of throughfall between infiltration and runoff • Water fluxes in soils (soil moisture and drainage) • Routing of runoff into river discharge • Human pressures, e. g. irrigation • Interactions with floodplains (fluxes and storage) • Wetlands • Snow pack processes • Permafrost (freeze/thaw in the soil) • Interactions with groundwater tables (fluxes and storage) 18
Driving data • Meteorological forcing (temp. , precip. , air humidity, surface pressure, wind speed, short- and longwave radiation) • Atmospheric CO 2 • Vegetation type (PFT) (when not using DGVM) • Soil Type • Land Cover Change 19
Parameters 20
ORCHIDEE model: current status Forest management module Nitrogen cycle Crops Fires Managed grass Forest Bare soil / desert Natural grass Multi-layer soil hydrology’ Assimilation Of variables Temperate Crops Tropical crops Modules implementation grassland - Generalization of PFT concept (number not limited) - A 11 -layer hydrological scheme - Scientific documentation 21
The LSCE - CCDAS Description of the LSCE - CCDAS 22
Structure of the LSCE CCDAS Forcing data Assimilation data Satellite data Atmos. Conc. Flux Tower Meteo. data IGBP LC ND EE A D L HI C OR Validation data Fossil Fuel & Biomass Burning fluxes Ocean flux Model OCVR Atm osp h LM DZ CCDAS Carbon Cycle Data Assimilation System CO 2 vertical Profiles Forest & Soil C stock Satellite data Ocean p. CO 2 data Optimized model parameters Carbon fluxes & pools (values & uncertainties) 23
Structure of the LSCE CCDAS PFT composition ecosystem parameters initial conditions climate NEE, LE, (H) J(X) parameters (X ) M(X) Yflux biomass data Optimizer BFGS J(X) and d. J(X)/X flux tower measurements Yf. APAR satellite f. APAR Atm CO 2 § Cost function: § Iterative minimization using either: - Variational approach (with Tangent Linear model for DJ/dx) - Monte Carlo approach 24
- Slides: 24