110 likes | 332 Vues
Part I. Uncertainties of modelling GPP over Europe: A systematic study on the effects of using different drivers and carbon cycle models.
E N D
Part I Uncertainties of modelling GPP over Europe: A systematic study on the effects of using different drivers and carbon cycle models Martin Jung, Mona Vetter, Nicolas Viovy, Guerric Le Maire, Soenke Zaehle, Alberte Bondeau, Sebastiaan Luyssaert, Frauke Feser, Markus Reichstein, Galina Churkina, Martin Herold, Martin Heimann Part II Confronting model simulations with observations: Preliminary results from the site level comparison Martin Jung1, Mona Vetter1,Galina Churkina1, Martin Herold2,Nicolas Viovy3, Soenke Zaehle3, Alberte Bondeau4, Frauke Feser5, Martin Heimann1 1 Max Planck Institute for Biogeochemistry, Jena, Germany 2 ESA GOFC GOLD PROJECT OFFICE, Department of Geography, Friedrich Schiller University Jena 3 Laboratory for Climate Sciences and the Environment (LSCE), Joint Unit of CEA-CNRS, Gif-sur-Yvette, France 4 Potsdam Institute for Climate Impact Research (PIK), Potsdam, Germany 5GKSS-Forschungszentrum Geesthacht GmbH, Germany
PART I Uncertainties of modelling GPP over Europe: A systematic study on the effects of using different drivers and carbon cycle models Objective: Effects of different driver data set options and carbon cycle models on GPP simulations over Europe • Magnitude • Spatial pattern • Interannual variation • 3 land cover maps • 3 spatial PFT resolutions • 2 meteorological data sets • Biome-BGC • LPJ • Orchidee
Modeling Strategy Reference (REF): Biome-BGC+REMO+SYNMAP+0.25°fract. Alternative Realisation (AR): One component changed from reference
mean(|ARi-REFi|) mean(REF) 1 – R2(AR, REF) Quantifying the effects Magnitude: Relative mean absolute difference Spatial variance not explained Spatial: Interannual variance not explained (1981-2000) mean(1 – R2(ARi, REFi)) Temporal: AR; REF: pixel based 1981-2000 mean i: Pixel index
Results Hierarchy of effects: Land cover map spatial resolution meteorology model
Biome-BGC vs. LPJ vs. ORCHIDEE How come the models are so different? What are the components to model GPP? • Soil vegetation atmosphere transfer scheme (SVAT) • Leaf level photosynthesis model (Farquhar) • Stomata conductance scheme • Upscaling scheme • LAI scheme • Phenology scheme Associated with PFT specific parameters
Conclusions How large is the effect of using different … • Land cover maps small • Land cover resolutions relatively small • Meteorological data sets large • Carbon cycle models very large Implications of uncertainties of meteorological input data : • It is an issue esp. when studying interannual variations of carbon fluxes with models • Model intercomparison initiatives should use the same meteo drivers Carbon cycle models should be improved in terms of simulating light harvesting (LAI, phenology) Thinking of alternatives to fixed generalised PFT parameterisations Better understanding of model’s uncertainties with data-model comparison strategies!
PART II Confronting model simulations with observations: Preliminary results from the site level comparison Objective: Testing the model’s adequacy to simulate carbon fluxes
Do the models reproduce observed patterns along spatial environmental gradients? Northern sites (> 50° latitude)