B21F-0534
Global Evaluation of Vegetation Carbon Residence Times as Simulated by ISI-IMP2 Global Vegetation Models (GVMs) Using a Data Product based on Satellite and Eddy Covariance Flux Measurements

Tuesday, 15 December 2015
Poster Hall (Moscone South)
Tim Tito Rademacher1, Nuno Carvalhais2, Andrew D Friend3 and Martin Thurner2, (1)University of Cambridge, Geography, Cambridge, United Kingdom, (2)Max Planck Institute for Biogeochemistry, Jena, Germany, (3)University of Cambridge, Cambridge, United Kingdom
Abstract:
Global vegetation carbon residence times are, by some measures, the single largest source of uncertainty in simulations of the future terrestrial global carbon cycle (Friend et al., 2013). This paper has two main aims : (i) evaluate ISI-MIP2 Global Vegetation Models (GVMs) with regard to their capacity to reproduce observation-derived current global patterns in vegetation carbon residence time and (ii) compare spatial relationships of vegetation carbon residence time with climate in GVMs and observation.
Recently, satellite-based measurements of biomass and eddy covariance-based estimates of gross primary production were combined to generate a contemporary snapshot of global ecosystem carbon residence times, henceforth 'observation-derived' data, by (Carvalhais et al., 2014). Emergent patterns and spatial relationships with few climatic variables, e.g. precipitation and temperature, were analysed. The vegetation component of this observation-derived data product is used here to evaluate global fields of current vegetation carbon residence time simulated by the ensemble of Global Vegetation Models that participated in the Inter-Sectoral Impat Model Intercomparison Project Phase 2 (ISI-MIP2). As well as, direct comparisons using simple statistics, spatial relationships between climatic forcings and simulated vegetation carbon residence times in the models and data are also investigated.
Finally, building on the results of the previous two sections, data products for further model evaluation and key areas for model development are suggested.