Integration of a Northern-Hemispheric Biomass Map Derived from Envisat ASAR and a Global Carbon Cycle Model
Thurner, Martin1; Beer, Christian2; Santoro, Maurizio3; Carvalhais, Nuno1; Schmullius, Christiane4
1Max Planck Institute for Biogeochemistry, GERMANY; 2Stockholm University, SWEDEN; 3Gamma Remote Sensing, SWITZERLAND; 4Friedrich-Schiller-University, GERMANY

Although Earth system models have been developed in order to investigate carbon cycle dynamics in the atmosphere, on land and in the oceans, we are still lacking reliable information on terrestrial carbon stocks. Improving our knowledge of current forest carbon stocks is critical because biomass provides important feedbacks on carbon fluxes and thus on climate. For those reasons, biomass is among the essential climate variables (ECV) identified by the Global Climate Observing System (GCOS). Complementing inventory data, only remote sensing has the potential to fulfill the need for consistent, high resolution biomass mapping over large areas.

A recently available northern-hemispheric forest growing stock volume (GSV) map retrieved from Envisat Advanced Synthetic Aperture Radar (ASAR) backscatter measurements allows for the integration with global carbon cycle models. Making use of a wood density and a biomass compartment database, a forest biomass density map at 0.01 degree resolution could be derived from the GSV product. At a regional scale, intercomparison of the map with inventory-based biomass products in Russia, the USA and Europe showed strong agreement (r2 = 0.70 - 0.90). The observation-based biomass map could be directly compared to wood carbon modeled within the Jena-Scheme for Biosphere-Atmosphere Coupling in Hamburg (JSBACH), a land surface scheme embedded into the Max Planck Institute Earth System Model (MPI-ESM). Plant functional type (PFT) specific biomass maps could be derived using land cover and biome distribution data. As a first application, these remote sensing based biomass data were used to adjust biomass turnover time in the model.

Since model performance was found to vary with PFT, a different turnover time was implemented for each PFT. The comparison of modeled and observation-based biomass showed an overestimation of modeled biomass for temperate broadleaf deciduous trees (PFT 7) and coniferous evergreen trees (PFT 8), while carbon density of coniferous deciduous trees (PFT 9) was underestimated on average (Fig. 1). In the JSBACH model, an adjustment of forest biomass turnover can have important consequences on litter flux, soil carbon and finally soil respiration and net ecosystem exchange (NEE). First results indicate a modification of NEE between -30 % and +30 % for 90 % of the study area. On average, NEE was found to be smaller compared to the previous model version with most important decreases occurring in boreal forests of Canada, Central Siberia and South Scandinavia and also in some temperate forests of Central and East Europe and in the South-East US.

The presented biomass map has the potential to significantly improve modeling of current carbon stocks and to reduce uncertainty in simulated carbon cycle dynamics. Further investigations will concentrate on identifying other factors (e.g. climatic conditions) contributing to observed and modeled biomass spatial patterns in addition to PFT. Since JSBACH lacks a direct influence of biomass on important processes like autotrophic respiration, an improvement of biomass effects on the carbon cycle in the JSBACH model is another possible future research area. Alternatively, different global carbon cycle models can be compared to the observation-based biomass map in order to test their performance.


Figure 1. Difference (left) and ratio (right) of modelled and observation-based wood carbon density [kg(C) m-2] per PFT at 0.5 degree resolution (pft7 = temperate broadleaf deciduous trees, pft8 = coniferous evergreen trees, pft9 = coniferous deciduous trees). Displayed values are limited to the range [-3, 3] (left) and [0, 3] (right) respectively.