Tropical forest response to elevated CO2: Model-experiment integration at the AmazonFACE site.

Abstract:

The terrestrial biosphere’s response to current and future elevated atmospheric carbon dioxide (eCO₂) is a large source of uncertainty in future projections of the C cycle, climate and ecosystem functioning. In particular, the sensitivity of tropical rainforest ecosystems to eCO_­2 is largely unknown even though the importance of tropical forests for biodiversity, carbon storage and regional and global climate feedbacks is unambiguously recognized. The AmazonFACE (Free-Air Carbon Enrichment) project will be the first ecosystem scale eCO₂ experiment undertaken in the tropics, as well as the first to be undertaken in a mature forest. AmazonFACE provides the opportunity to integrate ecosystem modeling with experimental observations right from the beginning of the experiment, harboring a two-way exchange, i.e. models provide hypotheses to be tested, and observations deliver the crucial data to test and improve ecosystem models. We present preliminary exploration of observed and expected process responses to eCO₂ at the AmazonFACE site from the dynamic global vegetation model LPJ-GUESS, highlighting opportunities and pitfalls for model integration of tropical FACE experiments. The preliminary analysis provides baseline hypotheses, which are to be further developed with a follow-up multiple model inter-comparison. The analysis builds on the recently undertaken FACE-MDS (Model-Data Synthesis) project, which was applied to two temperate FACE experiments and exceeds the traditional focus on comparing modeled end-target output. The approach has proven successful in identifying well (and less well) represented processes in models, which are separated for six clusters also here; (1) Carbon fluxes, (2) Carbon pools, (3) Energy balance, (4) Hydrology, (5) Nutrient cycling, and (6) Population dynamics. Simulation performance of observed conditions at the AmazonFACE site (a.o. from Manaus K34 eddy flux tower) will highlight process-based model deficiencies, and aid the separation of uncertainties arising from general ecosystem responses and those responses related to eCO₂.