Natural and drought scenarios in an east central Amazon forest: Fidelity of the Community Land Model 3.5 with three biogeochemical models

12 de março de 2011

mar 12, 2011

Koichi Sakaguchi, Xubin Zeng, Bradley J. Christoffersen, Natalia Restrepo‐Coupe, Scott R. Saleska, Paulo Brando

Recent development of general circulation models involves biogeochemical cycles: flows of carbon and other chemical species that circulate through the Earth system. Such models are valuable tools for future projections of climate, but still bear large uncertainties in the model simulations. One of the regions with especially high uncertainty is the Amazon forest where large‐scale dieback associated with the changing climate is predicted by several models.

In order to better understand the capability and weakness of global‐scale land‐biogeochemical models in simulating a tropical ecosystem under the present day as well as significantly drier climates, we analyzed the off‐line simulations for an east central Amazon forest by the Community Land Model version 3.5 of the National Center for Atmospheric Research and its three independent biogeochemical submodels (CASA’, CN, and DGVM). Intense field measurements carried out under Large Scale Biosphere‐Atmosphere Experiment in Amazonia, including forest response to drought from a throughfall exclusion experiment, are utilized to evaluate the whole spectrum of biogeophysical and biogeochemical aspects of the models.

Our analysis shows reasonable correspondence in momentum and energy turbulent fluxes, but it highlights three processes that are not in agreement with observations: (1) inconsistent seasonality in carbon fluxes, (2) biased biomass size and allocation, and (3) overestimation of vegetation stress to short‐term drought but underestimation of biomass loss from long‐term drought. Without resolving these issues the modeled feedbacks from the biosphere in future climate projections would be questionable. We suggest possible directions for model improvements and also emphasize the necessity of more studies using a variety of in situ data for both driving and evaluating land‐biogeochemical models.

Baixar (sujeito à disponibilidade)

Download (subject to availability)

Veja também

See also

Forest fragmentation, climate change and understory fire regimes on the Amazonian landscapes of the Xingu headwaters

Forest fragmentation, climate change and understory fire regimes on the Amazonian landscapes of the Xingu headwaters

Understory fire modeling is a key tool to investigate the cornerstone concept of landscape ecology, i.e. how ecological processes relate to landscape structure and dynamics. Within this context, we developed FISC—a model that simulates fire ignition and spread and...