Carbon dynamics in the Amazonian Basin: integration of eddy covariance and ecophysiological data with a land surface model. (Special issue on LBA data model.)

摘要

Information contained in eddy covariance flux tower data has multiple uses for the development and application of global land surface models, such as evaluation/validation, calibration and process parameterization for carbon stocks and fluxes. In this study, we combine Large Scale Biosphere-Atmosphere Experiment in Amazonia (LBA) project data collected from a network of eddy covariance flux towers deployed across the Amazonian basin to improve the carbon stocks and flux estimated by a land surface model, the Integrated Science Assessment Model (ISAM). We evaluate the key model parameters for carbon allocation factors, maintenance respiration as well as the autotrophic respiration using the LBA site measurements. Our model results show that the total biomass ranges from 0.7 kg C m-2 yr-1 for the pasture site to 20.8 kg C m-2 yr-1 for the forest site. The ISAM estimates for GPP and NPP are well within the uncertainty range of the site measurement data. Also, the results revealed that all, but one forest site have lower net primary productivity (NPP) to gross primary productivity (GPP) ratio (NPP/GPP=0.4) compared to the savanna and the pasture sites (NPP/GPP=0.5). This is because savanna and the pasture sites experienced the longest dry season and plants growing in such environmental conditions have stronger efficiency to store carbon compared to forests. The forest evergreen site (Km67) has a higher measured NPP/GPP ratio (0.5), because of higher carbon accumulation. Soil carbon is lowest for the pasture site (Km77) (7.2 kg C m-2 yr-1) and highest for the forest site (Km34) (12.8 kg C m-2 yr-1). The model results suggest that all the forest sites are a net sink for atmospheric CO₂, while the savanna (PDG) and pasture (FNS) sites are neutral and another pasture site (Km77) is net source for atmospheric CO₂. Meanwhile, the model results highlight the importance of the LBA site data to improve the model performance for the tropical Amazon region. The study also suggest the need for a network of long-term monitoring plans to measure changes in the vegetation and soil carbon biomass at the local and regional levels. Such programs will be necessary to make reliable global carbon emissions estimates.