Academic Journals Database
Disseminating quality controlled scientific knowledge

Uncertainties in carbon residence time and NPP-driven carbon uptake in terrestrial ecosystems of the conterminous USA: a Bayesian approach

ADD TO MY LIST
 
Author(s): Xuhui Zhou | Tao Zhou | Yiqi Luo

Journal: Tellus B
ISSN 0280-6509

Volume: 64;
Start page: 1;
Date: 2012;
Original page

Keywords: carbon residence time | carbon uptake | conterminous USA | inverse analysis | MCMC | terrestrial carbon cycle | uncertainties

ABSTRACT
Carbon (C) residence time is one of the key factors that determine the capacity of ecosystem C storage. However, its uncertainties have not been well quantified, especially at regional scales. Assessing uncertainties of C residence time is thus crucial for an improved understanding of terrestrial C sequestration. In this study, the Bayesian inversion and Markov Chain Monte Carlo (MCMC) technique were applied to a regional terrestrial ecosystem (TECO-R) model to quantify C residence times and net primary productivity (NPP)-driven ecosystem C uptake and assess their uncertainties in the conterminous USA. The uncertainty was represented by coefficient of variation (CV). The 13 spatially distributed data sets of C pools and fluxes have been used to constrain TECO-R model for each biome (totally eight biomes). Our results showed that estimated ecosystem C residence times ranged from 16.6±1.8 (cropland) to 85.9±15.3 yr (evergreen needleleaf forest) with an average of 56.8±8.8 yr in the conterminous USA. The ecosystem C residence times and their CV were spatially heterogeneous and varied with vegetation types and climate conditions. Large uncertainties appeared in the southern and eastern USA. Driven by NPP changes from 1982 to 1998, terrestrial ecosystems in the conterminous USA would absorb 0.20±0.06 Pg C yr−1. Their spatial pattern was closely related to the greenness map in the summer with larger uptake in central and southeast regions. The lack of data or timescale mismatching between the available data and the estimated parameters lead to uncertainties in the estimated C residence times, which together with initial NPP resulted in the uncertainties in the estimated NPP-driven C uptake. The Bayesian approach with MCMC inversion provides an effective tool to estimate spatially distributed C residence time and assess their uncertainties in the conterminous USA.
Why do you need a reservation system?      Save time & money - Smart Internet Solutions