Carbon (C) turnover time and increases in NPP together quantify the capacity for C storage in plant and soil pools. Accurate spatially distributed estimates of C turnover time over the conterminous USA are especially critical to the understanding of terrestrial C sequestration and prediction of climate change, when NPP changes were relatively well qualified. However, the spatial patterns of C turnover time have not been quantified for the conterminous USA. It is also largely unknown what probabilistic densities are of C turnover time and sequestration potential. In this study, the Bayesian probability inversion and Markov Chain Monte Carlo (MCMC) technique were applied to a regional terrestrial ecosystem (TECOR) model to generate probabilistic density function (PDF) of C turnover time based on 12 observed data sets of C pools and fluxes.
Results/Conclusions:
Ecosystem C turnover time and sequestration potential and their standard deviation (SD) were estimated using maximum likelihood estimates (MLEs) and SD of the parameters in the conterminous USA. Our results showed that over half of 22 parameters were well constrained by 12 data sets for each biome (totally 8). The poorly constrained parameters were attributable to either the lack of experimental data or the mismatch of timescales. Using MLEs or means of estimated parameters, ecosystem C turnover time ranged from 16.6 ± 1.8 (cropland) to 85.9 ± 15.3 years (Evergreen Needleleaf Forest, ENF) with an average of 56.8 ± 8.8 years. The C turnover times have highly spatial heterogeneity and their values largely depend on the vegetation type and climate condition. Along the latitude, C turnover time displayed a strong positive correlation (R2 = 0.91), suggesting that temperature may be the most important factor in influencing carbon turnover time. Under the current drive of increases in NPP (0.5% per year), terrestrial ecosystems in the conterminous USA sequestrates 0.22 ± 0.04 Pg C yr-1 with large portions in Mixed Forest, grassland, and cropland, which was similar to the estimation from inventory data. Our results suggest that the combination of Bayesian approach and MCMC inversion technique is an effective tool to estimate spatially distributed C turnover time and sequestration and their uncertainty in the conterminous USA.
