The Bayesian probability inversion and a Markov chain Monte Carlo (MCMC) technique was applied to CLM-CASA model under equilibrium conditions to estimate parameter distributions that reproduce most variability in the observed soil carbon (C). The optimized parameters were associated with soil C turnover rates and partitioning coefficients among litter and soil C pools. Most sensitive parameters to the observed soil C were temperature sensitivity, and parameters associated with effect of soil clay content on carbon partitioning among the C pools. The poor constraint of some parameters was attributed to lack of information in soil C data for those parameters; and skews of some parameter distributions against their maxima or minima were attributed to imperfections in model formulation. The CLM-CASA' model with calibrated parameters explained 41% of the global variability in the observed soil C from IGBP-DIS global gridded estimates, which was substantial improvement from the initial 27%.
Publication Hararuk O, Xia J, Luo Y (2014) Evaluation and Improvement of a Global Land Model against Soil Carbon Data Using a Bayesian MCMC Method. Journal of Geophysical Research: Biogeosciences, 2013JG002535
Data assimilation algorithm (Download Files)
Note: To run the routine one has to run the "main_script.m".