|Title||Mechanistic modelling of the vertical soil organic matter profile|
|Source||Wageningen University. Promotor(en): Pavel Kabat, co-promotor(en): C. Beer; M. Reichstein; Marcel Hoosbeek. - Wageningen : Wageningen University - ISBN 9789461738288 - 190|
Earth System Science
|Publication type||Dissertation, internally prepared|
|Keyword(s)||organisch bodemmateriaal - bodemprofielen - modelleren - modellen - bayesiaanse theorie - soil organic matter - soil profiles - modeling - models - bayesian theory|
|Categories||Soil Science (General)|
Soil organic matter (SOM) constitutes a large global pool of carbon that may play a considerable role for future climate. The vertical distribution of SOM in the profile may be important due to depth-dependence of physical, chemical, and biological conditions, and links to physical processes such as heat and moisture transport. The aim of this thesis is to develop a dynamic and mechanistic representation of the vertical SOM profile that can be applied for large scale simulations as a part of global ecosystem and earth system models.
A model structure called SOMPROF was developed that dynamically simulates the SOM profile based on above and below ground litter input, decomposition, bioturbation, and liquid phase transport. Furthermore, three organic surface horizons are explicitly represented.
Since the organic matter transport processes have been poorly quantified in the past and are difficult to observe directly, the model was calibrated with a Bayesian approach for two contrasting temperate forest sites in Europe. Different types of data were included in the parameter estimation, including: organic carbon stocks and concentrations, respiration rates, and excess lead-210 activity.
The calibrations yielded good fits to the observations, and showed that the two sites differ considerably with respect to the relevance of the different processes. These differences agree well with expectations based on local conditions. However, the results also demonstrate the difficulties arising from convolution of the processes. Several parameters are poorly constrained and for one of the sites, several distinct regions in parameter space exist that yield acceptable fit.
In a subsequent study it was found that radiocarbon observations can offer much additional constraint on several parameters, most importantly on the turnover rate of the slowest SOM fraction. Additionally, for one site, a prognostic simulation until 2100 was performed using the resulting a posterioriparameter distribution, This showed that different parts of the SOM profile can respond differently to increasing temperatures and litter input.
In conclusion, the SOMPROF model, combined with the Bayesian calibration scheme, offers valuable insights into the relevance of the different mechanisms to the SOM profile. However, equifinality remains a challenge, particularly for distinguishing different SOM transport processes. Improved representation of liquid phase transport and incorporation of additional observations may reduce these problems. In the future, SOMPROF can be incorporated into a terrestrial ecosystem model and calibration results can be used when deriving parameter sets for large scale application.