Application of the Bayesian calibration methodology for the parameter estimation in CoupModel

摘要

This study provides results for the optimization strategy of highlyparameterized models, especially with a high number of unknown inputparameters and joint problems in terms of sufficient parameter space.Consequently, the uncertainty in model parameterization and measurementsmust be considered when highly variable nitrogen losses, e.g. N leaching,are to be predicted. The Bayesian calibration methodology was used toinvestigate the parameter uncertainty of the process-based CoupModel.Bayesian methods link prior probability distributions of input parameters tolikelihood estimates of the simulation results by comparison with measuredvalues. The uncertainty in the updated posterior parameters can be used toconduct an uncertainty analysis of the model output. A number of 24 modelvariables were optimized during 20 000 simulations to find the "optimum"value for each parameter. The likelihood was computed by comparingsimulation results with observed values of 23 output variables includingsoil water contents, soil temperatures, groundwater level, soil mineralnitrogen, nitrate concentrations below the root zone, denitrification andharvested carbon from grassland plots in Northern Germany for the period1997–2002. The posterior parameter space was sampled with the Markov ChainMonte Carlo approach to obtain plot-specific posterior parameterdistributions for each system. Posterior distributions of the parametersnarrowed down in the accepted runs, thus uncertainty decreased. Results fromthe single-plot optimization showed a plausible reproduction of soiltemperatures, soil water contents and water tensions in different soildepths for both systems. The model performed better for these abioticsystem properties compared to the results for harvested carbon and soilmineral nitrogen dynamics. The high variability in modeled nitrogenleaching showed that the soil nitrogen conditions are highly uncertainassociated with low modeling efficiencies. Simulated nitrate leaching wascompared to more general, site-specific estimations, indicating a higherleaching during the seepage periods for both simulated grassland systems.