Applying CCB to predict management change affected long-term SOM turnover of the Extended Static Fertilization Experiment in Bad Lauchstadt

Long-term field experiments (LTEs) are invaluable in improving understanding of soil organic matter (SOM) turnover, as some of the involved processes have proceeded over centuries. Prediction of such slow carbon fluxes depends especially on the initialization of slow-reacting model pools and requires monitoring for a very long time for evaluation. This study reports soil organic matter (SOM) modelling covering more than 100 years for a dataset of 36 treatments of the "Extended Static Experiment" that is based on the classical "Static Experiment Bad Lauchstadt." The experimental scheme includes different fertilization treatments where the plots are arranged in two field stripes. A special feature of this experiment is the management change for each stripe after seven decades of continuous cultivation that includes a considerable alteration of organic amendments on different SOM levels. The Candy Carbon Balance (CCB) model was calibrated for soil organic carbon (SOC) dynamics regarding initial SOC content and one parameter describing the physical protection of SOM. Calibration was performed for each plot individually, for each stripe, and for the whole experiment, resulting in a relative root mean square error (rRMSE, (-)) of 0.050, 0.058 and 0.064, respectively. Model predictions for soil organic nitrogen (SON) dynamics were satisfactory using the calibration results for SOC and the average C/N ratio from all observations within this dataset (11.8) for model initialization. The model error for SON prediction reached from rRMSE = 0.09 (whole experiment) over rRMSE = 0.085 (each stripe) to rRMSE = 0.084 (each plot). Calibration results suggest some heterogeneity between the two field stripes. Restarting the model after the management change in 1978 using the plot-specific calculated SOC value from previous simulations for the whole time period provided different results of predicted SOC trends and values, suggesting that the precondition of SOM in steady state was still not fulfilled after seven decades. Highlights SOC and SON were modelled for 36 treatments over more than one century. Model restarts after seven decades indicated that SOM pools were not in steady state. Modelling SON required only an additional input of the average C/N ratio of SOM Effect of serious fertilization changes can be well predicted by the tested model setup