This thesis is written in order to create a model and investigate it with sensitivity analysis; the model will give a better understanding of the carbon sequestration dynamics after forest clearing for agricultural practices. Soil carbon sequestration is a very slow process, generally measured on a time scale of decades. In this thesis we study the chronosequence approach, which has as a main advantage that the observations to be carried out need not to cover a long time period but can be done in weeks only. With the use of chronosequencing techniques an estimation can be given for the carbon sequestration as a function of time when certain land management systems are applied.
With the model we try to estimate the parameters of an exponential decay function, which we assume to be the model description of carbon stock after forest clearing. We generated data for several years after forest clearing under various assumptions, including a Normal or a Double Exponential distribution, and the variance of the data could be equal for all years or could be different for each year. Analyzing the generated data with the model is done by means of the Ordinary Least Squares technique, the Ordinary Least Absolute Value technique and the Maximum Likelihood technique.
The results of the study indicate that the chosen estimators of the parameters of the exponential decay function have good properties, although they are a bit biased. When the variance of the generated data was not constant and was analyzed with the Maximum Likelihood technique assuming not a constant variance, these results included several outliers which resulted in a lower average with high standard errors. The Maximum Likelihood technique following a Normal distribution with a constant variance can be applied best to the dataset in order to find the optimal estimations of the parameters.