MODELING THINNING REGIMES WITH A FUZZY-LOGIC CONTROLLER

A very important issue in the development of a tree growth simulator is the implementation of a thinning model. This model should be suitable to represent formally defined thinning regimes as well as human decision making, which seems to be very ill structured in the context of thinning practice. In addition the model should be capable of giving a quite exact prognosis of real thinning regimes which have been documented on long term observation experimental plots. For this reason a thinning model is designed that can be implemented into distance dependent single tree growth simulators. The most central modul of this thinning model is a fuzzy logic controller which consists of a merely simple rule base with no more than seven rules. The only input variables are a relative tree diameter and a distance dependent competition index, the output variable prepares the thinning decision of a single tree. The rule base is designed using formally defined thinning instructions, empirical data from experimental plots and simulation results. The evaluation of the rule base is maintained by applying the max-prod-inference, the defuzzification of the resulting thinning decision follows the center of gravity method. The thinning model is validated using independent data from experimental plots. The simulation results give evidence of the good estimations that can be achieved using the model to predict the diameter distribution of the thinned trees in comparison with thinning practice on experimental plots. The fuzzy logic controller shows its computational efficiency and promises flexibility to serve as a mechanism which is suitable to model imprecise human decision making in ill structured situations as can be observed in thinning practice.