Forests have two functions; one is for economy the other is for public service. The problem in relation with forest policy is how to make these functions optimally and contemporaneously harmonizing with each other in order to get the maximum utility (or value) from forests. In this paper, KISINE's theory of optimal usage of forests is used to derive a nonlinear programming method (with restrictive conditions), then the method is used to empirically analyse OSAKA's forests. As the results, the values of optimal growth period and optimal division of whole forests area between the economic forests and the public service forests making the utility of OSAKA's forests maximum are found, also the efficiency of the theory is empirically verified.

View full abstract

Although Systems simulation has been widely used in agricultural research, it is still morean art than a science because of many problems such as uncertainty of model and limitation of its application. In this study, such problems concerning systems simulation were clarified and approaches to them were discussed. Three sources of model uncertainty were considered; natural variability, estimation error in model parameters and ambiguity of model structure. First, it was suggested that the natural variability can be reduced by collecting data from wide range and/or by use of adjustment coefficients for representing its variability quantitatively. Secondly, for a large-scale and complex model, the methods of evaluating effects of estimation error in model parameters on the model outputs were proposed and compared: sensitivity analysis and Monte Calro error analysis. The results indicated that Monte Calro error analysis using path coefficients as a measure for representing the relationship between model parameters and model outputs is the most adequate method among measures shown in this study, because use of path coefficient enables to consider not only interaction effects but also magnitude of parameter variabilities. Thirdly, with regard to ambiguity of model structure, the problems of use of multiple correlation coefficient as a measure of fitness of an equation were discussed and it was recommended that simulation model should be constructing using theoretical equations instead of empirical ones as possible. Lastly, limitation of using systems simulation with a deterministic model as a tool of decision making was pointed out and an objective method of evaluating alternatives in decision making was proposed, although it is available only when variance of a model output is known. Since all approaches shown in this study are not completed, it was indicated that further theoretical study on systems simulation would be required.

View full abstract