A model was developed to determine the flow size of intake water of headworks by considering the policies of the operator. As the flow size determination process involves various assumptions and considerations which are ambiguous and vague in the real world, the model was developed using the concept of fuzzy set theory. In this research, the model was developed with a principle to reflect the operator's thinking on the physical nature of the system regarding decision-making process on flow size determination. Questionnaire analysis showed that the factors for determining the flow size during increasing operations were different from those during decreasing operations. Hence two different algorithms for flow size increasing and decreasing operations were found to be necessary and gave smaller percentage of errors. The best ranges (minimum and maximum) of different factors were detected by a systematic approach to avoid the tedious effort. The model showed a good agreement between the actual and calculated flow size. Relative error for change in actual observed flow size (ΔQo) and calculated flow size (ΔQc) was 32.52%. Errors calculated by the model for increasing and decreasing operations were 6.06% and 11.85%, respectively. The model yielded better results than linear multiple regression analysis. Development of this model by considering the operator's feelings and measures on the physical nature of the system using fuzzy set theory was successful for determining the intake flow size.

View full abstract

Since the Geographic Information System (GIS) is useful for conducting spatial data analysis and integrations, it is often used in land evaluations. To apply GIS to the evaluation of land, logical and accurate criteria are must be used to develop evaluation models. Any evaluation model should express the whole relation between all the factors related to the evaluation. Though weighting or ranking methods are often used as GIS evaluation models, the weight and ranking values are subjectively determined by each operator. Therefore, it is difficult to objectively prove their validity. This study applies a neural network as a GIS evaluation model to evaluate the suitability of grassland development in Tochigi Prefecture. The neural network is an example of artificial intelligence technology, and it adjusts itself to fit the object. The network, which lets factors relate to results, is represented by a mathematical function like Sigmoid function. In this study, a neural network, which evaluates the suitability of grassland development by the natural factors, is produced. The relevant factors considered in the evaluation are topography, slope, elevation and soil productivity. As the result of trials to decide the number of unit on hidden layer, the neural network, which has 11 units on it, performs a good classification to supervised data sets. Using This network in conjunction with GIS, the land evaluation map for grassland development is produced by GIS.

View full abstract

In this paper, we used, as an example, water quality data of nine chemical compositions which were investigated at a river running urbanizing farmland in suburb, at 28 serial times over two years and nine fixed spots from upper-stream to down-stream. We tried to apply statistical methods to analyze time-sphere variations of environmental data in the 1) ～ 4) processing; 1) Applying principal components analysis for nine variates, the nine variates are summerized into a few synthetic characteristics. 2) By regression analysis, applying orthogonal polynomials, of principal component scores of each time to distance of nine spots from upper-stream, sphere effects are separated into orthogonal regression effects (0 degree - 8 degree). 3) By periodical regression analysis, applying Furie progression, of each orthogonal regression effect. to relative days from the first investigation time, periodical effects are analyzed. 4) Summarizing the results in an analysis of variance table, the contribution and the importance of analyzed time-sphere effects are evaluated. By analysis of the example data in accordance with this processing, two principal components (synthetic variates) meaningful in water quality were found. On each principal component, variations of among times, among spots and interaction times x spots were analyzed into more detail effects, and from contributions of those effects, an aspect of time-sphere variation was grasped. Also, it was shown that information about singular data isolating from periodicity were found on the way of analysis. From the results, we proposed that statistical approach was one of the effective technologies in order to analyze time-sphere variations of environmental data.

View full abstract