Abstract
The rational division of South Lake (Lake Biwa) into a few areas from the viewpoints of water quality similarity is attempted on the basis of horizontal distribution patterns of pollutants. A new statistical value, which is calculated by the difference of two concentrations measured at different points, is developed and calculated at every combination among 22 survey points to evaluate quantitatively the similarity of water quality between any two locations. The survey points are classified into 5 groups, based on the calculated value that is high in case of two survey points in the same group but very low in the different groups. Consequently, 22 sampling stations at South Lake can be divided into 5 characteristic areas, which are inlet, outlet, northern, southern and eastern parts.The propriety of the division is verified with the liniar discriminant analysis, in which the discriminant function are determined with water quality data, and the probabilitiys of correct discrimination at each area is calculated. The results show that the discriminat functions may express quantitatively the characteristics of each area on water quality changes, and that 65% of data at each area can be estimated to belong to their own area.
