@article{oai:soar-ir.repo.nii.ac.jp:00010749,
 author = {中路,  勉 and 港,  和行 and 高倉,  芳成 and 藤原,  みや子 and 入江,  鐐三},
 issue = {2},
 journal = {信州大学農学部紀要},
 month = {Dec},
 note = {We tried to explain the characteritics of the water quality and its variation in Shinano River by the use of time series analysis, correlation analysis, principal component analysis and factor analysis. The values obtained from the survey of the water quality of the public water area in Nagano and Niigata prefecture from 1973 to 1979 were used. As the variables, water temperature, pH, DO, BOD and SS were employed. The results were summarized as follows: 1. To remove the seasonal variation in water temperature and DO, which showed a twelve month-cycle in time series analysis, the revised values (revised series) were calculated, by dividing the original values from the survey (original series) by the seasonal index, which was obtained by transfer mean method. It was recognized that the seasonal variation was removed at all measuring places on the run test of the revised series. 2. By the use of the correlation analysis, effects of the variables on the other variables was reduced by removing the seasonal variation. In the revised series, correlation cofficients of individual variables were close to partial correlation cofficients of these variables. 3. By the use of the principal component analysis, both the eigen value and the ratio of contribution of the first principal component were reduced by removing the seasonal variation. The revised series required the fourth principal component to obtain the cumulative contribution above 80%. On the eigen vector in the original series, water temperature and DO showed the high value in the first principal component, pH and SS in the second principal component and BOD in the third principal component. Since, in the revised series, water temperature was under the second principal component, and BOD and SS showed high value in the first principal component, the effect of the removal of the seasonal variation was clear. 4. On the revised series, the factor loading calculated was projected along the axis of the principal components. These projections showed that BOD and SS were in a group at every four places of the midstream. By the rotation by the standard varimax method, the character of every factor became clearer. It became clear that the first factor was the index of DO, the second was of water temperature, the third was of the organic corruption and the fourth was of SS. 5. From the sqare pattarn with the factor examination marks, places of measurment of the main stream of Shinano river could be classified into three groups: the upper and the lower stream where the organic corruption increased, the midstream where the water quality was good and clear, and the middle to lower stream where SS was main factor., Article, 信州大学農学部紀要 19(2): 169-181(1982)},
 pages = {169--181},
 title = {河川水質の統計学的研究(IV) 信濃川水質への多変量解析の適用},
 volume = {19},
 year = {1982}
}