1985 Volume 20 Issue 5 Pages 384-393
The purpose of this study is to examine a method for establishing the representativeness of an air monitoring station by applying a method of estimating the regional distribution of air pollutant concentration based on a statistical model.
It is assumed that the air pollutant concentration at a station, which has been proposed for elimination under a reduction or rearrangement plan, can be estimated from the data obtained at other stations. A method for determining the most important representative stations from a set of existing stations is tested through its application to the 35 stations of the air monitoring network of the Tokyo Metropolitan Government.
The results are as follows.
(1) Five of the 35 stations are selected as candidates of reduction under the criterion of multiple correlation coefficients of 0.9 by applying the method of step-wise multiple regression analysis.
(2) When hierarchical structural analysis of the stations is done by applying the method of Interpretive Structural Modeling (ISM) based on the ad jacency matrix among the stations, the 35 stations are divided into three levels: four stations of the candidates were in the lowest level, of which stations could be explained by the data of air pollutant concentration of other 31 stations including one of the candidates; the remain of the candidates is in second level.
(3) In consequence, when three stations: two stations of the lowest level and the station of the second level are eliminated from the network and air pollutant concentrations at the station are estimated by the data obtained at other stations, the correlation coefficients between calculation and observation are 0.74, 0.89 respectively.
