Development and evaluation of statistical prediction model for photochemical smog

Abstract

In order to predict the occurrence of photochemical smog, the statistical model for oxidant concentration was developed. According to the result of principal component analysis for 37 stations in Tokyo, it was found that total area in Tokyo can be covered by only one model. Two types of prediction models were developed and compared. One is based on the discriminant analysis and another is based on the multiple regression analysis. These models were constructed from the data of 1982 and validated by the data of 1981. Though result by multiple regression model is better as a whole. discriminant model is better for prediction of high concentration day.