Author(s)

U. Schlink, O. Herbarth & M. Richter

Abstract

A method is proposed for short-term ambient ozone prediction. The basic assumption is that ozone concentration profiles can be implied from preceeding parameters, the predictors. For this purpose, certain properties have to be identified during the time period prior to the forecasting point in order to calculate the probability that the threshold of 0.18 mg/nf ozone will be exceeded. Data analyses indicate that ambient ozone time series show periodicities and autocorrelation that can be used for predictive purposes. On the one hand, the wanted parameters are suitable features of the measurement recordings as the Os and the NOz values at the forecasting point. These are the direct predictors. On the other hand, it is possi

Keywords