Prediction of daily ozone concentration maxima in the urban atmosphere

A model for the prediction of daily O3 concentration maxima several hours in advance is reported. The model is developed for an urban area of warm oceanic climate and can be easily integrated with automatic pollution and meteorological stations. It provides a basis for issuing advance warnings to the public before O3 peaks are reached. It also affords straightforward interpretation of the processes leading to O3 level changes as the predicted O3 concentrations are directly related to variables with physical meaning. The correlations between O3 concentrations and other pollutants, e.g. CO, NO, NO2, SO2, suspended particles as well as meteorological variables, e.g. wind speed, temperature, relative humidity, rainfall, solar radiation, cloud cover and light intensity, were evaluated prior to the selection of variables for the model. Multiple linear regression with forward stepwise method has defined seasonal, photochemical and persistence components that account for the seasonal influence of O3 production, the relations between O3 and NOx, and previous day O3 levels, respectively. The model was calibrated using data collected over a period of two years and its performance for prediction was evaluated by calculation of the daily O3 concentration maxima over the subsequent two years and comparing the predictions to measured values.