A hybrid model for predicting the distribution of pollutants dispersed from line sources

This paper applies a hybrid modeling methodology to the problem of the dispersion of pollutants from line sources. The model combines a deterministic component, the GM model (Chock, 1978), with a statistical component, the two-parameter Weibull distribution, to produce estimates of the entire distribution of pollutant concentration. The approach is demonstrated using hourly average carbon monoxide data recorded in Melbourne, Australia. Using a second data set a model validation exercise is performed. The hybrid model has worked well, producing estimates of pollutant concentration with an accuracy better than a factor of two over all percentiles of the distribution of pollutant concentration.