A NEURAL-NETWORK-BASED METHOD FOR SHORT-TERM PREDICTIONS OF AMBIENT SO2 CONCENTRATIONS IN HIGHLY POLLUTED INDUSTRIAL-AREAS OF COMPLEX TERRAIN

A new method for short-term air pollution prediction is described, bas ed on the neural network. It was developed for prediction for SO2 poll ution around the biggest Slovenian thermal power plant at Sostanj. Bec ause of the high SO2 emissions, there is a need for a reliable air pol lution prediction method that would enable lowering the peaks of pollu tant concentrations in critical meteorological situations. In complex topography, classical methods for air pollution modelling are not reli able enough. The results obtained by this new method are very promisin g. The method can also be used, with slight modifications, for other i mportant air pollutants, the concentrations of which can be measured c ontinuously.