SHORT-TERM PREDICTION OF SO2 CONCENTRATION IN MACAO WITH ARTIFICIAL NEURAL NETWORKS

The air quality of Macau is deteriorating in the recent years due to t he rapid economic and population growth of itself and its surrounding areas. One of the main air pollutant which is of great concern to Maca u, as well as to many urban cities in the world, is sulfur dioxide (SO 2). In view of the yearly averaged SO2 concentration, it is found that Macau may be classified as an uncontaminated area. Nevertheless, the daily averaged SO2 concentrations of the year 1995 show that 39% of th e last 3-month values recorded at A. Preta exceeded the Chinese primar y standard, and it is selected as the representing period for the inve stigation of SO2 pollution in Macau. Using just the records of the pas t month, a three-layered feed-forward artificial neural networks is de veloped to predict the daily SO2 concentration 5 days in advance. The results show that the accuracy of the ANN model is within 14.45% and 1 3.71% for two testing periods, respectively. The promising results ind icate that the ANN could be used to develop efficient air-quality anal ysis and prediction models in the future. (C) 1998 Published by Elsevi er Science S.A. All rights reserved.