Development of a model for forecasting of PM10 concentrations in Salamanca, Mexico

Salamanca has been ranked as one of the most polluted cities in Mexico. The industry in the area led to a major economic development and rapid population growth in the second half of the twentieth century. The concerning registered pollutants are sulfur dioxide (SO2) and particles in the order of 10 micrometers or less (PM10). The prediction of concentrations of those pollutants can be a powerful tool in order to take preventive measures such as the reduction of emissions and alerting the affected population. This work presents a forecasting model to predict average concentration of PM10 for the next 24 hours. The proposed model uses a combination of Multilayer Perceptron Neural Network and clustering algorithm. The source database used contains historical time series of meteorological variables and concentrations of PM10 collected in three different stations in Salamanca. The clustering algorithms have been implemented in order to find relationships between PM10 and meteorological variables. These relationships will help us to get additional information that will be used in the prediction model. The proposed model was compared, for accuracy and validation purposes, with a simple Multilayer Perceptron and a multiple Linear Regression. The performance estimation is determined using the Root Mean Square Error (RMSE) and the Mean Absolute Error (MAE). The obtained results show the importance of this set of meteorological variables in the prediction of pollutant concentrations.