Combined model for PM10 forecasting in a large city

► The performance of statistical PM10 forecasting models is shown. ► Reasonable accuracy is achieved with a neural network model (ANN). ► Improvement in accuracy of high concentrations forecasting is obtained with a nearest neighbor model (NNM). ► Increase of false alarms episodes is observed with NNM. ► Combination of the ANN and NNM provides a near optimal PM10 operational forecasting system.