Time series analysis techniques applied to transportation noise

Acoustic noise assessment is a crucial problem in areas in which transportation means, such as motorway, railway, airport, etc., are present. Dwelled areas, in fact, represent a sensible point, that is affected by several externalities, among which, acoustic noise is very important. In this paper, the techniques known as Time Series Analysis (TSA), are used to analyze datasets of noise level produced by transport systems. This approach is based on the analysis of trend and seasonality of the series, and on the implementation of a function of the time that can provide predictions for future time periods. According to the choice and to the input of each model, the forecast horizon can vary from few days further to any time period in the future. Two techniques will be presented: one is based on a Deterministic Decomposition (DD-TSA), able to predict at any future time period; the second is based on a stochastic approach, and adopt the so called SARIMA (Seasonal AutoRegressive Integrated Moving Average) models, to provide prediction on a short time range. Both techniques will be applied to a road traffic noise dataset and to an airport noise levels time series. Results will show that the typology of transportation system does not affect the prediction performances of both the DD-TSA and the SARIMA techniques, even though the time basis of the data is different, being daily for traffic noise and hourly for airport.