Acoustical noise is one of the most relevant problem to be faced in urban areas. Since a large network of monitoring stations would be very expensive, the adoption of predictive model is a very common practice in environmental impact assessment. Usually, the standard models are based on the fit of field measurements in certain areas, in order to evaluate the model parameters and be able to give predictions in any other condition. A new approach, based on Time Series analysis, is presented in this paper. This approach considers that a time series can be composed by a trend (long term behaviour), a seasonality periodicity) and an irregular term (random non deterministic variation). The evaluation of these components is made in the calibration phase and a validation can be performed evaluating the difference between observed and forecasted values. This scheme is applied to a case study time series, that is a daily noise levels dataset detected in the city of Messina, Italy. The building model procedure and results are discussed and presented in details for different calibration and validation subsets, in order to highlight the variation in the model predictive capabilities and to optimize the forecast, by means of minimization of the difference between actual data and forecasts. The easiness in the implementation and the good predictive performances will be the strengths of the model.

Development and Application of a Time Series Predictive Model to Acoustical Noise Levels

GUARNACCIA, CLAUDIO;QUARTIERI, Joseph;TEPEDINO, CARMINE
2014

Abstract

Acoustical noise is one of the most relevant problem to be faced in urban areas. Since a large network of monitoring stations would be very expensive, the adoption of predictive model is a very common practice in environmental impact assessment. Usually, the standard models are based on the fit of field measurements in certain areas, in order to evaluate the model parameters and be able to give predictions in any other condition. A new approach, based on Time Series analysis, is presented in this paper. This approach considers that a time series can be composed by a trend (long term behaviour), a seasonality periodicity) and an irregular term (random non deterministic variation). The evaluation of these components is made in the calibration phase and a validation can be performed evaluating the difference between observed and forecasted values. This scheme is applied to a case study time series, that is a daily noise levels dataset detected in the city of Messina, Italy. The building model procedure and results are discussed and presented in details for different calibration and validation subsets, in order to highlight the variation in the model predictive capabilities and to optimize the forecast, by means of minimization of the difference between actual data and forecasts. The easiness in the implementation and the good predictive performances will be the strengths of the model.
File in questo prodotto:
Non ci sono file associati a questo prodotto.

I documenti in IRIS sono protetti da copyright e tutti i diritti sono riservati, salvo diversa indicazione.

Utilizza questo identificativo per citare o creare un link a questo documento: http://hdl.handle.net/11386/4516262
 Attenzione

Attenzione! I dati visualizzati non sono stati sottoposti a validazione da parte dell'ateneo

Citazioni
  • ???jsp.display-item.citation.pmc??? ND
  • Scopus 26
  • ???jsp.display-item.citation.isi??? ND
social impact