Statistical Road Classification Applied to Stratified Spatial Sampling of Road Traffic Noise in Urban Areas

Abstract

Monitoring of road traffic noise is becoming an important issue in modern cities due to the spreading of noise pollution and the extension of monitored areas. Thus, the stratified spatial sampling is frequently applied to reduce the costs and provide adequate accuracy in order to obtain reliable noise maps. The definition of the strata in the sampling may refer to the legislative classification of roads: in Italy 8 classes of roads are defined. Generally, this classification often does not reflect the actual use of roads in the mobility network, as it is mainly based on the their geometrical characteristics. In order to improve the efficiency of stratification, an alternative criterion is proposed, based on clustering of 24 h patterns of road traffic noise. To explain this criterion, a preliminary analysis of 74 patterns of 24 h continuous monitoring of the hourly equivalent levels LAeqh taken in the city of Milan, Italy, in 35 different sites has been performed. The applied agglomerative algorithms provided two groups and the mean profile of each cluster was associated with the available traffic flow data, namely the rate at morning rush hour. By means of ROC curve, the first cluster was associated with traffic flow greater than 1500 vehicles/hour and the second with less than 1500 vehicles/hour. The proposed criterion of road stratification performed better than the one based on the legislative classification of roads as, for a given accuracy, it needs a lower number of sites to estimate the noise indicators.