TRAFFIC NOISE MODELLING AT URBAN INTERSECTIONS IN INDIA

Abstract

Traffic noise has emerged as a major contributor to noise pollution in urban areas. The problem of traffic noise is particularly pronounced in mid-sized cities in developing countries like India, where poor infrastructure, vehicle maintenance, policy implementation, complex traffic flow, and road geometry exacerbate the issue, especially at intersections. Traffic noise modelling is the initial step in the direction of applying measures to control traffic noise and its impact on the health of urban dwellers. While efforts have been made to develop traffic noise models, limited studies have been conducted for modeling intersection-specific models, particularly for mid-sized cities like those in India. Therefore, the primary aim of this study is to develop traffic noise models for intersections in mid-sized Indian cities. Additionally, a model is developed to analyse traffic noise-induced annoyance and its impact on the psychological health of dwellers at intersections. The research aids in predicting traffic noise levels, analysing the impact of variables on traffic noise levels, and assessing the impact of traffic noise on annoyance and psychological health conditions. The study is based on data collected from different intersections in Kanpur, Uttar Pradesh, India. Data collection occurs in two stages: initially, 342 hours of data are collected from 19 intersections with varying land uses, followed by interviews with 487 dwellers at 15 of the selected intersections. The present research is structured into four main sections. The first section focuses on developing traffic noise models using multiple linear regression (MLR) and artificial neural network (ANN) techniques. The MLR model establishes linear relationships and is designed for wider applicability due to its simple methodological approach. In contrast, the ANN model aims for high prediction accuracy by capturing the nonlinear relationship between traffic noise levels and influencing variables. To account for variations in traffic flow characteristics and road geometry at the entrance and exit arms of intersections, separate arm analysis is conducted to develop the traffic noise model using the MLR approach. The significant variables identified in the MLR model include entrance arm traffic volume, exit arm traffic volume, exit arm speed, percentage of heavy vehicle at entrance arm and exit arm, percentage of Vikram-rikshaw at entrance, and geometric mean distance of sound level meter from the carriageway. Although honking is found to be significant at a 95% confidence interval, it is ignored in the final MLR model due to collinearity issues with traffic volume. The model illustrates that all variables positively impact traffic noise level, while exit arm speed and geometric mean distance of sound