Correlation of Psychoacoustic Indices with Traffic Conditions Across Cluster-Classified Intersections in a Developing Tier-2 Indian City

This study investigates the relationship between traffic conditions and psychoacoustic indices in a developing city, using half-hourly data collected in the morning and evening at different intersections. The psychoacoustic indices were computed using standardized methodologies, with loudness calculated through Zwicker's model based on sound pressure levels, while sharpness was derived from frequency-weighted contributions above 2 kHz. Roughness and fluctuation strength were extracted using signal processing techniques, applying band-pass and low-pass filtering to isolate rapid and slow sound fluctuations. Spearman's correlation analysis revealed significant associations, particularly highlighting strong relationships between queue length and psychoacoustic discomfort indicators such as loudness (r = 0.78) and roughness (r = 0.81). To assess multicollinearity effects, a Variance Inflation Factor (VIF) analysis was conducted, revealing a notable interaction between roughness and queue length, suggesting overlapping influences in congested environments. Cluster classification using K-Means Clustering identified three distinct noise exposure zones based on psychoacoustic profiles and traffic characteristics, validated by a Silhouette Score of 0.62. The analysis revealed that residential commuting clusters exhibited the highest psychoacoustic discomfort, while educational commuting clusters demonstrated higher sharpness due to frequent honking, emphasizing the significant influence of commuting patterns on urban noise exposure.