Develop a land use regression model to predict noise levels in the urban residential neighbourhood of Bhopal, India
摘要
Environmental noise pollution has a detrimental impact on people’s quality of life, particularly in rapidly developing countries like India. The study aims to measure noise levels, understand their association with the surrounding built environment using Pearson correlation, and develop a Land Use Regression (LUR) model to predict noise in the Chunabhatti neighbourhood, Bhopal, India. Class I sound level meter at 255 sites was used to obtain noise levels (Leq 5 min) during peak traffic from 4:30 p.m. to 9 p.m. in the month of May. The LUR model was used to examine the relationship between noise levels and various predictor variables of the built environment, including proximity to arterial roads, land cover, land use, settlement typology, and road hierarchy. Then, ten-fold cross-validation was conducted to evaluate the model’s performance. The mean noise level in the neighbourhood was 58.76 dB(A), often surpassing the permissible noise limit in most areas. The supervised forward stepwise regression method was applied to develop a LUR model that fits well (R2 = 0.723), effectively addressing small-scale noise fluctuations. The R2 from the ten-fold cross-validation was 0.83, with an RMSE of 1.30 dB(A), indicating robust and accurate noise level prediction. The results demonstrated that noise level varied from 47.60 dB(A) to 78.60 dB(A), with considerable fluctuations depending on proximity to arterial roads, vegetation, and settlement typology. Gated societies had lower noise levels, with a mean value of 56.36 dB(A), whereas informal settlements recorded the highest, averaging 60.17 dB(A). Additionally, non-residential land use, collector roads, and areas with greater land use diversity (as indicated by SHDI) significantly increased noise levels. This study emphasizes the significance of localized noise assessments in guiding public health and urban policy in developing countries.