Development of a Quantitative Methodology to Assess the Impacts of Urban Transport Interventions and Related Noise on Well-Being

Well-being impact assessments of urban interventions are a difficult challenge, as there is no agreed methodology and scarce evidence on the relationship between environmental conditions and well-being. The European Union (EU) project “Urban Reduction of Greenhouse Gas Emissions in China and Europe” (URGENCHE) explored a methodological approach to assess traffic noise-related well-being impacts of transport interventions in three European cities (Basel, Rotterdam and Thessaloniki) linking modeled traffic noise reduction effects with survey data indicating noise-well-being associations. Local noise models showed a reduction of high traffic noise levels in all cities as a result of different urban interventions. Survey data indicated that perception of high noise levels was associated with lower probability of well-being. Connecting the local noise exposure profiles with the noise-well-being associations suggests that the urban transport interventions may have a marginal but positive effect on population well-being. This paper also provides insight into the methodological challenges of well-being assessments and highlights the range of limitations arising from the current lack of reliable evidence on environmental conditions and well-being. Due to these limitations, the results should be interpreted with caution.


Supplementary File 3: Coverage of Covariates Potentially Associated With Wellbeing
Several covariates expected to have an effect on wellbeing were included in the EQLS2012 datasets, such as gender, age (5 categories), income (quartiles), education (3 categories), employment (7 categories), making ends meet financially (6 categories) and household structure (5 categories). These covariates were used in their original format, except when the variable values did not provide a clear direction. This was the case for employment (recoded into 3 categories) and household structure (recoded into 4 categories). All other values of the covariate variables were coded "system missing". Within each dataset, the selected covariates were screened for significant bivariate associations with subjective wellbeing using chi-square tests. Significant variables (p < 0.05) were then applied as covariates for the binary logistic regression with "good wellbeing" as outcome (see Table below).
For the SHP2012 dataset, similar covariates were available but for household structure and making ends meet financially, some adjustments were necessary: "Household structure" was replaced by "Single household (yes/no)" and "Making ends meet financially" was replaced by "Satisfaction with financial situation (0 (not at all) to 10 (completely satisfied))".
 not significant not significant Making ends meet financially *    * For Switzerland, household structure was replaced by "Single household (yes/no)" and making ends meet financially was replaced by "Satisfaction with financial situation (0 (not at all) to 10 (completely satisfied))".

Supplementary File 5: Establishing Noise Cut-Offs on City Level on the Basis of National Datasets
The matching of the local noise model data and the noise perception categories derived from EQLS2012 and SHP2012 data is explained in the Table below using the example of Thessaloniki.
The EQLS2012 indicated that in Greece, about 14.4% of the urban population report major problems with noise. Transferring these 14.4% to the modeled noise exposure data from Thessaloniki would suggest 65 dB Lden as the most suitable noise cut-off of, as 15.2% of Thessaloniki's population are exposed to 65 dB Lden and beyond. This is the noise level affecting the population percentage closest to 14.4% (cut-off levels at 64 and 66 dB would be less close to the 14.4%). The same approach was applied for moderate and no problems with noise. The main limitation associated with this approach is that the noise perception data taken from the national surveys results from all noise sources while the noise models provided by the cities reflect only traffic-related noise. Although traffic noise accounts for the largest share of overall urban noise exposure, this approach is therefore problematic and indicates one of the many methodological problems arising for a noise-related wellbeing assessment of urban interventions. Especially unclear is the contribution of neighbourhood noise on the overall perception of noise problems as not many studies provide insight into the relative influence of neighbourhood noise and traffic noise on overall noise perception. This limitation is addressed in more detail in the discussion section of the main paper.  Figure S1. Comparison of traffic noise exposure distribution in the population of Basel at Baseline2010 and under BAU2020 and Intervention2020.