Data Insights for Sustainable Cities: Associations between Google Street View-Derived Urban Greenspace and Google Air View-Derived Pollution Levels

Unprecedented levels of urbanization have escalated urban environmental health issues, including increased air pollution in cities globally. Strategies for mitigating air pollution, including green urban planning, are essential for sustainable and healthy cities. State-of-the-art research investigating urban greenspace and pollution metrics has accelerated through the use of vast digital data sets and new analytical tools. In this study, we examined associations between Google Street View-derived urban greenspace levels and Google Air View-derived air quality, where both have been resolved in extremely high resolution, accuracy, and scale along the entire road network of Dublin City. Particulate matter of size fraction less than 2.5 μm (PM2.5), nitrogen dioxide, nitric oxide, carbon monoxide, and carbon dioxide were quantified using 5,030,143 Google Air View measurements, and greenspace was quantified using 403,409 Google Street View images. Significant (p < 0.001) negative associations between urban greenspace and pollution were observed. For example, an interquartile range increase in the Green View Index was associated with a 7.4% [95% confidence interval: −13.1%, −1.3%] decrease in NO2 at the point location spatial resolution. We provide insights into how large-scale digital data can be harnessed to elucidate urban environmental interactions that will have important planning and policy implications for sustainable future cities.

Table S2.Lagrange Multiplier and Robust LM tests for all models analysing associations between urban greenspace (GVI, NDVI, GVI:NDVI) and PM 2.5 , for all spatial scenarios studied (all point locations and for buffer zones of radius 100m, 300m, 500m, 1000m, and 2000m) (n=24,694).

GVI LM lag
Table S3.Lagrange Multiplier and Robust LM tests for all models analysing associations between urban greenspace (GVI, NDVI, GVI:NDVI) and NO 2 , for all spatial scenarios studied (all point locations and for buffer zones of radius 100m, 300m, 500m, 1000m, and 2000m) (n=24,694).S7.Moran's I test for the residuals of the Spatial Lag regression analysis for all models analysing associations between urban greenspace (GVI, NDVI, GVI:NDVI) and air pollution (PM 2.5 , NO 2 , NO, CO, and CO 2 ) levels.Models analysed these associations for all for all spatial scenarios studied (all point locations and for buffer zones of radius 100m, 300m, 500m, 1000m, and 2000m) (n=24,694).All models were adjusted for traffic, precipitation, temperature and population density.Table S8.Spatial lag Regression coefficients for all models analysing associations between urban greenspace (GVI, NDVI, GVI:NDVI) and air pollution (PM 2.5 , NO 2 , NO, CO and CO 2 ) levels.Models analysed these associations for point locations and all buffer zones of radius 100m, 300m, 500m, 1000m, and 2000m (n=24,694).All models were adjusted for traffic, precipitation, temperature and population density.Note that these results consider the natural log of all pollution dependent variables.

Figure S1 .
Figure S1.High-resolution version of figure 1a in the main manuscript: Map of urban greenspace Green View Index (GVI) computed in high spatial resolution (67,265 point locations) on the entire road network of Dublin City.

Figure S2 .
Figure S2.High-resolution version of figure 1b in the main manuscript.Map of urban greenspace Normalised Difference Vegetation Index (NDVI) computed in high spatial resolution (67,265 point locations) on the entire road network of Dublin City.

Figure S3 .
Figure S3.High-resolution version of figure 1c in the main manuscript.Map of urban greenspace metric (GVI:NDVI) computed in high spatial resolution (67,265 point locations) on the entire road network of Dublin City.

Figure S4 .
Figure S4.High-resolution version of figure 1d in the main manuscript.The urban greenspace Green View Index (GVI) (map D) highlights specific sites representing a dense urban City Centre area and a major urban park (Phoenix Park) which assist in the visual interpretation of the urban greenspace metrics computed across the city.

Figure S5 .
Figure S5.High-resolution version of figure 2a in the main manuscript.Map of Google Air View-derived PM 2.5 air pollution quantified in high spatial resolution using 5,030,143 point measurements on 24,694 road links (50m road segments) across the entire road network of Dublin City.

Figure S6 .
Figure S6.High-resolution version of Figure 2b in the main manuscript.Map of Google Air View-derived NO air pollution quantified in high spatial resolution using 5,030,143 point measurements on 24,694 road links (50m road segments) across the entire road network of Dublin City.

Figure S7 .
Figure S7.High-resolution version of figure 2c in the main manuscript.Map of Google Air View CO2 quantified in high spatial resolution using 5,030,143 point measurements on 24,694 road links (50m road segments) across the entire road network of Dublin City.

Figure S8 .
Figure S8.High-resolution version of Figure 2d in the main manuscript.Map of Google Air View-derived NO 2 air pollution quantified in high spatial resolution using 5,030,143 point measurements on 24,694 road links (50m road segments) across the entire road network of Dublin City.

Figure S9 .
Figure S9.High-resolution version of Figure 2e in the main manuscript.Map of Google Air View-derived CO air pollution quantified in high spatial resolution using 5,030,143 point measurements on 24,694 road links (50m road segments) across the entire road network of Dublin City.

Figure S10 .
Figure S10.High-resolution version of Figure 2f in the main manuscript.Graphs show comparisons of citywide PM 2.5 , NO 2 and CO pollution distributions with national Irish and international World Health Organization limits.

Figure S16 .
Figure S16.NO 2 measurements for the EPA Station Dublin 2 and Google Air View measurements at the same location for September 30, 2021.EPA measurements are hourly-averaged and available for the whole day, while Air View measurements were only taken between 14:00h and 15:00h.The graph allows the visual inspection of the spatial variation of NO 2 since different concentrations were recorded for each hour.

Figure S17 .
Figure S17.Distribution of NO 2 Google Air View measurements vs. hour-average concentrations recorded by the EPA Station Dublin 2. The boxplot demonstrates the distribution of NO 2 values recorded by Google Air View between 14:00h and 15:00h for September 30, 2021.The dotted blue line represents the NO 2 concentrations recorded by EPA Station Dublin 2 during the same time period.

Figure S18 .
Figure S18.PM 2.5 measurements for the EPA Station Dublin 6 and Google Air View measurements at the same location for May 06, 2022.EPA measurements are hourly-averaged and available for the whole day, while Google Air View measurements were only taken between 12:00h and 13:00h.The graph allows the visual inspection of the spatial variation of PM 2.5 since different concentrations were recorded for each hour.

Figure S19 .
Figure S19.Distribution of PM 2.5 Google Air View measurements vs. hour-average concentrations recorded by the EPA Station Dublin 6.The boxplot demonstrates the distribution of PM 2.5 values recorded by Google Air View between 12:00h and 13:00h for May 06, 2022.The dotted blue line represents the PM 2.5 concentrations recorded by EPA Station Dublin 6 during the same time period.

Figure S20 .
Figure S20.CO measurements for the EPA Station Dublin 8 and Google Air View measurements at the same location for March 16, 2022.EPA measurements are hourly-averaged and available for the whole day, while Google Air View measurements were only taken between 09:00h and 10:00h.The graph allows the visual inspection of the spatial variation of CO since different concentrations were recorded during the day.

Figure S21 .
Figure S21.Distribution of CO Google Air View measurements vs. hour-average concentrations recorded by the EPA Station Dublin 8.The boxplot demonstrates the distribution of CO values recorded by Google Air View between 09:00h and 10:00h for March 16, 2022.The dotted blue line represents the CO concentrations recorded by EPA Station Dublin 8 during the same time period.

Figure S22 .
Figure S22.PM 2.5 measurements for the EPA Station Dublin 12 and Google Air View measurements at the same location for July 13, 2021.EPA measurements are hourly-averaged and available for the whole day, while Google Air View measurements were only taken between 12:00h and 13:00h.The graph allows the visual inspection of the spatial variation of PM 2.5 since different concentrations were recorded for each hour.

Figure S23 .
Figure S23.Distribution of PM 2.5 Google Air View measurements vs. hour-average concentrations recorded by the EPA Station Dublin 12.The boxplot demonstrates the distribution of PM 2.5 values recorded by Google Air View between 12:00h and 13:00h for July 13, 2021.The dotted blue line represents the PM 2.5 concentration recorded by EPA Station Dublin 6 during the same time period.

Figure S24 .Figure S25 .Figure S26 .Figure S27 .Figure S28 .Figure S29 .
Figure S24.Scatterplots of actual values vs. predicted values for point locations (n=24,694).Spatial lag regression models were used to analyse associations between urban greenspace (GVI, NDVI, GVI:NDVI) and air pollution (PM 2.5 , NO 2 , NO, CO and CO 2 ) levels.All models were adjusted for traffic, precipitation, temperature, and population density.Graphs were plotted with a random sample of 150 points.

Table S4 .
Lagrange Multiplier and Robust LM tests for all models analysing associations between urban greenspace (GVI, NDVI, GVI:NDVI) and NO, for all spatial scenarios studied (all point locations and for buffer zones of radius 100m, 300m, 500m, 1000m, and 2000m) (n=24,694).