How Green Is Green? Modeling Urban Greenness Exposure in Environmental Health Research

摘要

Exposure to greenness has several health benefits, yet analyses are limited by uncertainty about accuracy of greenness metrics, often derived from different remotely-sensed data sources and using different spatial methods. This research (1) assesses the strengths and weaknesses of multiple greenness data sources and metrics for application in environmental health research, and (2) develops and tests an alternative greenness exposure metric that can be applied worldwide. Methods: We analyzed 5 data sources: Landsat time series, Landsat 8, Sentinel-2, RapidEye, and the Green View Index (GVI), derived from Google Street View. These data sets span various time series, resolutions and costs, and represent aerial and perspective views. We compared the Normalized Difference Vegetation Index (NDVI) using different imagery types and the GVI for various buffer distances around postal codes and examined sensitivity to spatial metrics and data sources. Based on these analyses, we constructed a spatially-weighted greenness metric combining data from Sentinel-2 and the GVI that incorporates neighbourhood street-level and at-home greenness. Results: NDVI showed correlations of between 0.65 and 0.85 among satellite types and demonstrated significant inter-variability. GVI showed low correlations with all other data types (0.25-0.40), suggesting an important new source of greenness data. Metrics were spatially sensitive, particularly at small distances and high resolutions, but lacked temporal sensitivity. Initial analyses suggest that the proposed metric is superior to traditional measures that overestimate neighbourhood greenness exposure and underestimate at-home exposure. Conclusions: This research is the first comparison of multiple remote sensing data sources and metrics, including both aerial and novel perspective views. It presents an alternative greenness exposure metric based on freely accessible data sources that may be applied in public health research internationally.