Can’t see the wood for the trees? An assessment of street view- and satellite-derived greenness measures in relation to mental health

Abstract

Greenness in the urban living environment is inconsistently associated with mental health. Satellite-derived measures of greenness may inadequately characterize how people encounter greenness visually on site, but systematic comparisons are lacking. We aimed 1) to compare associations between remotely sensed and street view (SV) greenness, and 2) to examine whether these greenness metrics are differently associated with mental health outcomes. We used cross-sectional depressive and anxiety symptoms data on adults in Amsterdam, the Netherlands. We employed a convolutional neural network to segment greenness in SV panoramas. Greenness was measured top-down by normalized difference vegetation indices (NDVI) from 1 m resolution orthophotos (OP) and 30 m resolution Landsat-8 (LS) imagery per postal code, and 100 and 300 m concentric and street-network buffers at the home address. Correlation analyses assessed associations across greenness measures. Covariate-adjusted regressions (e.g., noise, air pollution, deprivation) were conducted to assess associations between each greenness metric and mental health outcomes. Correlations between greenness metrics were significantly positive and moderately high. SV greenness was less sensitive across scales and residential contexts than OP and LS greenness. There was no statistically significant evidence that people with less urban residential greenness had higher depression or anxiety scores than those exposed to higher levels. Nor did different greenness measures, scales, or residential context definitions alter our null associations. This suggests that even though SV and remotely sensed measures capture different aspects of greenness, these differences across exposure metrics did not translate into an association with mental health outcomes.