Urban Greenness Extracted from Pedestrian Video and Its Relationship with Surrounding Air Temperatures

Abstract Urban greenness has been associated with a wide range of health benefits, partially due to local cooling. Several studies on these health benefits have assessed individual and population exposure to urban greenness using the Normalized Difference Vegetation Index (NDVI) from different satellite platforms. Recent comparisons between birds-eye NDVI and street-level measurements suggest that NDVI can severely misclassify individual exposure which, in turn, can bias epidemiologic effect estimates. Pedestrian video data may provide a novel source of individual, eye-level information on both indoor and outdoor exposure to vegetation. The objective of this pilot study was to examine the potential of pedestrian video data for assessing exposure to urban greenness using secondary data collected for a different study on microscale urban air temperatures. Image processing was used to extract green, yellow, and shaded pixels from ˜10 million frames of video footage collected during 40 sampling runs of 20 urban routes measuring 8–10 km each. Resulting greenness values (combined total of green, yellow, and shaded pixels) were compared with concurrent air temperatures using correlations, time series plots, and maps. Correlations ranged from -0.61 to 0.34 and were in the expected direction for 31 of 40 runs. Time series plots and overlay maps showed clear inverse relationships in many cases. Most routes originally chosen to characterize higher temperature areas had negative correlations. Flat and weakly positive relationships tended to occur when conditions were overcast or routes were closer to large waterways. Secondary data are limited for such assessments, but the methods described here are promising for future exposure assessment at the individual level and evaluation of methods applied at the population level.