Tree abundance, species richness, or species mix? Exploring the relationship between features of urban street trees and pedestrian volume in Jinan, China

Abstract

Accumulating evidence has confirmed that urban greenery, especially street trees, is beneficial to walking behaviors. Existing street greenery exposure-walking behavior studies have focused on the quantity of greenery, which is often measured by normalized difference vegetation index (NDVI) or green view index (GVI). However, some important qualities of street trees, the dominant component of urban greenery, were often overlooked, due to the labor-intensive and expensive nature of on-site surveys. To address this issue, we proposed a cutting-edge deep learning technique to identify street tree species at the individual tree level. We established a citywide dataset of all street trees with species information via 185,831 Baidu Street View images (BSV) in Jinan, China. Population-level walking intensity, measured by pedestrian volume, was retrieved from BSV images using Baidu AI. We further adopted spatial regression models to investigate the association between pedestrian volume and street tree characteristics, including street tree abundance (number of street trees), species richness (number of unique tree species) and species mix (the degree of diversity of tree species). The built environment and urban greenery covariates were adjusted in the models. The results indicate that the street tree abundance and species mix are positively associated with pedestrian volume. Species richness is not associated with it. Besides, spatial mismatch is identified between abundance and species mix of street trees in the study area. Hence, to facilitate walking behavior and deliver related health benefits, it is necessary to develop fine-grained measures of street greenery features.