Characterizing street trees in three metropolises of central China by using Street View data: From individual trees to landscape mapping
Liang, Chentao; Jiang, Huan; Yang, Sijia; Tian, Panli; Ma, Xiang; Tang, Zhonghua; Wang, Huimei; Wang, Wenjie
Peer reviewed, Journal article
Published version
View/ Open
Date
2024Metadata
Show full item recordCollections
- Publikasjoner fra CRIStin - SINTEF AS [5911]
- SINTEF Industri [1593]
Abstract
Street trees can benefit urban spaces, and the characterization of them is a basis for evaluating their functions and proper management. In this paper, we speculate that Street View imagery-based measurement favors characterizing street trees from individual tree sizes and growth status to spatial mapping and risk region identification at multiple city levels. Using Wuhan, Changsha, and Nanchang as examples, nearly 6000 sites and over 111,000 Baidu Street View (BSV) images were used to measure tree sizes, growth status, and vertical structure. Together with BSV-derived tree data, we also calculated the street-view greenness index (GVI) and the Normalized Difference Vegetation Index (NDVI) from remote sensing data, and then city-level spatial mapping was produced to identify risk regions for urban afforestation. The results showed that BSV measurement could compare city-level street trees, i.e., Changsha had a 1.2–1.3-fold higher diameter, tree height, and canopy size than the lowest at Nanchang; Changsha had the highest GVI, but the lowest NDVI; while Nanchang had the lowest GVI, but the highest NDVI, showing different significance between street-view and remote-sensing greenness. BSV data could identify urbanization effects, i.e., the tree sizes decreased from the urban center to the edge, and three cities showed the same pattern. By using BSV data, the growth status score was summarized from leaf blade color, tree-supported, dieback, and dead-tree percents, and thereafter, urbanization-induced decreases of growth status scores can be observed from the city center to the suburbs. At the same urban-rural gradient, BSV data found the vertical structure tended to be complex, and the sky and middle GVI decreased while the ground GVI rose in all three cities in general. The BSV data made landscape mapping available, and the map can be used to identify risk areas of big-sized tree conservation, health problems, greenness deficiency, and the over-simplified forest structure within a city or among different cities. Together with BSV-derived street development data, complex associations between tree characteristics and geo-climates and street social development could be decoupled by redundancy ordination. In general, streets with more pedestrians were often accompanied by taller, larger, healthier trees, more complex communities, and an increasing sky/ground ratio of GVI greenery. Our findings highlighted that street-view data could be used to assess street tree characteristics, geographic distribution, and possible associations with other factors, favoring the city-level urban tree inventory from individuals to the landscape for urban tree management and policymaking.