Mapping sky, tree, and building view factors of street canyons in a high-density urban environment
Publication in refereed journal


Times Cited
Altmetrics Information
.

Other information
AbstractView factors for sky, trees, and buildings are three important parameters of the urban outdoor environment that describe the geometrical relationship between different surfaces from the perspective of radiative energy transfer. This study develops an approach for accurately estimating sky view factor (SVF), tree view factor (TVF), and building view factor (BVF) of street canyons in the high-density urban environment of Hong Kong using publicly available Google Street View (GSV) images and a deep-learning algorithm for extraction of street features (sky, trees, and buildings). As a result, SVF, TVF, and BVF maps of street canyons are generated. Verification using reference data of hemispheric photography from field surveys in compact high-rise and low-rise areas shows that the GSV-based VF estimates have a satisfying agreement with the reference data (all with R2 > 0.95), suggesting the effectiveness and high accuracy of the developed method. This is the first reported use of hemispheric photography for direct verification in a GSV-based streetscape study. Furthermore, a comparison between GSV-based and 3D-GIS-based SVFs shows that the two SVF estimates are significantly correlated (R2 = 0.40, p < 0.01) and show better agreement in high-density areas. However, the latter overestimates SVF by 0.11 on average, and the differences between them are significantly correlated with street trees (R2 = 0.53): the more street trees, the larger the difference. This suggests that a lack of street trees in a 3D-GIS model of street environments is the dominant factor contributing to the large discrepancies between the two datasets.
Acceptance Date26/02/2018
All Author(s) ListFang-Ying Gong, Zhao-Cheng Zeng, Fan Zhang, Xiaojiang Li, Edward Ng, Leslie K. Norford
Journal nameBuilding and Environment
Year2018
Month4
Day15
Volume Number134
Pages155 - 167
ISSN0360-1323
eISSN1873-684X
LanguagesEnglish-United States
KeywordsView factor, Google Street View, Deep learning, Street trees, Street canyon, High density

Last updated on 2020-28-03 at 02:09