Environments for Climatic Studies Using Google Street View Fang-Ying - PowerPoint PPT Presentation

10th 10 th Inter Interna nationa tional Conf Confer erence ence on on Urban Urban Clima limate te (ICUC ICUC 10 10) Quantifying Street View Factors of High-Density Urban Environments for Climatic Studies Using Google Street View Fang-Ying Gong ( 龚芳颖 ) 1,2 * Collaborators: Zhao-Cheng Zeng 3 , Fan Zhang 4 , Edward Ng 1 , Leslie Norford 2 1 School of Architecture, The Chinese University of Hong Kong (CUHK) 2 Department of Architecture, Massachusetts Institute of Technology (MIT) 3 Division of Geological and Planetary Sciences, California Institute of Technology (Caltech) 4 Institute of Remote Sensing & Geographical Information System, Peking University (PKU)

Bac Backg kground ound Resear esearch Pr Problems oblems A typical street in high-density urban areas of Hong Kong is characterized by: • High-rise buildings; • Narrow compacted streets; • Heavy travel volume and pedestrian flow; • Complex streetscapes with colorful overhanging signboards that block sunlight and air paths; • Limited openness to the sky. Fig. 1. An example of the deep street canyon in the Mong Kok area in Hong Kong (Google Street View, 2016)

Bac Backg kground ound Resear esearch Pr Problems oblems One typical high-density high-rise urban areas with trees of Hong Kong. • It’s difficult to quantify the street features (tree canopy cover, building overhangs, and shade structures) using model methods in complex street environments. • In particular, street tree canopy, a major component of streetscapes, is hard to parameterize in models. ➢ An effective and accurate method for mapping the street features is Fig. 2. An example of deep street canyon in the Tsim Shi Tsui area crucial for studying its urban in Hong Kong. (Source: Google Street View, 2016) climate and assessing the relevant outdoor thermal comfort.

Backg Bac kground ound Resear esearch Questions Questions ➢ Question 1 ➢ Research Purpose How to use public-access Google Street View images to estimate the view factors? To develop an approach for estimating ➢ Question 2 and mapping sky, building, and tree What’s is the spatial patterns of the sky, tree, features of street canyons in complex and building features of street canyons in high- urban living environments. density urban environments? ➢ Question 3 What’s the differences of GSV-based and 3D- GIS-model estimate methods?

Method Metho Study Ar Study Area ea: Hong Hong Kong ong • High-density urban areas of Hong Kong are characterized by high-rise compact building blocks and deep street canyons with a high H/W ratio. • Tall buildings of some 40-60 stories with narrow streets. Fig. (a) Location of Hong Kong in south-eastern China; (b) High-density urban areas in Hong Kong (Kowloon and northern Hong Kong Island); (c) Building density map.

Method Metho Data Da ta Collection Collection ： Goog Google le Str Street eet View iew (GSV) (GSV) • Google Street View serves millions of Google users daily with panoramic imagery captured in hundreds of cities in 20 countries across four continents (Anguelov et al., 2010). • Streets with Street View imagery available are shown as blue lines on Google Maps. Fig. Google Street View coverage 3D map of Kowloon Area of Hong Kong (Google, 2016).

Metho Method View F iew Factor actor Calcula Calculations tions using using GSV ima GSV images ges Fig. Workflow procedure for calculation of view factors using Google Street View images ➢ Panorama images from Google Street View ➢ Features extraction using deep-learning framework. Sky (in blue ), tree (in green ) and building (in grey ) are extracted using the scene parsing method in a deep-learning framework (Zhou et al., 2016). ➢ Fisheye images obtained by projecting the panorama images ➢ Based on the classified fisheye image, view factors for sky ( SVF ), tree ( TVF ), and building ( BVF ) are calculated using the classical photographic method developed by Johnson and Watson (1984).

Method Metho Semant Semantic Scene P ic Scene Par arsing sing using using PSPNet PSPNet Fig. Workflow of semantic scene parsing using Pyramid Scene Parsing Network (PSPNet). The study uses the filtered 29.264 GSV images for further analysis. ➢ For a given input street view image in (a) , ➢ the network extracts the feature map in (b) , and then ➢ the pyramid parsing module is applied to form the final feature representation of the streetscape in (c) . ➢ Finally, a pixel-wise classified output street view image with semantic categories in (d).

Method Metho Accurac Accur acy y Asses Assessment of sment of Classi Classifica fication tion • This assessment is implemented by using 100 randomly street points (cover low-to-high building densities); • Comparing their calculated SVF, TVF, and BVF from sky, tree, and building features extracted using: (1) Scene parsing deep learning technique; (2) Manual delineation by eye inspection (as reference data). Fig. Accuracy assessment of feature extraction using the PSPNet in a deep-learning framework.

Results esults Tree V ee View F iew Factor (TVF) actor (TVF) • The TVF is dominated by values less than 0.1 , which is limited by the high building density and narrow street environment. • The high-density residential areas, which cover about 58% of the study area, are dominated by low TVF (0.0 – 0.3) , because of the high- density construction and narrow streets that limit space for greenery. Total Kowloon HK Area Island Fig. Mapping of Tree View Factor (TVF) estimates of SVF 0.49 0.53 0.41 street canyons derived using 29,264 Google Street TVF 0.14 0.12 0.19 View images along the streets at 30-meter intervals;

Results esults Building View F Building iew Factor (BVF) actor (BVF) • The coastline regions and low-rise areas, which cover about 20% of the study area, show much higher SVF (0.7 – 1.0), and lower BVF (0.0 – 0.3), because of fewer buildings and more sky openness. Total Kowloon HK Area Island SVF 0.49 0.53 0.41 BVF 0.33 0.31 0.36 Fig. Mapping of Building View Factor (BVF) estimates of street canyons derived using 29,264 Google Street View images along the streets at 30-meter intervals;

Results esults Sky V Sk y View F iew Factor actor (SVF) (SVF) • The spatial patterns of GSV-based SVF estimates are similar and consistent with the corresponding building height and density in build-up areas. • Areas with higher building density have lower SVF , lower TVF , and higher BVF , and vice versa. • Areas with higher tree canopy also have lower SVF , but higher TVF , and lower BVF . Fig. Mapping of GSV-based Sky View Factor (SVF) estimates of street canyons derived using 29,264 GSV images along the streets at 30-meter intervals;

Results esults Sky V Sk y View F iew Factor actor (SVF) (SVF) 3D-GIS SVF GSV-based SVF ➢ The map of 3D-GIS-based SVF shows a similar pattern to that of GSV-based SVF estimates. They are correlated ( R 2 = 0.40 ) and have a better agreement in high-building-density areas. ➢ The mean SVF value of 3D-GIS-based estimates ( 0.59 ) is about 0.11 (about 20%) higher than that of GSV-based estimates ( 0.49 ). ➢ There are large differences in the low-rise areas with large amount of street trees .

Results esults Verifica erification tion of of GSV GSV-based V based View F iew Factor actor Estima Estimates tes • Scatter plot of view factor data from field survey and the corresponding GSV- based estimates. • The sampling view factor data include 20 samples in Mong Kok within high-rise building area (in triangles), and 20 samples in Kowloon Tong within median-rise area (in circles). ➢ GSV-based estimations is the Fig. 1. Validation of TVF estimates Fig. 2. Validation of BVF estimates in high and low-density areas using in high and low-density areas using effectiveness and high accuracy hemispheric photography hemispheric photography method to quantify the tree canopy and building density.

Results esults Verifica erification tion of of GSV GSV-based V based View F iew Factor actor Estima Estimates tes • Scatter plot of SVF data from field survey and the corresponding GSV- based (in blue) and 3D-GIS-based (in red) SVF data. • The sampling SVF data include 20 samples in Mong Kok within high- rise building area (in triangles), and 20 samples in Kowloon Tong within median-rise area (in circles). ➢ GSV-based streetscape study is effective and accurate in high- density urban environments. Fig. 1. Validation of SVF estimates in high and low-density areas using hemispheric photography

Results esults Differ Dif erence betw ence between een 3D 3D-GIS GIS and GSV and GSV-based SVF based SVF Impact of Building View Factor Impact of Tree View Factor Fig. A. The bivariate histogram of GSV-based TVF estimates Fig. B. The bivariate histogram of GSV-based BVF estimates and difference between 3D GIS-based and GSV-based SVF and difference between 3D GIS-based and GSV-based SVF ➢ The higher of the H/W ratio, ➢ The higher of the amount of street trees, the smaller of the uncertainty the larger of the uncertainty of model of model simulation of SVF. simulation of SVF.