Data-Driven Research On Street Environmental Qualities And Vitality Using Gis Mapping And Machine Learning, A Case Study Of Ma On Shan, Hong Kong
Xinyu Liu The Chinese University of Hong Kong
Jeroen Van Ameijde The Chinese University of Hong Kong
As the global pandemic and extreme weather continue to impact lives (Walton et al., 2020), an increased awareness of impact of industrialization and urbanization on climate change is emerging. In this context, the construction of low-carbon smart cities is promoted an effective way to improve the quality of urban spaces and implement sustainable development. In the post-carbon framework, data-driven methods can be applied to assess or improve the quality and sustainability of the urban environment. Streets form an important part of everyday life and enable daily travel, recreation and socializing. Improving walkability is of great significance to the construction of low-carbon and sustainable cities, and walkability is becoming an increasingly important concept in sustainable urban design practice (Grignaffini et al., 2008). One proposed definition for walkability is: “The extent to which the built environment is friendly to the presence of people living, shopping, visiting, enjoying or spending time in an area” (Abley, 2005, p.3). Making cities walkable is an important solution for adapting cities to climate change: it reduces carbon emissions, and improves quality of life (Project Drawdown, 2020). This study focuses on Ma On Shan new town, Hong Kong, as a case study area, to test a data-driven approach to evaluate the spatial qualities of and usage characteristics of the streets. The specific study covers an area of 31 streets, which were firstly analysed through a GIS platform to evaluate the street morphological characteristics. The spatial quality of the streets was analysed through gathering and processing a large number of Google Street View images, which were analysed through a Machine Learning algorithm designed to perform image recognition and semantic segmentation. Machine learning enables computers to learn without being explicitly programmed (Luo, 2018). Built on prior studies in measuring walkability (Ewing & Handy, 2009) and with emerging applications in urban perceptions using deep learning and street view images (Li et al., 2015; Qiu et al., 2021), the algorithms can recognize and calculate the amounts of greenery, sidewalk space, and pedestrians. Secondly, using SPSS statistical analysis software, the correlation between street morphological characteristics, street space quality and street users was analysed. Finally, randomly selected street users were asked to answer a questionnaire, to document their subjective evaluation on the morphological characteristics and spatial qualities of the street. These subjective evaluations helped to supplement the objective measurement results and improve the reliability and validity of the study. The results showed that greenness, safety and connectivity those indexes had positive and significant effects on walkability. The study demonstrates how the improvement of walkability can start from the quantitative design of urban street environmental indicators, and provides a reference for the construction of a green, low-carbon and healthy walking city. References: Walton, D. and Van Aalst, M.K. (2020). Climate-related extreme weather events and COVID-19. A first look at the number of people affected by intersecting disasters. IFRC, Geneva. 21 pp. ISBN ISBN/EAN: 978-90-818668-1-10. Grignaffini, S., Cappellanti, S. & Cefalo, A. (2008). Visualizing sustainability in urban conditions. WIT Transactions on Ecology and the Environment, Vol. 1, pp. 253-262, 10 Jun 2008. Abley, S. (2005) Walkability scoping paper. Christchurch, Nova Zelândia: Land Transport New Zealand Project Drawdown. (2020). Walkable Cities .Climate Solutions at Work. Project Drawdown. Climate Solutions at Work | Project Drawdown Luo, D., Wang, J., and Xu, W. (2018). Applied automatic machine learning process for material computation. International Conference on Education and Research in Computer Aided Architectural Design in Europe, Lodz, Poland, 17-21 September 2018. Lodz, Poland: eCAADe and Faculty of Civil Engineering, Architecture and Environmental Engineering, Lodz University of Technology. Ewing, R., & Handy, S. (2009). Measuring the Unmeasurable: Urban Design Qualities Related to Walkability. Journal of Urban Design, 14(1), 65–84. https://doi.org/10.1080/13574800802451155 Li, X., Zhang, C., Li, W., Ricard, R., Meng, Q. & Zhang, W. (2015). Assessing Street-Level Urban Greenery Using Google Street View and a Modified Green View Index. Urban Forestry & Urban Greening. 10.1016/j.ufug.2015.06.006. Qiu, W., Li, W., Zhang, Z., Li, X., Liu, X. & Huang, X. (2021). Subjective and Objective Measures of Streetscape Perceptions: Relationships with Property Value in Shanghai. 10.20944/preprints202103.0506.v1.
Keywords: Street Walkability, Geographic Information System, Machine Learning, Image Segmentation, Sdg11: Sustainable Cities And Communities