An approach to urban landscape character assessment: Linking urban big data and machine learning

Abstract

Diverse urban landscape is an important cultural driving force for urban sustainable development. Although characterizing landscape to protect landscape diversity is widely used in regional landscape and preservation practices, it is difficult to apply to urban landscape character assessment, which needs fine-scale data support, explicit study units, and effective clustering models. Therefore, this study uses urban big geospatial data and machine learning technology to establish a technical system for character assessment of urban landscape applicable to the block scale and complete the landscape assessment of urban areas of Beijing and Shanghai, China. A total of 64 landscape character types were identified in Beijing, and 61 in Shanghai. We find that (1) urban landscape characters are different with the ring road as the boundary, but each zone presents a combination of different proportions of landscape characters. (2) Beijing's city wall demolition policy is affected by historical protection policy. Landscape differentiation on both sides of the Huangpu River in Shanghai has yet to be realized. This study extends the theory of LCA and realizes the research exploration of urban built environment. It can also be used to guide urban zoning control, evaluate planning policy, and provide assistance in practice for sustainable urban development and management.