Dynamics of street environmental features and emotional responses in urban areas: Implications for public health and sustainable development

Yangfei Huang^1,2*Chenjian Zhong^1,3Tangtao He⁴Yuyang Jiang⁵

• ¹School of Civil Engineering & Architecture, Zhejiang University of Science & Technology, Hangzhou, Jiangsu Province, China
• ²Center of Urban and Rural Development, Zhejiang University of Science & Technology, Hangzhou, Jiangsu Province, China
• ³Zhejiang - Singapore Joint Laboratory for Urban Renewal and Future City, Hangzhou, Jiangsu Province, China
• ⁴College of Architecture and Urban Planning, Guangzhou University, Guangzhou, China
• ⁵School of Geography and Planning, Cardiff University, McKenzie House, Cardiff, United Kingdom

Introduce: Urban street spatial quality, as an intervenable environmental factor from the perspective of public health, significantly affects residents' mental health and emotional well-being. Accurately identifying emotional hot spots in urban environment and exploring the mechanism of environmental features affecting emotions are crucial for improving residents' mental health level, promoting healthy urban planning and creating a sustainable urban environment.Methods: This study employed an interdisciplinary approach, utilizing street view images from Liwan District, Guangzhou, China. A Pyramid Scene Parsing Network (PSPNet) was applied to quantify 18 key environmental features, including the Green View Index (GVI), Space Openness (SO), Enclosure Index (EI), etc. By integrating an emotion dataset assessed by 40 experts, a random forest model was constructed to predict emotional responses to different street spaces. Emotional distribution maps were generated using ArcGIS Pro to identify emotional hotspots. Subsequently, SHAP (SHapley Additive exPlanations) analysis was conducted to explore how environmental features influence emotional responses.The analysis revealed the following: (1) Positive emotions were significantly associated with areas of well-vegetated, while negative emotions were predominantly concentrated in industrial zones and narrow alleys. (2) GVI, Sky-Green ratio, EI, and SO had a notable impact on emotional responses. (3) The optimal range for the GVI (0.27-0.3) was found to maximize positive emotional valence. Beyond this range, further increases in the GVI did not result in significant emotional changes.Discussion: This study demonstrates the feasibility of predicting public emotional responses from street view images using machine learning. Optimizing green spaces and improving pedestrian environments can promote emotional health. To effectively balance the distribution of urban green spaces and maximize public health benefits, it is recommended that governments collaborate with communities, leveraging fiscal incentives and green infrastructure investments to promote equitable and sustainable development of green spaces. These findings play a crucial role in advancing both public health and environmental sustainability.