关键词: 城市热岛, 蓝绿空间, 降温效应, 景观配置, 景观构形, 情景模拟

Abstract: Due to climate change, the urban temperature in summer has increased gradually as a result of higher temperatures and the urban heat island effect. This poses a threat to the physical and mental health of residents. Optimizing the layout of green-blue spaces to achieve a cooling effect has become a crucial issue, especially in high-density cities with land resource shortages. The cooling effect of green-blue spaces is influenced by the built environment, its characteristics, and multiple other factors. Given smart growth, measures to improve the cooling effect across various dimensions, such as the occupancy, composition, and configuration of blue-green spaces within different built environments, are essential for optimizing outdoor thermal comfort and improving land-use efficiency in urban areas. Based on a review of the factors that influence the cooling effect of blue-green spaces, a multi-dimensional collaborative research framework and method system was established. After sample division by unsupervised algorithms based on architectural environmental features, a scenario simulation model was constructed using the machine learning algorithm to analyze the influences of occupancy, composition, and configuration on the cooling effects of blue-green spaces. Moreover, the interactive effects of different dimensions were explored through the generalized additive model. Based on the above analysis results, the optimization of the blue-green space layout in accordance with local conditions was proposed. The proposed methodological framework was implemented in the Central City of Shanghai, which is a typical built area with a high density and high mixing degree. The samples used to build the scenario simulation model included 5,112 blocks enclosed by lowest-level roads. These blocks were divided into four types, including an open space group, a low building intensity group, a medium building intensity group, and a high building intensity group. The basic model for scenario simulation was built based on the random forest model with good prediction ability. The fitting degree between the scenario simulation model and real temperature was 0.9074, indicating that the scenario simulation model had good reliability. The single-dimensional optimization scenarios included increasing blue-green space occupancy, optimizing composition, and optimizing configuration. The comprehensive improvement scenario involved optimization and adjustment of both composition and configuration while improving blue-green space occupancy. According to simulation results for the cooling effect of different scenarios, the degree of improvement in the cooling effect by blue-green space layout optimization was influenced by the built environment. Generally speaking, the cooling effect brought about by blue-green space, especially with optimization of occupancy, decreased with increases in construction intensity. In the single-dimensional improvement scenario, increasing the occupancy of blue-green space improved the cooling effect the most, followed by increasing the composition. However, it can be seen from the multi-dimensional interactive analysis curves, that the cooling effect brought about by increasing the occupancy and configuration of blue-green space was unsatisfactory when the sample block was relatively small. In summary, the multi-dimensional comprehensive scenario has better effects, especially in communities with a low construction intensity.

Key words: urban heat island, green-blue spaces, cooling effect, landscape composition, landscape configuration, scenario simulation