关键词: 城市绿地, 热环境, 空间格局, 季节差异, 增强回归树

Abstract: Rapid urbanization has become inevitable in the current stage of human development. Accelerated urbanization has led to significant changes in vegetation. The increasingly impermeable surface, decreasing urban green space, and intense landscape fragmentation may exacerbate the effects of urban heat islands, resulting in potential adverse effects on human life. Urban green space is an important component of the urban ecosystem. Many studies have extensively demonstrated the important role of urban green space in regulating regional climate and mitigating urban heat via field measurements, numerical simulations, and remote sensing inversions. Specifically, the surface temperature determined by remote sensing inversion not only reflects the spatial distribution of the thermal environment in the urban green space instantaneously, but also effectively connects the various methods of landscape pattern analysis. Hence, remote sensing inversion represents an important method used to study thermal environment in urban green space currently. Although the associated studies involve relatively rich overall content, they provide no systematic analysis of the nonlinear relationship between green space pattern and urban thermal environment as well as the threshold characteristics. They only provide guidance for the optimal design of urban green land by measuring the nonlinear characteristics of different landscape pattern indices and urban thermal environment as well as the threshold. Therefore, it is necessary to introduce a new analytic method based on different spatiotemporal scales in order to fully understand the effect of urban green landscape on urban thermal environment.
　　A case study based on the main urban area of Hefei City was carried out. The quantitative relationship between green space pattern and urban thermal environment in different seasons was analyzed via landscape pattern analysis and Sentinel-2A and Landsat8 OLI/TIRS remote sensing images in the GEE cloud platform. First, the green space layout and surface temperature distribution characteristics of the study area in different seasons were determined based on Sentinel-2A and Landsat8 OLI/TIRS remote sensing images. Subsequently, the landscape patterns of the urban green land in the study area under different research scales were investigated by moving window method in the Fragstats 4.2 software. Finally, the optimal scale to analyze the thermal environment of green space in Hefei was explored using the ArcGIS Pro software and the boosted regression tree model. Meanwhile, the relationship between the green space layout and urban thermal environment as well as seasonal differences including relative contribution and marginal effect were analyzed.
　　The results showed that the urban green space significantly regulated the thermal environment. It decreased the temperature in spring, summer and autumn, but increased the temperature slightly in winter. Second, the importance of green landscape composition and configuration to thermal environment varied across different seasons. Landscape composition affected the thermal environment more than the landscape configuration in warm seasons, whereas the opposite phenomenon was observed during winter. The marginal effect of green landscape composition and configuration index on the thermal environment varied. Based on the landscape composition, PLAND was negatively correlated with LST in all seasons, which suggested a uniform cooling effect. Based on landscape configuration, AI, AERA_MN and ENN_MN were negatively correlated with LST in most seasons, while LSI, ED and PD showed positive and negative correlations before and after the threshold. Thus, the structures with simple shape, low fragmentation, high degree of aggregation, and low connectivity can be used in small green space, whereas structures with complex shapes, high fragmentation, high degree of aggregation, and low connectivity can be utilized in large green space, which decreased the local LST. In brief,, future urban planning and landscape design to ensure optimal thermal environment should ensure a reasonable balance between urban green landscape composition and configuration. This study provides a standard of reference for urban planners to manage high urban temperatures and adapt to climate change.

Key words: urban green space, thermal environment, spatial layout, seasonal difference, boosted regression tree