公园内外景观格局对城市公园降温效应的影响研究

landscape architecture; park cooling effect; landscape patternindex; random forest regression; Hangzhou; urban green space system planning

关键词:  风景园林  公园降温效应  景观格局指数  随机森林回归  杭州  绿地系统规划设计

Key words:  landscape architecture  park cooling effect  landscape pattern index  random forest regression  Hangzhou  urban green space system planning

摘要:

城市公园在对周边环境的降温中起到关键作用。过去的研究过度关注公园内部环境对降温效应的影响，却忽略了外部环境的潜在作用。研究杭州主城区 公园内外景观格局指数对降温距离的影响，以及确定了最优降温距离测量方法和最佳的回归模型，同时分析了公园内外景观格局指数与降温强度、梯度和效率之 间的关系。利用卫星遥感影像、杭州250个公园AOI数据和杭州土地覆盖数据，建立联立方程模型和随机森林回归模型，探讨了公园内外土地覆盖的景观格局指 数与降温效应的相关性和各变量在模型中的重要性。1)发现随机森林模型在解释公园降温距离方面更为出色，其中以斜率法-随机森林模型的降温距离解释度最 高(R2=0.97)。2)在随机森林模型下，公园的TA和LSI指数仍然是影响降温强度的重要因素；降温效率更加关注于公园内景观格局的影响，而降温梯度更注重缓 冲区景观格局的作用。3)缓冲区AIWater(99.69%)对降温强度的影响最为显著；在降温效率方面，内外乔木覆盖的作用更为突出，能够有效减缓温度上升；在降 温梯度的讨论中，乔灌草层次展现出实现更佳降温条件的潜力。斜率法是最佳的公园降温距离测定方法，非线性的随机森林回归模型能对降温效应做出更好的解 释；杭州城市绿地规划设计过程中要注重公园中优势斑块的DIVISION(54.48%)扩大有限条件下的降温效应，并在公园外绿地中注重LSICropland(94.69%)，使 得公园降温效应影响更大的范围。

Abstract:

Urban parks serve as vital green infrastructure in combating urban heat islands (UHI), yet the interplay between their internal and external landscape configurations in driving cooling effects remains underexplored. While prior research has extensively examined intra-park factors such as vegetation composition and waterbody distribution, the role of peri-park landscapes in modulating thermal mitigation remains poorly understood. This study addresses this knowledge gap through a comprehensive investigation of 250 parks in Hangzhou's core urban area, integrating multi-source geospatial data with advanced modeling techniques to unravel how synergistic interactions between park interiors and their buffer zones govern cooling performance. Methodologically, the research combines Landsat-8 thermal imagery (30 m resolution) with high-precision park AOI boundaries and land cover classification data (1 m resolution) to establish three key cooling metrics: intensity (maximum temperature differential), gradient (spatial decay rate), and efficiency (cooling per unit area). A novel dual-model framework was implemented, pairing traditional simultaneous equation modeling (SEM) with machine learning-based Random Forest (RF) regression. Landscape pattern indices were calculated at multiple scales using FRAGSTATS, encompassing class-level metrics (TA: total area; LSI: landscape shape index) and configuration parameters (AI: aggregation index) for both park cores and 500 m buffer zones. The RF model demonstrated exceptional explanatory power, particularly when coupled with the slope method for cooling distance determination (R2=0.97), outperforming conventional SEM approaches by 22%-35% across all metrics. Three critical dimensions of cooling dynamics emerged: 1) Cooling Intensity showed the strongest correlation with buffer zone Water aggregation (AI-Water: 99.69% importance), where strategically distributed aquatic features in peri-park areas amplified thermal differentials by enhancing evaporative cooling pathways. Intra-park TA and LSI maintained secondary influence (78.42% combined), suggesting that larger, irregularly shaped parks better disrupt heat accumulation patterns. 2) Cooling Efficiency predominantly responded to vertical vegetation structure, with tree canopy coverage in both park interiors (63.15%) and buffers (58.92%) proving crucial. The synergy between core green spaces and surrounding arboreal networks created continuous cooling corridors, reducing energy input required for temperature regulation. 3) Cooling Gradient optimization relied on multi-layered vegetation (tree-shrub-grass composites), which extended thermal decay distances by 18%-27% compared to mono-structured green spaces. Buffer zone cropland edge complexity (LSI-Cropland: 94.69%) unexpectedly emerged as a key gradient modulator, likely through enhanced air flow turbulence at agricultural-urban interfaces. Spatial analysis revealed non-linear threshold effects: parks exceeding 5ha exhibited disproportionate cooling benefits (+0.38/hm2 beyond this size), while buffer zone green coverage below 30% triggered rapid cooling capacity deterioration. The slope method's superiority in distance measurement stems from its capacity to capture inflection points in thermal decay curves, particularly where abrupt land use transitions occur at park peripheries. From an urban planning perspective, the findings challenge conventional park-centric design paradigms. While maximizing core area division (54.48%) remains essential for localized cooling, strategic buffer zone interventions yield system-level improvements. For instance, increasing LSI-Cropland complexity through agroforestry belts could extend park influence radii by 150-200 m. The study further proposes a tiered design framework: For micro-scale (<2 hm2 parks), prioritize vertical vegetation stratification and water feature integration to compensate for the limited area. For meso-scale (2-10 hm2), optimize edge permeability through sinuous boundaries and mixed-use buffer plantings. For macro-scale (>10 hm2), implement radial green wedges aligned with prevailing winds to amplify regional cooling effects. Notably, the research identifies critical trade-offs: excessive pursuit of park shape complexity (LSI>35) paradoxically reduces cooling efficiency due to edge effect proliferation, while over-aggregated buffer zone greenspaces (AI>75) inhibit cool air dispersion. These findings carry particular relevance for Hangzhou's ongoing ecological modernization, suggesting that retrofitting 10%-15% of peri-park impervious surfaces with stratified vegetation could yield 1.2-1.8, neighborhood cooling during peak heat events. This study advances UHI mitigation theory by establishing a holistic "core-buffer-system" analytical framework, demonstrating that park cooling efficacy constitutes an emergent property of landscape matrix interactions rather than isolated green space characteristics. Future research directions include temporal analysis of seasonal variation effects and 3D urban canopy layer modeling to refine vertical vegetation cooling mechanisms. The methodology and insights provide actionable pathways for climate-resilient city planning, particularly in high-density Asian megacities facing escalating thermal stress challenges.