【摘要】 在全民健康背景下，步道作为公共服务设施的核心载体，其规划需兼顾多维景观信息与使用者偏好。然而，传统方法难以全面获取步道三维景观数据，且缺乏对用户偏好的精准评估。大数据显著提升了景观感知研究的精准性与全面性，但其与景观设计实践的深度融合亟待推进。以北京小西山为例，构建“数据获取—偏好评价—决策支撑”框架，结合地理信息数据与问卷调研，应对当下的实践决策难点。大数据的景观感知研究能够：（1）提供多维的步道信息，包括常用起止点、打卡点、步道网络以及区域层面的步道景观信息，揭示铺装前的自然路网中土石路（62.92%）与土路（22.90%）占主导；（2）揭示全域范围内三类使用者的偏好满足程度，发现铺装改造后（铺装路占39.44%），非徒步者满意度提升，但长途徒步者因偏好步道类型被破坏而满意度下降；（3）支撑步道空间规划优化。基于偏好差异提出空间优化策略，如恢复高偏好的原始步道类型，在低密度路网区开发针对不同人群的偏好道路类型，实现路线差异化设计。研究证实，多源数据融合能有效解决规划前端数据匮乏与实践中偏好落位难题，并通过“信息获取—需求诊断—决策支撑”路径推动景观感知研究向循证设计转化，为步道建设提供科学依据。更多还原

【Abstract】 Within the context of public health initiatives, trails serve as critical infrastructure in public service systems, requiring planning that integrates multidimensional landscape information and user preferences. However, conventional methodologies struggle to comprehensively capture three-dimensional(3D) trail landscape data and lack the capacity to quantify user preferences precisely. While big data has significantly enhanced the accuracy and comprehensiveness of landscape perception research, its deep integration with landscape design practice remains an urgent need. This study employs Beijing’s Xiaoxishan area as a case study to establish a “Data Acquisition-Preference Evaluation-Decision Support” framework that integrates geospatial data with questionnaire surveys to address current practical decision-making challenges. Big data-enabled landscape perception research demonstrates its capacity to:(1) Provide multi-dimensional trail information, including frequently used start/end points, photo-taking hotspots, trail networks, and regional trail landscape information, revealing that unpaved trails dominates the natural network before paving(rock-dirt trails: 62.92%; dirt trails: 22.90%);(2) Quantify preference fulfillment levels among three distinct user groups across the entire area, revealing that paving(39.44% paved trails) improves non-hiker satisfaction but decreases long-distance hiker satisfaction due to altered trail types;(3) Inform spatial planning optimization. Based on preference-based disparities, spatial optimization strategies are formulated, including restoration of high-preference natural trail types and development of user-specific trail typologies in low-density network zones to achieve route diff erentiation. This study confirms that multi-source data fusion eff ectively addresses foundational planning data scarcity and the challenge of preference localization during implementation. Furthermore, it advances the translation of landscape perception into evidence-based design through the “Information Acquisition-Demand Diagnosis-Decision Support”, thereby establishing a robust scientific foundation for trail development.更多还原