Abstract: With the increasing urban traffic problems, urban managers are seeking the way to guide the transformation of traffic patterns through urban spatial planning, and to balance the relationship between supply and demand and alleviate traffic problems. As an important step of urban spatial planning, however, center structure identification is still limited in analytical methods and data applications. Using the commuting data from Baidu location dataset, this paper established the grid-based DBSCAN(Density-Based Spatial Clustering of Applications with Noise) density clustering algorithm to identify the center layout of 35 major metropolitan areas in China. The paper also analyzed the center structures of metropolitan areas through developing five commuting theoretical models and determining corresponding quantitative indicators. A regression analysis was also performed to identify the factors that affect the commuting efficiency of the metropolitan areas. Results show that center layouts of different metropolitan areas are obviously different and most metropolitan areas exhibit an unbalanced polycentric structure. The distribution of the center structure has some regional characteristics. For example, the monocentric metropolitan center structure is mainly found in the central and western cities of China. The constrained diffusion and balanced polycentric structure are mostlydistributed in the east coastal cities. The city size and commuting time shows the most significant correlation. The workhousing balance also has a great impact on commuting time. The results of this study help to determine the effective strategies for resource allocation and commuting efficiency optimization for metropolitan areas which also provides useful references for the spatial planning and sustainable development of transportation in metropolitan areas.

Key words: urban traffic, center structure, commuting patterns, metropolitan areas, commuting efficiency

摘要： 面对日益严重的城市交通问题，城市管理者尝试从城市空间规划层面引导交通出行模式的转型，调节供需关系，缓解交通问题，而中心结构识别作为城市空间规划的重要步骤，在分析方法与数据应用等方面仍然受限。本文利用百度位置数据挖掘出的都市圈通勤数据，采用基于网格的DBSCAN(Density-Based Spatial Clustering of Applications with Noise)密度聚类算法，识别出我国35个主要都市圈的中心布局，并根据5种通勤模式的理论模型和相应的量化指标对都市圈中心结构进行判别，最后对主要中心结构下影响通勤效率的因素进行回归分析。研究发现，虽然不同都市圈的中心布局差异明显，但大部分都市圈表现为非均衡多中心结构。不同中心结构的 分布呈现出一定的地域特征，单中心结构都市圈主要位于中西部城市，而约束扩散结构和均衡多中心结构都市圈大多是由东部沿海开放城市发展形成。城市规模与通勤时耗的相关性最为显著，职住平衡度对通勤时耗也有较大影响。上述研究结果为针对性地制定不同都市圈的资源配 置策略和通勤效率优化策略提供了有效支撑，对都市圈空间规划和交通可持续发展具有一定的理论价值和现实指导意义。

关键词: 城市交通, 中心结构, 通勤模式, 都市圈, 通勤效率