Abstract:

Using Tencent location data in three times: working days, rest days, and holidays, we studied the temporal variation characteristics of the activity space in Nanchang City. First, the activity spatial cohesion structure of the study area is analyzed by the nuclear density analysis method. Then, the specific urban activity area is divided and analyzed by the method of heat peak point extraction and Tyson polygon.The research found that: 1)The area of ??high-frequency activity area on working day is obviously higher than that of rest day and holiday. The larger the thermal value of the classification interval is, the smaller the number of pixels is. 2)There are four kinds of condensed structures of ??urban activity space in the study area, namely “mononuclear structure”, “dual-core structure”, “trinuclear structure” and “aggregate structure”. Each structure has unique morphological characteristics and representativeness. 3)Divide the study area into six kinds of activity space function zoning: comprehensive activity area, residential area, learning activity area, leisure and entertainment area, transportation hub area, employment area, and different activity patterns correspond to their respective activities frequency changes. Using the method of superposition analysis, the core area and the marginal area of ??the six functional modes are divided. 4)The activity function areas in the study area are unevenly distributed. Most of the activity types are residential activity areas and learning activity areas. 5)According to the distribution characteristics of the activity space structure and the distribution characteristics of the activity area discovered by big data mining, combined with the urban development plan of Nanchang City, the current activity space structure of Nanchang City was evaluated, and based on the above analysis, we made some recommendations on improving the planning of the city.

Key words: time and space location data, urban activity space structure, activity function zoning, Tencent, Nanchang

摘要：

近年来地理信息大数据运用在人文地理学领域成为了新热潮。本文借助腾讯位置数据，得到了工作日、休息日、节假日三种类型的数据，以南昌市为例，从居民实际活动的视角出发，首先对研究区域的活动空间时序变化特征进行研究，在此基础上，通过核密度分析法总结了研究区域的活动空间凝聚结构，最后利用热度山顶点提取和泰森多边形等方法对具体的城市活动空间功能区进行划分与分析。研究发现：①工作日的高频活动区域面积明显高于休息日和节假日，分类区间的热力值越大像元数量越小。②研究区域城市活动空间的凝聚结构有四种，分别为“单核结构”、“双核结构”、“三核结构”、“集聚结构”，每种结构都有独特的形态特征以及代表性的典型区域。③研究区域可划分为六种活动空间功能区划：综合活动区、居住活动区、学习活动区、休闲娱乐区、交通枢纽区、就业活动区，不同的活动模式对应着各自的活动频度变化规律，且运用叠加分析法识别出六种功能模式的核心区和边缘区。④研究区域的活动功能区分布不均，大部分的活动类型为居住活动区和学习活动区，但主城区的居民活动类型更加丰富，主城区的集聚程度高于远城区，且活动功能区与空间凝聚结构具有一定联系。⑤根据大数据挖掘发现的南昌市活动空间凝聚结构分布特征、活动功能区的分布特征，结合南昌市城市发展规划，对南昌市目前的活动空间结构进行评价，从而对昌南、昌北、城东的发展提出相应建议。

关键词: 时空位置数据, 城市活动空间结构, 活动功能区划, 腾讯, 南昌