Research on the Spatial Clustering of Chinese Cities Based on Complex Network Analysis

The process to obtain the built-up area data involves the following steps: first, generating standard false-color images through band fusion; second, conducting land-use classification using the Support Vector Machine (SVM) method; finally, extracting built-up areas by combining administrative boundaries, with accuracy verified via a confusion matrix.Road raw dataRoad data of all class-level and above highways in China was obtained from the OpenStreetMap platform.

The processing of this raw road network includes data cleaning and deduplication. A distance measurement tool is used to calculate multi-route distances between cities, and the shortest path between each pair of cities is selected to form the Processed road network. Subsequently, cities are abstracted as nodes and the shortest-path road network as edges.

The max spanning tree is constructed based on the Kruskal algorithm, and pruning is performed by setting a threshold using betweenness centrality. Finally, the spatial scope of urban clusters is determined in combination with the convex hull algorithm.