Road Network Centrality Reveals Architectural Heritage-Driven Road Patterns in Ming–Qing Ganzhou Ancient City

Technical Summary: Road Network Centrality Reveals Architectural Heritage-Driven Road Patterns in Ming–Qing Ganzhou Ancient City

Problem Statement
The reconstruction of historical urban road networks is frequently hindered by the limitations of schematic historical maps, which often prioritize city walls and principal thoroughfares while omitting secondary streets and alleys. Furthermore, existing methodologies often treat the reconstruction of street networks and the quantitative analysis of heritage–network relationships as separate tasks. This separation leaves a gap in understanding how persistent architectural heritage (such as government offices, religious buildings, and landmarks) spatially corresponds with reconstructed road structures, particularly when historical cartographic evidence is incomplete or non-metric.

Methodology
The study proposes a "reverse-inference" method based on the spatial anchoring between architectural heritage and road networks, utilizing Ganzhou Ancient City (Ming Wanli to Qing Qianlong periods) as a case study. The methodology integrates Historical GIS, multi-centrality analysis, and spatial statistics through the following steps:

1. Data Reconstruction and Source Criticism:

   • Baseline: A high-precision road network for the early People's Republic of China (PRC) was reconstructed using 1963/1969 satellite imagery and stable surviving features.
   • Historical Layers: Road networks for the Ming Wanli and Qing Qianlong periods were reconstructed using historical maps (e.g., Reprinted Gazetteer of Ganzhen), local gazetteers, and documentary records.
   • Workflow: Georeferencing was performed using stable control features (walls, gates, landmarks) to establish relative spatial correspondence rather than modern metric precision. Road centerlines were manually digitized, and topological validation was applied. Architectural heritage sites were maintained as a separate analytical layer with chronological and functional attributes.

2. Network Analysis (Multi-Centrality):

   • Using a primal node–edge representation, three centrality metrics were calculated for each period using the Urban Network Analysis (UNA) toolbox:
     • Straightness: Measures route efficiency (deviation from Euclidean distance).
     • Closeness: Measures overall accessibility to the network.
     • Betweenness: Measures intermediary control and flow mediation.

3. Spatial Association Analysis:

   • Kernel Density Estimation (KDE): Both heritage point data and centrality values were transformed into continuous raster surfaces to enable spatial comparison.
   • Bivariate Spatial Autocorrelation: Global bivariate Moran's I was used to test the systematic spatial association between heritage density and road centrality.
   • Local Indicators of Spatial Association (LISA): Local analysis identified specific clusters (High-High, Low-Low, High-Low, Low-High) to interpret local structural differences among heritage types (Buddhist, Daoist, Confucian, Administrative).

4. Reverse Inference Framework:

   • The study utilized the verified heritage–road coupling from the early PRC period as a quantitative constraint.
   • Multiple candidate road network scenarios for the Ming and Qing periods were generated based on historical constraints (walls, gates, axes).
   • Each candidate was evaluated using the same centrality and autocorrelation workflow. The scenario demonstrating the strongest statistically supported heritage–centrality association, while remaining consistent with historical evidence, was selected as the preferred reconstruction. The framework explicitly avoids assuming that buildings generated roads or using spatial association as proof of a uniquely correct geometry; instead, it evaluates how spatial correspondence changes across periods.

Key Results

• Validation of Coupling: In the early PRC period, a measurable and statistically significant positive spatial correlation was found between architectural heritage distribution and road network centrality (Straightness, Closeness, and Betweenness). This confirmed the existence of a heritage–road network coupling relationship and provided a quantitative basis for using heritage distribution as a spatial constraint in reconstructing earlier road networks.
• Evolution of Spatial Structure:
  • Ming Wanli Period: The city exhibited a dispersed, multi-core pattern. Architectural heritage was distributed across multiple nodes (Daoist temples in the northwest, religious clusters in the southwest/southeast), and road centrality showed weak centralization with dispersed high-value areas.
  • Qing Qianlong Period: The structure shifted to a "one-core, two-zone" pattern centered on the Zhenyuan Tower. Heritage density became more compact and orderly, with religious functions forming a spatial axis along the main thoroughfare.
• Changes in Heritage–Network Association:
  • Ming Period: Strong positive coupling (Moran's I: Straightness 0.418, Betweenness 0.385) indicated that buildings were deeply embedded in high-grade arterial roads and transport hubs, reflecting a layout controlled by military and administrative order.
  • Qing Period: The coupling strength weakened (Moran's I: Straightness 0.230, Betweenness 0.116). While the skeletal road structure remained, the built environment expanded into secondary streets and alleys. The decline in correlation suggests a shift from a rigid, axis-centered model to a more organic, multi-node distributed system driven by commerce and daily life.
• Heritage Type Specifics: LISA analysis revealed that while administrative and religious buildings maintained strong ties to main axes in the Ming period, their control over building layout weakened in the Qing period. Daoist temples showed increased penetration into secondary residential spaces, indicating greater functional flexibility.

Significance and Claims
The paper claims to provide a transparent analytical framework for reconstructing historical urban structures under conditions of incomplete cartographic evidence. Its primary contributions are:

1. Methodological Innovation: It demonstrates a quantitative approach to using existing architectural heritage as a relational constraint to evaluate and select among candidate historical road networks, moving beyond simple geometric alignment of historical maps without claiming to prove a uniquely correct geometry.
2. Quantitative Historical Insight: The study offers a quantitative framework to trace the evolution of urban spatial organization, revealing the transition of Ganzhou from a highly controlled, military-defensive spatial form (Ming) to a networked, polycentric, and organically expanded form (Qing).
3. Conservation Application: The findings suggest that contemporary historic city conservation should shift from protecting individual buildings to preserving the overall spatial structure and ecological relationships. By identifying structural elements that remain stable (such as the coupling between heritage and high-centrality roads), planners can better evaluate the impacts of modern interventions on historical urban character.

The authors conclude that the road network is not merely transport infrastructure but a spatial framework produced by long-term urban functions and social order. The persistence of the heritage–road network integration in Ganzhou explains the stability of the city's spatial pattern over centuries.