Aggregated data for "Integrating spatial field and network perspectives for intra-city tourism flow analysis: GPS trajectory evidence from Huangshi, China"

## 1. Overview All files in this repository are derived from raw GPS trajectories collected from the **Liang Bu Lu (Two-Step Road, www.2bulu.com)** outdoor platform for four equally spaced years (2016, 2019, 2022, 2025). Raw GPS trajectories cannot be redistributed in this repository due to the platform's terms of service. Researchers who require raw trajectory data should contact the platform directly via https://www.2bulu.com. The aggregated files provided here are sufficient to reproduce all tables and figure values reported in the manuscript, including the buffer-radius sensitivity analysis and the sample-size down-sampling robustness check. **Note on reproducibility**: All node-assignment and network indicators were generated using a unified pipeline in which each GPS point is assigned to its nearest tourism node within the buffer radius (nearest-node assignment via KD-tree). All spatial-autocorrelation statistics (Moran's I, LISA, Gi*) use an administratively clipped fishnet. Permutation-based computations use fixed random seeds to ensure full reproducibility. --- ## 2. File Index ### 2.1 OD matrices (`od_matrices/`) | File | Content | |------|---------| | `od_matrix_2016_v32.csv` | 32 × 32 directed weighted origin-destination matrix for 2016. Rows = origin, columns = destination. Cell value = number of trajectories observed flowing from row node to column node. | | `od_matrix_2019_v32.csv` | Same structure, year 2019. | | `od_matrix_2022_v32.csv` | Same structure, year 2022. | | `od_matrix_2025_v32.csv` | Same structure, year 2025. | Nodes are labeled N01–N32 (see `tourism_nodes/tourism_nodes_32.csv`). Main analysis uses a 1.5 km buffer with nearest-node assignment. ### 2.2 Tourism nodes (`tourism_nodes/`) | File | Content | |------|---------| | `tourism_nodes_32.csv` | List of 32 tourism nodes. Columns: ID, Name (English), Type (4A / 3A / 2A / Non-A / Transport hub), Longitude (°E, WGS84), Latitude (°N, WGS84), District/County. Corresponds to Supplementary Table S3. | ### 2.3 Network indicators (`network_indicators/`) | File | Content | |------|---------| | `SNA_results_v32.xlsx` | Multi-sheet workbook. Per-node centrality (in-degree, out-degree, total degree, in-strength, out-strength, total flow, betweenness, closeness, PageRank, eigenvector centrality), core/semi-periphery/periphery classification, and community membership for each of the four years. The "网络整体指标 / overall" sheet contains the cross-year overall network indicators corresponding to **Table 4**. | | `network_buffer_sensitivity.csv` | Overall network indicators under three buffer radii (1.0 / 1.5 / 2.0 km) for each year. The 1.5 km group matches Table 4. Corresponds to **Supplementary Table S4**. | | `subsample_network_robustness.csv` | Sample-size down-sampling robustness check for network indicators. Each year's trajectories are randomly down-sampled to the smallest-year baseline (3,500 trajectories), repeated 50 times; network density, weakly connected components, and modularity Q are recomputed per replicate. Supports the down-sampling analysis in **Section 4.3**. | ### 2.4 Spatial statistics (`spatial_statistics/`) | File | Content | |------|---------| | `sde_summary.csv` | Standard deviational ellipse and centroid parameters for each year (semi-major axis, semi-minor axis, azimuth, area, flattening, centroid coordinates, migration distance). Corresponds to **Table 2**. | | `moran_sensitivity_clipped.csv` | Global Moran's I under the administratively clipped fishnet at three resolutions (0.005° / 0.01° / 0.02°) for each year. Corresponds to **Supplementary Table S1**. | | `gistar_hotspot_counts.csv` | Getis-Ord Gi* significant hotspot cell counts per year at the main scale (0.01°): total hotspots, Hot 95%, Hot 90%, total coldspots. Corresponds to **Table 3**. | | `hotspot_2016_10.csv` to `hotspot_2025_10.csv` | Per-cell results at the main scale (0.01°, administratively clipped fishnet). Columns: cell ID, centroid longitude, centroid latitude, point count, log-count, Gi* z-score, Gi* p-value, Gi* classification, local Moran quadrant, LISA p-value, LISA classification. Supports **Supplementary Table S2** and the Gi*/LISA discussion in Section 4.1. | | `lisa_subsample_robustness.csv` | Sample-size down-sampling robustness check for the LISA high-high (HH) cluster count, using the same 3,500-trajectory baseline and 50 replicates. Supports the down-sampling analysis in **Section 4.3**. | | `kde_2016.csv` to `kde_2025.csv` | KDE grid values per year. Columns: longitude, latitude, raw count, KDE value, normalized KDE. Supports **Figure 2**. |