EUVIMEDEUVIMED entdecken
Open Access · CC BY 4.0

Spatial patterns and nonlinear drivers of traditional villages: a grid-based machine learning analysis in Yunnan, China

Xinyu Huang, Yalan Cheng, Linhan Jiang, Zhuoran Li, Yue Deng, Xueyue Bai

Frontiers in Environmental Science · 08.07.2026 · 14

DOI: 10.3389/fenvs.2026.1834584 ↗

Kurz gefasst

Fachartikel in der Zeitschrift Frontiers in Environmental Science, online erschienen am 08.07.2026. Verfasst von Xinyu Huang und 5 weiteren Autor:innen. Die Arbeit nennt 36 Literaturangaben. Der Volltext ist frei zugänglich (Open Access, CC BY 4.0). Gefördert von: Beijing Forestry University.

AbstractBibliografische AngabenZitierenVerwandte Studien

Abstract (Originalsprache)

Traditional villages represent critical components of the historic built environment, reflecting long-term human–environment adaptation in complex mountainous systems. Understanding their spatial distribution is crucial for effective rural governance and the conservation of traditional village cultural heritage. Focusing on Yunnan Province, China, home to 777 nationally recognized traditional villages, this study aims to examine spatial heterogeneity and nonlinear driving mechanisms at the regional scale. A grid-based analytical framework was constructed using 10 km × 10 km units, generating 4,994 valid grid cells across the province. Landforms were classified into six geomorphological categories representing a continuous terrain gradient. Multiple environmental and socio-economic variables, including elevation, slope, precipitation, temperature, river density, road density, population density, GDP, and distance to towns, were analyzed using nonparametric statistics, Spearman’s correlation, Random Forest regression, and Partial Dependence Plots (PDP). The results show that significant distributional differences exist across landform types (p ≪ 0.001). Traditional villages exhibit a pronounced intermediate peak pattern, concentrating in moderately undulating mountainous areas rather than plains or extremely rugged terrains. Random Forest results indicate that road density (FI = 0.144), mean annual precipitation (FI = 0.128), GDP (FI = 0.122), and elevation (FI = 0.121) contribute most to explaining spatial variation. PDP analysis further shows nonlinear response trends, including mid-elevation preferences and inverted-U responses to accessibility. Field investigations of 25 traditional villages across different geomorphological contexts suggest that observed settlement patterns are broadly consistent with the modeled spatial trends. These findings suggest that traditional village distribution reflects complex human–environment interactions shaped by terrain structure, hydrological conditions, and moderated accessibility. This study provides a grid-based and data-driven analytical approach for understanding the spatial patterns of traditional villages in mountainous regions and offers spatial insights for heritage conservation and regional planning.

Förderung

  • Beijing Forestry University

Zitieren

APA

Huang, X., Cheng, Y., Jiang, L., Li, Z., Deng, Y., & Bai, X. (2026). Spatial patterns and nonlinear drivers of traditional villages: a grid-based machine learning analysis in Yunnan, China. Frontiers in Environmental Science, 14. https://doi.org/10.3389/fenvs.2026.1834584

Vancouver

Huang X, Cheng Y, Jiang L, Li Z, Deng Y, Bai X. Spatial patterns and nonlinear drivers of traditional villages: a grid-based machine learning analysis in Yunnan, China. Frontiers in Environmental Science. 2026;14. doi:10.3389/fenvs.2026.1834584

BibTeX
@article{Huang2026,
  author = {Huang, Xinyu and Cheng, Yalan and Jiang, Linhan and Li, Zhuoran and Deng, Yue and Bai, Xueyue},
  title = {Spatial patterns and nonlinear drivers of traditional villages: a grid-based machine learning analysis in Yunnan, China},
  journal = {Frontiers in Environmental Science},
  year = {2026},
  volume = {14},
  doi = {10.3389/fenvs.2026.1834584}
}
Kostenlos starten →