Understanding the village-scale expansion of rural settlements in China from a topographic perspective

Abstract

Topography critically shapes the distribution of Rural Settlements (RS). However, previous studies have often neglected the systematic role of topographic gradients, typically focusing on macro scales, which obscures the nuanced patterns and underlying mechanisms at the village level. To address this, we developed a two-dimensional elevation-slope framework to reconstruct the 40-year evolution of China's RS at the administrative village scale. We then quantified its morphological changes at the village level and employed a Geographically Weighted Machine Learning (GWML) framework, which integrates geographically weighted principles with machine learning capabilities to capture the spatial heterogeneity and non-linear effects of the driving factors. Our findings reveal a highly uneven RS distribution. By 2020, 78.49% of the settlement area was concentrated in Low elevation-Low slope (L-L) regions, comprising just 21.74% of China's landmass. Over the past four decades, expansion has trended towards higher elevations and steeper slopes, though patterns and land sources varied significantly by terrain. Plains expansion was dominated by edge-expansion onto Cultivated Land, whereas in topographically complex regions, it was more dispersed with diverse sources. Furthermore, settlement density in L-L villages was over a hundredfold greater than in High elevation-High slope (HH) villages. The optimal Geographically Weighted Random Forest (GWRF) model shows that expansion in plains is driven by land use intensity and village scale, while in complex terrains, it is governed by ecological constraints or economic density. This study systematically dissects the dynamic patterns and morphological differentiation of rural settlements under topographic constraints, offering scientific insights for rural revitalisation and regional planning.