Non-linear drivers of urban-to-rural mobility in China’s Greater Bay Area: insights from mobile phone big data

Persistent urban–rural disparities hinder global sustainable development goals, while rural revitalization plays a pivotal role in advancing these objectives. High-frequency urban-to-rural mobility strengthens functional linkages between cities and rural areas by facilitating recurrent movements of people and their associated activities. However, existing research predominantly focuses on rural-to-urban migration, with less attention to high-frequency, recurrent and non-permanent urban-to-rural mobility. Moreover, current studies on reverse mobility mainly describe flow patterns rather than examining driving mechanisms. These knowledge gaps result in inefficient population attraction, resource misallocation, and policy misalignment. Here, using mobile signaling big data and the XGBoost model integrated with SHAP interpretability analysis, we identify the factors associated with urban-to-rural mobility in China’s Greater Bay Area, revealing their nonlinear responses and threshold effects. Results demonstrate that, beyond the common effects of logistics and accessibility, driving mechanisms vary significantly across rural area types: ecological-agricultural areas remain significantly influenced by traditional factors such as destination population size and road; the number of enterprises and GDP exert the strongest positive effect in industrial-productive areas; while the betweenness centrality of township nodes within the regional road network contributes significantly to population inflows in suburban-touristic areas. These insights advance theoretical understanding of urban-to-rural mobility drivers, calling for differentiated strategies to be adopted for distinct rural area types, effectively promoting urban-to-rural mobility and achieving rural revitalization.