Abstract:
Objective Traditional villages in the Li River Basin are key cultural heritage assets, representing the historical evolution of local societies, architectural traditions, and the complex interactions between human activities and natural landscapes. These villages, which exhibit distinct spatial patterns and cultural expressions, are central to understanding rural cultural landscapes in China. However, rapid urbanization has intensified pressures on these villages, resulting in challenges such as spatial homogenization, functional decline, and disordered development, all of which contribute to the erosion of the unique spatial characteristics of such villages. To address these critical issues, this research aims to identify and analyze the spatial genes of traditional villages in the Li River Basin. By constructing a comprehensive framework that integrates production, living, and ecological spaces, the research seeks to provide a theoretical foundation for the conservation, adaptive reuse, and sustainable management of these traditional landscapes within the broader framework of national and regional territorial spatial planning.
Methods The research employs an innovative mixed-methods approach that combines deep learning techniques, spatial feature quantification, and extensive field investigations. A comprehensive spatial information database is developed, drawing information from multiple sources such as remote sensing imagery, historical documents, architectural surveys, and field observations. This database serves as the foundation for spatial analysis, facilitating the identification of spatial genes using a deep learning model, the Dual Attention Network (DANet). The model is designed to enhance spatial feature extraction through a combination of spatial and channel attention mechanisms, with a view to enabling more precise identification of spatial characteristics. The research categorizes spatial genes into three main types: Production space, living space, and ecological space. Production space mainly involves the forms of farmland and the relationship between village and farmland, reflecting the close integration of village with agricultural land. Living space encompasses settlement scale, settlement morphology, and road network patterns. Ecological space includes terrain type, landform type, village – mountain relationship, and village – water relationship, emphasizing the spatial relationship between village location and natural elements like mountains and water. The research also constructs a detailed spatial gene map that visually represents the organization and integration of production, living, and ecological spaces within traditional villages, providing insights into the dynamic interactions among these elements.
Results The research identifies 32 distinct spatial gene types across traditional villages in the Li River Basin, which are systematically categorized into four main village types based on their spatial gene combinations and adaptation strategies. The first type, Surrounding Solitary Peak Village, is typically located in low-elevation karst landscapes, characterized by a circular spatial layout around isolated peaks and close proximity to water sources. These villages exhibit a strategic use of natural terrain, enabling effective agricultural production and compact living environments that are adapted to local topography. The second type, Peak Forest Depression Village, is found in hilly and plain karst areas. These villages feature compact living spaces but more dispersed production areas, reflecting adaptations to limited ecological space and a strong reliance on local natural resources. The third type, River Valley Plain Village, is situated along river valleys and plains, with residential and production spaces closely aligned with water channels. These villages often serve as commercial centers with well-developed waterway networks, demonstrating a functional balance between agricultural production and trade activities. The fourth type, Mountainous Hill Village, is located at higher elevations and characterized by stepped terraces, tiered housing, and adaptive land use patterns that integrate with the steep terrain. These villages employ innovative land management strategies, such as contour farming and hillside construction, to maximize the use of available land while maintaining ecological balance. The spatial gene map constructed in this research offers a comprehensive visualization of the spatial organization and integration patterns within these villages, highlighting how different types of traditional villages maintain a balance among production, living, and ecological spaces. The map also reveals varying adaptation strategies shaped by local geographic, cultural, and economic factors, providing valuable insights into the sustainability of traditional village landscapes.
Conclusion This research makes significant contributions to traditional village research by introducing an innovative framework for spatial gene identification and analysis, specifically focusing on the Li River Basin. Through the systematic classification of spatial genes into production, living, and ecological spaces, the research not only refines the theoretical understanding of traditional village landscapes, but also provides practical guidance for their conservation and revitalization. The use of the DANet deep learning model enhances the objectivity, precision, and efficiency of spatial gene identification and mapping, thereby establishing a comprehensive spatial gene database. The spatial gene map constructed in this research effectively visualizes the interactions and combination patterns of spatial elements, offering clear insights into the strategies for classification and sustainable development of villages. The findings underscore the need to harmonize traditional spatial characteristics with modern development requirements, advocating for a balanced integration of historical heritage within the context of contemporary societal demands. By supporting policy formulation aligned with territorial spatial governance and regional planning, this research contributes to the sustainable development of traditional villages, as well as the preservation of cultural identity and the accommodation of modern needs.