Objective Aiming at the problems of separation and inadequate integration between natural and cultural heritage protection in current territorial spatial planning, this study aims to construct a regional heritage holistic protection spatial network that integrates natural and cultural values. Taking the six central districts of Tianjin as an example, this research explores a systematic approach that integrates local theories, historical documents, and multi-source spatial data to identify heritage elements, quantify their natural and cultural associations, and ultimately form a spatial structure that supports collaborative governance and planning optimization, providing a solution to break the protection dilemma of the natural-cultural dualistic division.

Methods Based on the “nature-culture” linkage mechanism reflected in traditional GUJI concepts and cultural landscape theory, this study integrates textual, visual, and geospatial multi-source data to construct a technical framework of “text mining−feature quantification−network building.” The research data includes text data, map data, and spatial data.The research methods include: 1) Natural base classification: Based on four types of indicators—terrain, hydrological conditions, climate features, and vegetation/soil—100  m  ×  100  m grid analysis was applied. GIS spatial analysis was employed to delineated five types of natural substrates. 2) Natural and cultural feature identification: Natural elements (rivers, lakes, wetlands, terraces) and cultural themes (transportation and shipping, commercial markets, agricultural production, religion and folklore, military defense) surrounding heritage sites were extracted. Keyword matching and inverse term frequency co-occurrence were used to compute element intensity. 3) Feature clustering: Principal component analysis (PCA) and Gaussian mixture model (GMM) clustering were applied to identify four types of natural features and seven types of cultural features. 4) Cultural influence calculation: Kernel density estimation (KDE) was used to transform discrete heritage points into continuous density surfaces, quantifying the spatial influence ranges of the five cultural themes. 5) Network construction: Based on feature clustering results, cultural landscape units were delineated, and the minimum spanning tree (MST) algorithm was employed to construct heritage linkage corridors, forming an integrated “point-line-area” holistic conservation network.

Results The study identifies five types of natural bases, with significant micro-scale differentiation between the units of each base. Additionally, there are distinct functional differences in the heritage types supported by different natural bases. A macro understanding of the geographical features and hydrological conditions in the six districts of Tianjin, along with the spatial distribution of heritage sites, helps in developing preventive strategies based on the natural geographical features of each region in the protection planning. Through text mining, four natural elements and five cultural themes are identified. Clustering results in seven cultural features and four natural features. The cultural features show significant spatial differentiation: The transportation and shipping culture is distributed in a belt along the Haihe River, with the core concentrated around the Sancha River confluence; the commercial market culture exhibits a dual-core pattern (traditional center of the old city and the modern commercial belt of the foreign concessions); agricultural production culture is distributed as points on the periphery; religious and folklore culture is scattered along the old city streets; military defense culture is concentrated in city walls, garrisons, and river defense points. Among the natural elements, rivers are distributed in bands, wetlands are concentrated in waterfront areas, and terraces are distributed in the concentric areas and historical construction nodes such as ferry terminals. Ultimately, an integrated protection network was constructed, including nine typical cultural landscape units and 311 spatial association paths. The network is centered around the Haihe River confluence, radiating outward along the water system, presenting a “core concentration, peripheral diffusion” pattern. The association corridors are densely distributed from the Haihe River confluence to the core area of Heping District, extending radially along historical streets and water systems. This distribution reflects the spatial logic of religious folklore and commercial market cultures following the street axis, and the interaction between military defense and transportation cultures relying on waterways.

Conclusion This study constructs an integrated conservation spatial network for natural and cultural heritage from a holistic perspective of nature-culture interactive symbiosis. Theoretically, it integrates the traditional Chinese concept of “Guji” with cultural landscape theory, proposing an analytical framework of “spatial substrate–textual narrative–social identity” to provide indigenous theoretical support for the cognition of “correlation” and “integrity” in natural and cultural heritage conservation. Methodologically, it establishes a technical route of “text mining–cluster analysis–geographic modeling–network analysis”, whose core lies in integrating multi-source heterogeneous historical materials (local gazetteers, historical maps) with contemporary geospatial data, revealing the spatial correlation and structural characteristics of “nature-culture” elements through quantitative models. The research findings supplement potential natural and cultural heritage and their interrelations, delineating the boundaries of cultural landscape units. The network clarifies the element composition and spatial scope of holistic conservation, providing a basis for constructing integrated conservation systems in current planning; the relational diagram among heritage points can serve as a “relational base map” for cross-departmental collaboration, promoting coordinated governance of heritage.