Abstract: Objectives: Realistic representation of tree bark textures is essential for achieving high-fidelity 3D tree modeling in digital twin cities. However, the surface textures of tree bark exhibit significant variability due to species diversity and environmental influences. Existing methods, particularly those based on image acquisition or fixed templates, often suffer from limitations such as low resolution, insufficient structural detail, and poor adaptability to different tree types. Methods: A hierarchical modeling framework based on “Feature–Attribute–Parameter” is constructed to analyze and represent the structural composition of tree bark textures. Seven typical texture features are defined—smooth, lenticels, cracks, furrows, ridges, scales, and strips—and are further characterized using five attributes: shape, number, position, depth, and color. A node-based procedural generation workflow is designed, in which each attribute is parameterized and mapped to dedicated procedural nodes, enabling the flexible combination of texture features. The framework supports fine-grained control of appearance through adjustable parameters. Results: The procedural generation method was applied to five representative tree species, including oak, litchi, camphor, coconut, and kapok. The generated textures closely resembled real-world bark photographs in terms of structure, visual depth, and color distribution. Compared with oblique photogrammetry and generative AI, the proposed method produced textures with richer detail, greater visual clarity across scales, and consistent appearance across occluded or inaccessible regions. Conclusions: The procedural approach offers high realism, strong controllability, and broad adaptability for various tree species. It addresses key limitations of existing methods, particularly in resolution independence and feature completeness. This method offers a scalable, parameter-driven solution for tree bark texture generation, and can be integrated into digital twin platforms to support realistic, editable, and high-quality 3D vegetation modeling.