Dataset for Comprehensive Analysis of Desiccation Cracks in Soils

Soil cracking poses significant challenges to the integrity of slopes and earthen infrastructure. This study presents a Dataset for Comprehensive Analysis of Desiccation Cracks in Soils (D-CRACKS), aiming to enhance the understanding of this complex phenomenon. The dataset compiles 1,000 images from laboratory tests of soil desiccation cracking collected from 45 studies in the literature. D-CRACKS is built using a Structured Query Language (SQL)-based database management system, enabling efficient data retrieval, filtering, and querying. Each image is linked to detailed metadata, including testing conditions, soil properties, boundary constraints of the tested samples, environmental conditions, and admixture properties. Following data cleaning, images were analyzed to extract key crack properties such as crack area, crack ratio, crack length, and average crack width. Statistical analyses were performed to examine variability and trends across soil types and testing conditions. D-CRACKS offers a structured and scalable foundation for investigating soil cracking and offers substantial value for future research, particularly for developing data-driven predictive models and validating physics-based numerical simulations of soil cracking.

土体开裂对边坡与土工构筑物的完整性构成严峻挑战。本研究构建了土壤干燥开裂综合分析数据集（D-CRACKS），旨在加深对这一复杂现象的认知。该数据集整合了来自45项已发表文献中实验室土体干燥开裂试验的1000幅图像。D-CRACKS采用基于结构化查询语言（Structured Query Language, SQL）的数据库管理系统搭建，可实现高效的数据检索、筛选与查询操作。每幅图像均关联详细的元数据，涵盖试验条件、土体特性、试样边界约束、环境条件及外加剂特性。完成数据清洗后，研究人员对图像开展分析，提取开裂面积、开裂率、开裂长度与平均开裂宽度等关键开裂特征参数。随后通过统计分析，探究不同土体类型与试验条件下的参数变异性及变化趋势。D-CRACKS为土体开裂研究提供了结构化且可扩展的研究基础，对未来相关研究具有重要价值，尤其可为开发数据驱动的预测模型以及验证土体开裂的物理驱动数值模拟提供有力支撑。