Scripts underlying the publication: Predicting subsurface classification in 2D from cone penetration test data

This dataset contains the script underlying the manuscript titled 'Predicting subsurface classification in 2D from cone penetration test data'. The script can be used for predicting the most likely 1D classification of the subsurface at CPT locations. The uncertainties in the CPT measurements and transformation to soil units are accounted for by integrating the Robertson chart with the Bayesian approach. The classification is carried out in two steps: (a) for a given number of layers, identifying the most probable soil layer thicknesses, and (b) identifying the most likely number of layers which, together with the most probable thicknesses, maximises the likelihood of observing the CPT measurements. The script can be used with any CPT data in .gef format.

本数据集包含题为《基于圆锥贯入试验数据开展二维地下空间分类预测》的手稿所依托的脚本程序。该脚本可用于预测圆锥贯入试验（Cone Penetration Test, CPT）测点处地下空间最可能的一维分类结果。通过将罗伯逊图表（Robertson Chart）与贝叶斯方法（Bayesian Approach）相结合，可对圆锥贯入试验测量误差及土层转换过程中的不确定性进行量化与纳入考量。该分类流程分为两个阶段：(a) 给定土层层数时，确定最具可能性的土层厚度；(b) 确定最合理的土层总层数，结合前述最优厚度值，使观测到圆锥贯入试验测量数据的似然值最大化。该脚本可兼容所有.gef格式的圆锥贯入试验数据。