Neural Network for Predicting Soil Thermal Conductivity

The uploaded files are the referenced dataset for the article "Neural network prediction of thermal conductivity across full saturation for various soil types" submitted to Agricultural Forest and Meteorology. It comprises two folders: 1. “Model prediction” Folder: This folder contains the trained neural network (py and pth files). It enables users to reproduce the results presented in Figure 8 (training and testing outcomes) and facilitates the application of the model to additional soil datasets. 2. “Parameters map” Folder: This folder demonstrates an example application of the trained model, where the authors used 14064 SSCBD data at surfacical horizons from WoSIS Soil Profile Database to generate and visualize the global distribution of the MLD model parameters (Sf, p, λdry and λsat).上传的文件为投稿至《Agricultural and Forest Meteorology》（农业与森林气象学）的论文《不同土壤类型全饱和状态下热导率的神经网络预测》所引用的数据集。该数据集包含两个文件夹： 1. "模型预测（Model prediction）"文件夹：该文件夹内存储有训练完成的神经网络（neural network）模型文件（格式为py和pth）。使用者可通过该文件夹复现论文中图8展示的训练与测试结果，同时可将该模型应用于其他土壤数据集。 2. "参数分布图（Parameters map）"文件夹：该文件夹展示了训练完成模型的示例应用：研究人员使用来自WoSIS土壤剖面数据库（WoSIS Soil Profile Database）的14064份表层土层SSCBD数据，生成并可视化了MLD模型参数（Sf、p、λdry及λsat）的全球分布情况。