DL-SOC

Deep learning code for digital soil organic carbon (SOC) mapping and covariate importance analysis

Overview

• convert_multiscale_features/ constains a script to generate multiscale covariate layers using Gaussian Pyramids
• dl_models/ contains scripts to build deep learning models for digital soil mapping as well as the weights of the trained models
• permutation_analysis/ contains scripts to perform local and continental scales feature permutation analysis
• shap/ contains a script to compute the contribution of covariates to SOC using SHapley Additive exPlanation (SHAP)
• To make the scripts work correctly, please create a data/ folder in the main/ directory and download the processed datasets from DL-SOC-data and paste all the subfolders into data/

Package Requirements

• python 3.8.2
• pandas 1.4.1
• shap 0.41.0
• scikit-learn 1.2.2
• numpy 1.20.1
• tensorflow 2.6.0
• evidential-deep-learning 0.4.0
• opencv-python 4.7.0.72

Citation

TBD