Combination of Convolutional Neural Networks and Recurrent Neural Networks for predicting soil properties using Vis–NIR spectroscopy

• Published in: Geoderma Vol. 380; p. 114616
• Main Authors: Yang, Jiechao, Wang, Xuelei, Wang, Ruihua, Wang, Huanjie
• Format: Journal Article
• Language: English
• Published: Elsevier B.V 15.12.2020
• Subjects: calibration; Convolutional Neural Network; land cover; land use; mineral soils; model validation; neural networks; nondestructive methods; prediction; Recurrent Neural Network; reflectance spectroscopy; Soil properties estimation; soil quality; surveys; Transfer learning; Vis–NIR spectroscopy; Convolutional Neural Network; Soil properties estimation; Vis–NIR spectroscopy; Recurrent Neural Network; Transfer learning

•Combining the advantages of CNN and RNN for Vis-NIR spectral analysis.•Developing a joined CNN and RNN model called CCNVR for predicting soil properties.•CCNVR exhibits great superiority in the prediction accuracy and robustness to noise.•CCNVR shows good transferability across different soil types...