Deep Learning Application for Predicting Soil Organic Matter Content by VIS-NIR Spectroscopy

Abstract

Deep learning is characterized by its strong ability of data feature extraction. This method can provide unique advantages when applying it to visible and near-infrared spectroscopy for predicting soil organic matter (SOM) content in those cases where the SOM content is negatively correlated with the spectral reflectance of soil. This study relied on the SOM content data of 248 red soil samples and their spectral reflectance data of 400–2450 nm in Fengxin County, Jiangxi Province (China) to meet three objectives. First, a multilayer perceptron and two convolutional neural networks (LeNet5 and DenseNet10) were used to predict the SOM content based on spectral variation and variable selection, and the outcomes were compared with that from the traditional back-propagation neural network (BPN). Second, the four methods were applied to full-spectrum modeling to test the difference to selected feature variables. Finally, the potential of direct modeling was evaluated using spectral reflectance data without any spectral variation. The results of prediction accuracy showed that deep learning performed better at predicting the SOM content than did the traditional BPN. Based on full-spectrum data, deep learning was able to obtain more feature information, thus achieving better and more stable results (i.e., similar average accuracy and far lower standard deviation) than those obtained through variable selection. DenseNet achieved the best prediction result, with a coefficient of determination (R2) = 0.892 ± 0.004 and a ratio of performance to deviation (RPD) = 3.053 ± 0.056 in validation. Based on DenseNet, the application of spectral reflectance data (without spectral variation) produced robust results for application-level purposes (validation R2 = 0.853 ± 0.007 and validation RPD = 2.639 ± 0.056). In conclusion, deep learning provides an effective approach to predict the SOM content by visible and near-infrared spectroscopy and DenseNet is a promising method for reducing the amount of data preprocessing.

Application and interpretation of deep learning methods for the geographical origin identification of Radix Glycyrrhizae using hyperspectral imaging

“…39 It is used by some scholars in the eld of spectroscopy. 40,41 In the eld of spectroscopy, the one-dimensional (1D) CNN model is used to learn and predict spectra, and has achieved good performances. 1D CNN is a non-linear model; however, given a spectrum (SPEC) of category c, it can be expanded by calculating the rst-order Taylor approximation to approximate the score value S c (SPEC) with a linear function.…”

Predicting and Mapping of Soil Organic Carbon Using Machine Learning Algorithms in Northern Iran

“…The hierarchical structure and high learning capacity make DNN models quite flexible and adaptable for a wide variety of highly complex problems such as SOC prediction [11,39,40]. The DNNs have recently been used for the prediction of soil properties [40][41][42] and particularly for SOC prediction [43,44]. Xu et al [43], for instance, indicated that the DNN method had a high performance for the prediction of SOC with the effective abstraction of complex covariates for learning by using visible and near-infrared soil spectra.…”

Prediction of the Soil Organic Matter (SOM) Content from Moist Soil Using Synchronous Two-Dimensional Correlation Spectroscopy (2D-COS) Analysis

“…Thus, the technology for the rapid detection of SOM content in soil has been a hot topic in the soil research community within the past couple decades [ 2 , 3 , 4 , 5 , 6 , 7 , 8 , 9 , 10 , 11 , 12 , 13 , 14 ]. Visible-near infrared (Vis-NIR) spectroscopy has been a simple and efficient analytical technique which has been widely deployed for the detection of SOM content [ 15 , 16 , 17 , 18 , 19 , 20 , 21 , 22 , 23 ]. The high sensitivity of vis-NIR spectroscopy has been shown to be capable to detect trace minerals such as arsenic contamination in soil [ 24 ].…”

“…Amongst these studies [ 27 , 28 , 29 , 30 , 31 , 32 , 40 ] the best coefficient of determination ( R 2 ) and the ratio of prediction to deviation (RPD) for assessing the SOM in moist soil (<35% w / w moisture) were about 0.86–0.88 [ 27 ] and 2.66 [ 29 ] respectively. Note that the assessments of SOM from relatively dry soil have achieved much better result with R 2 and RPD of 0.892 ± 0.004 and 3.053 ± 0.056 respectively [ 15 ].…”