Search

[Display omitted] • A new means for multi-source spectral fusion based on deep learning was proposed. • Soil quality index developed from 21 physical, chemical, environmental properties. • Continuous wavelet transform highlighted the spectra related to soil quality index. • The proposed method was superior to the commonly used data fusion methods. Soil quality assessment, as a means to assess the impact of human activities on soil, is of great significance to achieve sustainable development. Proximal sensing offers a rapid and cost-effective means for soil spectral acquisition, which resolves the shortcomings of traditional laboratory analysis and provides a new way for soil quality assessment. However, using a singular spectroscopic technique may not encompass adequate information needed for soil quality assessment. Therefore, in this research, the visible-near-infrared (vis-NIR) and portable X-ray fluorescence (pXRF) spectral data were fused, and the continuous wavelet transform (CWT) was combined with the parallel input 2D convolutional neural network (PI-2D-CNN) algorithm, to enhance the prediction accuracy of soil quality index (SQI). A total of 287 soil samples were collected from the Yarkand-Kashgar River Basin in Xinjiang, China for proximal sensing in the laboratory, and 21 soil physicochemical properties were measured for SQI calculation simultaneously. After preprocessing the vis-NIR and pXRF spectral data with the CWT, partial least squares regression (PLSR) and CNN models were constructed to predict SQI. The results showed that the SQI prediction accuracy based on vis-NIR spectra (coefficient of determination for validation set (R2) = 0.32–0.54) was higher than that based on pXRF spectra (R2 = 0.17–0.36). The prediction accuracy of both vis-NIR and pXRF spectra were improved by CWT combined with CNN. Compared with the traditional spectral data fusion methods (concatenation, sequential and orthogonalised-partial least squares (SO-PLS)), the PI-CNN algorithm demonstrated superior SQI prediction accuracy, and the constructed PI-2D-CNN model obtained the highest prediction accuracy (R2 = 0.67). Overall, this study proved that the combination of CWT and PI-CNN could increase the SQI prediction accuracy. This research will provide a rapid and low-cost approach for large-scale soil quality assessment. [ABSTRACT FROM AUTHOR]

• Soil TN estimation accuracy using vis-NIR spectra is higher than pXRF spectra. • Feature selection & geographical stratification improve soil TN estimation accuracy. • CARS is the optimal feature selection method for multi-sensor data fusion models. • SO-PLS fusion method increases model accuracy by fully using multiple sensor data. Fast and accurate monitoring of soil total nitrogen (TN) content is particularly important to optimize agricultural inputs (e.g. fertilizers) and inhibit N loss-induced pollution. Proximal soil sensing combined with multi-sensor fusion has been considered to be a promising alternative to traditional laboratory analysis because it can achieve fast, non-destructive, environmentally friendly, and low-cost monitoring. However, the accuracy of this technique depends on the heterogeneity of the dataset and the data fusion strategy. In this study, a total of 500 soil samples were collected from two locations with high degree of soil and environment heterogeneity in Xinjiang, China, and then visible-near-infrared spectroscopy (vis-NIR), portable X-ray fluorescence (pXRF) spectroscopy and soil TN measurement were conducted in the laboratory. Based on partial least squares regression algorithm, direct concatenation, outer-product matrix analysis, and sequentially orthogonalized partial least-square (SO-PLS) were applied for multi-sensor data fusion by using full spectra or spectral features. The results showed that the estimation accuracy using vis-NIR spectral data were higher than pXRF spectral data. Compared with single sensor data and full-spectrum data fusion, the feature selection combined with data fusion contributed to a higher soil TN estimation accuracy, and the competitive adaptive reweighted sampling combined with SO-PLS fusion and geographical stratification modeling strategy had the highest soil TN estimation accuracy, with a root mean square error (RMSE) of 0.1838 g kg−1, and a Lin's concordance correlation coefficient of 0.86. It was worth noting that geographical stratification was an effective modeling strategy to improve the TN estimation accuracy based on multi-sensor data fusion, and its RMSE was 0.10 % ∼ 11.70 % lower than that of global modeling. This study highlights the potential of feature selection combined with geographic stratification to increase the soil TN estimation accuracy based on multi-sensor fusion, especially in regions with high soil and environmental heterogeneity. [ABSTRACT FROM AUTHOR]

[Display omitted] • Vis-NIR and pXRF spectral data fusion enhances soil organic matter estimation. • 1.6th-order derivative is the optimal preprocessing method for vis-NIR spectra. • Baseline correction is the optimal preprocessing method for pXRF spectra. • Feature selection lifts the single/multi sensor data fusion estimation accuracy. • CARS combined with GRA fusion provides the optimal estimation of SOM. Rapid and accurate estimation of soil organic matter (SOM) content is of great significance for agricultural production and carbon stock estimation. Visible-near-infrared spectroscopy (Vis-NIR) and portable X-ray fluorescence spectroscopy (pXRF), with the fast, low-cost, and environmental friendly characteristics, can realize the rapid large-scale detection of SOM. However, due to the heterogeneity and complexity of soil, estimation of SOM using a single sensor often leads to high uncertainty in result. Multi-sensor data fusion is one of the approaches to solve this problem, but suitable spectral preprocessing methods and data fusion methods still need further research. In this study, 500 soil samples were collected from two watersheds with large differences in soil parent material, soil type, and climate in the Xinjiang Uygur Autonomous Region of China. The effects of different spectral preprocessing methods for vis-NIR and pXRF spectra on SOM estimation were compared, and the effects of five feature selection methods on single sensor estimation and three different multi-sensor fusion estimations were evaluated. The results showed that vis-NIR spectrum was better than pXRF spectrum for SOM estimation, and the 1.6th-order derivative and baseline correction were the optimal preprocessing method for vis-NIR and pXRF spectra (Beam 3), respectively. Feature selection could improve the accuracy of the single sensor and multi-sensor fusion estimations. The competitive adaptive reweighted sampling (CARS) method was the optimal feature selection method for single sensor estimation and Granger–Ramanathan averaging (GRA) fusion estimation, and the variable importance in projection (VIP) was the optimal feature selection method for direct concatenation and outer-product analysis fusion estimation. In general, the optimal SOM estimation (RMSE p = 3.59 g kg−1, LCCC p = 0.84, RPIQ p = 2.67) could be achieved based on CARS feature selection method and GRA fusion. This study proves that the combination of spectral preprocessing and feature selection can effectively improve the accuracy of the single sensor and multi-sensor fusion estimations. Especially, the multi-sensor fusion estimation has higher accuracy than the single sensor estimation. This study will provide a reference for the application of proximal soil sensing. [ABSTRACT FROM AUTHOR]