A Transfer Learning Approach Utilizing Combined Artificial Samples for Improved Robustness of Model to Estimate Heavy Metal Contamination in Soil

Abstract

Benefiting from the nanoscale sampling intervals and subtle spectral information in the visible and near-infrared band, hyperspectral technology is considered as an efficient means for monitoring soil heavy metal contamination whereby the good robustness of prediction model is driven by the increase to spectral dimension in model analysis. Considering the positive correlation between samples size and spectral dimension, we focuses on a novel derivation of enlarging samples size in this study to improve model performance by i) preparing artificial samples taking into account of flexibility and control over the laboratory environment compared with collecting wild samples, and ii) using transfer learning method called transfer component analysis (TCA) for reducing spectral feature differences caused by soil heterogeneity to train model in the same data distribution. The proposed approach was tested on three heavy metals, namely copper (Cu), cadmium (Cd) and lead (Pb), collected in the mining area located in the Xiangjiang Basin, Hunan Province, China. The experiments showed that the initial model constructed by a small number of wild samples performed strong prediction sensitivity as the training samples change. In contrast, a modified model with TCA could showed good robustness with excellent predicted ability, the average prediction accuracy of the determinable coefficient (R 2) and the ratio of prediction to deviation (RPD) improved to 0.73 and 1.90, 0.74 and 1.92, 0.72 and 1.73, respectively. The results illustrated there was a more reliable modeling method in potential to predict soil heavy metals based on hyperspectral analysis at low cost.

Time-Domain Transfer Learning for Accurate Heavy Metal Concentration Retrieval Using Remote Sensing and TrAdaBoost Algorithm: A Case Study of Daxigou, China

“…developed using samples from one region to other regions. Wang et al [20] adopted the transfer component analysis method to analyse the distribution differences between artificial and natural samples to improve the regression accuracy under limited heavy metal samples data using an expanded training dataset.…”

“…For example, Tao et al [19] used the transfer component analysis method to reduce the differences in the probability distribution of soil samples from two or more regions, enabling the application of a model developed using samples from one region to other regions. Wang et al [20] adopted the transfer component analysis method to analyse the distribution differences between artificial and natural samples to improve the regression accuracy under limited heavy metal samples data using an expanded training dataset.…”

An Improved Vis-NIR Estimation Model of Soil Organic Matter Through the Artificial Samples Enhanced Calibration Set

“…Generally, the coefficient of determination (R 2 ) and ratio of performance to interquartile distance (RPIQ) are the key parameters for evaluating model accuracy and stability [26], [45], [46]. R 2 indicates the level to which the target variables are fully explained by the predictor variables.…”

“…These samples were primarily used to determine the spectral absorption features of the soil components because the near-natural soil samples were not affected by the heterogeneous constituents of the soils. Furthermore, Wang et al [26] attempted to enlarge the field sample size with samples prepared under controlled laboratory conditions. Still, a model built with mixed samples did not show satisfactory performance and was even lower than the performance of the initial model, suggesting that the spectral difference between field-obtained spectra of natural soil samples and near-natural soil sample spectra may damage the generalization and robustness of the model.…”

“…The near-natural samples method is regarded as an easy and effective experiment for supporting quantitative spectroscopic analysis of soil attributes [18], [23]. Farifteh et al [24], Linsinger et al [25], Zou et al [20], and Wang et al [26] created near-natural samples to characterize the spectral signals of the key attributes in soils, and developed a spectroscopic technique to recognize the content status of soil chemical components. The near-natural soil samples with various known SOM contents created under controlled laboratory conditions exhibit the standard behaviors of SOM absorption features that are in agreement with the natural soils, and therefore can be help to compensate for sampling pattern shortages under field conditions.…”