Identifying cadmium and lead co-accumulation from living rice blade spectrum

Rapid assessment of heavy metal accumulation capability of Sedum alfredii using hyperspectral imaging and deep learning

Hyperaccumulators are the material basis and key to the phytoremediation of heavy metal contaminated soils. Conventional methods for screening hyperaccumulators are highly dependent on the time- and labor-consuming sampling and chemical analysis. In this study, a novel spectral approach assisted with multi-task deep learning was proposed to streamline accumulating ecotype screening, heavy metal stress discrimination, and heavy metals quantification in plants. The significant Cd/Zn co-hyperaccumulator Sedum alfredii and its non-accumulating ecotype were stressed by Cd, Zn, and Pb. Spectral images of leaves were rapidly acquired by hyperspectral imaging. The self-designed deep learning architecture was composed of a shallow network (ENet) for accumulating ecotype identification, and a multi-task network (HMNet) for heavy metal stress type and accumulation prediction simultaneously. To further assess the robustness of the networks, they were compared with conventional machine learning models (i.e., partial least squares (PLS) and support vector machine (SVM)) on a series of evaluation metrics of classification, multi-label classification, and regression. S. alfredii with heavy metals accumulation capability was identified by ENet with 100 % accuracy. HMNet reduced overfitting and outperformed machine learning models with the average exact match ratio (EMR) of heavy metal stress discrimination increased by 7.46 %, and residual prediction deviations (RPD) of heavy metal concentrations prediction increased by 53.59 %. The method succeeded in rapidly and accurately discriminating heavy metal stress with EMRs over 91 % and accuracies over 96 %, and in predicting heavy metals accumulation with an average RPD of 3.29 for Zn, 2.57 for Cd, and 2.53 for Pb, indicating the satisfactory practicability and potential for sensing heavy metals accumulation. This study provides a relatively novel spectral method to facilitate hyperaccumulator screening and heavy metals accumulation prediction in the phytoremediation process.

Improving the estimation accuracy of rapeseed leaf photosynthetic characteristics under salinity stress using continuous wavelet transform and successive projections algorithm

Soil salinization greatly restricts crop production in arid areas for salinity stress can inhibit crop photosynthesis and growth. Chlorophyll fluorescence and photosynthetic gas exchange (CFPGE) parameters are important indicators of crop photosynthesis and have been widely used to evaluate the impacts of salinity stress on crop photosynthesis and growth. Remote sensing technology can quickly and non-destructively obtain crop information under salinity stress, however, at present, the distribution of spectral features of CFPGE parameters in different regions is still unclear. In this study (2019-2020), under salinity stress conditions, the spectral data of rapeseed leaves were acquired and the CFPGE parameters were simultaneously determined. Then, continuous wavelet transformation (CWT) and standard normal variate (SNV) transformation were utilized to preprocess the raw spectral data. After that, a CFPGE parameter estimation model was constructed by using the partial least squares regression (PLSR) algorithm and the support vector machines (SVM) algorithm based on the spectral features in the red region (600-800 nm) and those in the red, blue-green (350-600 nm), and near-infrared (800-2500 nm) regions. The results showed that the spectral features of CFPGE parameters could be extracted by successive projections algorithm (SPA) based on the CWT preprocessing. The CFPGE parameter estimation model constructed based on the spectral features in the red region (675 nm, 680 nm, 688 nm, 749 nm, and 782 nm) had the highest Fv/Fm estimation accuracy on day 30, with R2c, R2p, and RPD of 0.723, 0.585, and 1.68, respectively. Based on this, the spectral features (578 nm, 976 nm, 1088 nm, 1476 nm, and 2250 nm) in the blue-green and near-infrared regions were added in the variables for modeling, which significantly improved the accuracy and stability of the model, with R2c, R2p, and RPD of 0.886, 0.815, and 2.58, respectively. Therefore, the fusion of the spectral features in the red, blue-green, and near-infrared regions could improve the estimation accuracy of rapeseed leaf CFPGE parameters. This study will provide technical reference for rapid estimation of photosynthetic performance of crops under salinity stress in arid and semi-arid areas.