A modeling method based on discrete wavelet transform (DWT) was introduced to analyze the concentration of chromium, copper, zinc, arsenic and lead in soil with a portable X-ray fluorescence (XRF) spectrometer. A total of 111 soil samples were collected and observed. Denoising and baseline correction were performed on each spectrum before modeling. The optimum conditions for pre-processing were denoising with Coiflet 3 on the 3rd level and baseline correction with Coiflet 3 on the 9th level. Calibration curves were established for the five heavy metals (HMs). The detection limits were compared before and after the application of DWT, the qualitative detection limits and the quantitative detection limits were calculated to be three and ten times as high as the standard deviation with silicon dioxide (blank), respectively. The results showed that the detection limits of the instrument using DWT were lower, and that they were below national soil standards; the determination coefficients (R²) based on DWT-processed spectra were higher, and ranged from 0.990 to 0.996, indicating a high degree of linearity between the contents of the HMs in soil and the XRF spectral characteristic peak intensity with the instrument measurement.

The accuracy, repeatability and detection limits of the energy-dispersive X-ray fluorescence (XRF) spectrometer used in this study were tested to verify its suitability for rapid screening of cadmium in samples. Concentrations of cadmium in rice grain samples were tested by the XRF spectrometer. The results showed that the apparatus had good precision around the national limit value (0.2 mg/kg). Raman spectroscopy has been analyzed in the discrimination of rice grain samples from different geographical origins within China. Scanning time has been discussed in order to obtain better Raman features of rice samples. A total of 31 rice samples were analyzed. After spectral data pre-treatment, principal component analysis (PCA), K-means clustering (KMC), hierarchical clustering (HC) and support vector machine (SVM) were performed to discriminate origins of rice samples. The results showed that the geographical origins of rice could be classified using Raman spectroscopy combined with multivariate analysis.

X-ray fluorescence (XRF) technology is a widely used method forrapid detection of heavy metals in soil.