Soil Microplastics Spectrum Based on Visible Near-Infrared Spectroscopy
DOI:
https://doi.org/10.3329/bjb.v51i40.63841Keywords:
Soil, Microplastics, Visible near-infrared spectroscopy, MachinelearningAbstract
As the largest "repository" of microplastics, soil is affected by soil structure, microplastics category and particle size. In this study, the method of combining indoor simulation modeling and field verification is proposed. The reflectance of soil microplastic samples was collected by ASD FieldSpec4 Hi-Res spectrometer, NOR, MSC, SNV were used for spectral pretreatment, and differential transformations of different orders are used to enhance the spectral signal-to-noise ratio. The results showed that the spectral reflectance of microplastics decreased with the increase of microplastics content in soil. After FD and SD transform, the spectral features are enhanced obviously. The regression model based on NOR transformation of reflectivity combined with first deviation is the best, Rc2, Rv2, RMSEC and RMSEPare 0.75, 0.77, 0.16, 0.12,respectively.This study can provide a scientific basis for quantitative research on microplastics in farmland soil in northern Shaanxi, China.
Bangladesh J. Bot. 51(4): 971-977, 2022 (December) Special
Downloads
54
54
Downloads
Published
How to Cite
Issue
Section
License
© Bangladesh Botanical Society
Authors are required to transfer the copyright of their articles to the journal. The Declaration form is available here http://www.bdbotsociety.org/journal/journal_pdf/declaration_form.pdf