Comparison of learning models to predict LDPE, PET, and ABS concentrations in beach sediment based on spectral reflectance

Faisal Raiyan Huda; Florina Stephanie Richard; Ishraq Rahman; Saeid Moradi; Clarence Tay Yuen Hua; Christabel Anfield Sim Wanwen; Ting Lik Fong; Aazani Mujahid; Moritz Müller

doi:10.1038/s41598-023-33207-x

Comparison of learning models to predict LDPE, PET, and ABS concentrations in beach sediment based on spectral reflectance

Sci Rep. 2023 Apr 17;13(1):6258. doi: 10.1038/s41598-023-33207-x.

Authors

Faisal Raiyan Huda^{1

2}, Florina Stephanie Richard¹, Ishraq Rahman¹, Saeid Moradi², Clarence Tay Yuen Hua¹, Christabel Anfield Sim Wanwen¹, Ting Lik Fong¹, Aazani Mujahid³, Moritz Müller⁴

Affiliations

¹ Faculty of Engineering, Computing and Science, Swinburne University of Technology, Sarawak Campus, 93350, Kuching, Sarawak, Malaysia.
² Faculty of Science, Thompson Rivers University, 805 TRU Way, Kamloops, BC, V2C0C8, Canada.
³ Institute of Sustainable and Renewable Energy (ISuRE), Universiti Malaysia Sarawak, Kota Samarahan, 94300, Sarawak, Malaysia.
⁴ Faculty of Engineering, Computing and Science, Swinburne University of Technology, Sarawak Campus, 93350, Kuching, Sarawak, Malaysia. mmueller@swinburne.edu.my.

Abstract

Microplastic (MP) contamination on land has been estimated to be 32 times higher than in the oceans, and yet there is a distinct lack of research on soil MPs compared to marine MPs. Beaches are bridges between land and ocean and present equally understudied sites of microplastic pollution. Visible-near-infrared (vis-NIR) has been applied successfully for the measurement of reflectance and prediction of low-density polyethylene (LDPE), polyethylene terephthalate (PET), and polyvinyl chloride (PVC) concentrations in soil. The rapidity and precision associated with this method make vis-NIR promising. The present study explores PCA regression and machine learning approaches for developing learning models. First, using a spectroradiometer, the spectral reflectance data was measured from treated beach sediment spiked with virgin microplastic pellets [LDPE, PET, and acrylonitrile butadiene styrene (ABS)]. Using the recorded spectral data, predictive models were developed for each microplastic using both the approaches. Both approaches generated models of good accuracy with R² values greater than 0.7, root mean squared error (RMSE) values less than 3 and mean absolute error (MAE) < 2.2. Therefore, using this study's method, it is possible to rapidly develop accurate predictive models without the need of comprehensive sample preparation, using the low-cost option ASD HandHeld 2 VNIR Spectroradiometer.