Soil Nitrate Prediction Using FTIR-ATR, Vis-NIR, and Raman Spectroscopy

Authors

Sabiha Ferdous, University of Nebraska-LincolnFollow

First Advisor

Yugeng Ge

Date of this Version

3-2024

Document Type

Article

Citation

A thesis presented to the faculty of the Graduate College at the University of Nebraska in partial fulfilment of requirements for the degree of Master of Science

Major: Agricultural and Biological Systems Engineering

Under the supervision of Professor Yufeng Ge

Lincoln, Nebraska, March 2024

This work was supported by a grant from USDA-NIFA (Award #2018-67007-28529)

Comments

Copyright 2024, Sabiha Ferdous. Used by permission

Abstract

Fourier transform infrared spectroscopy (FTIR) combined with Attenuated total reflectance (ATR), Visible-Near Infrared spectroscopy (Vis-NIR), and Raman spectroscopy (RS) are non-destructive techniques for rapid determination of nitrogen compounds in soil. Leveraging FTIR-ATR and Vis-NIR spectra using partial least squares regression (PLSR) modeling, the study aims to predict soil nitrate content and explored the feasibility of Raman spectroscopy to detect nitrate (NO₃^-), nitrite (NO₂^-), and ammonium (NH₄⁺) in soil. Soil samples were collected from four different fields, dried, sieved (2mm), and then used for collecting spectra (FTIR-ATR and Vis-NIR). Laboratory analysis was done to determine nitrate content. Another set of moist soil samples was used for FTIR-ATR spectra collection, and the nitrate content was determined using chemical analysis. A PLSR model was trained and validated with obtained spectra with the corresponding soil NO₃^- dataset. For FTIR-ATR spectra, results demonstrated that, with dry soil samples (n = 170), the R² = 0.69 and RMSE = 11.15 ppm (for 1^st phase), where dry soil (n=75) showed R² = 0.37 and RMSE = 10.03 ppm whereas the moist soil (n = 75) showed R² = 0.29 and RMSE = 10.43 ppm (in 2^nd phase). Vis-NIR spectra was collected using FieldSpec and spectral evolution (SE) instruments for dry soil, where a comparatively better result was found: R² = 0.66, RMSE = 1.27 ppm using SE instrument. This study also found that, RS and UV-Vis spectrophotometry methods can be potential methods for detecting NO₃^-, NO₂^-, and NH₄⁺ in salt solutions. However, these methods did not show NO₃^- peak for dry soil samples due to low NO₃^- content. For FTIR-ATR and Vis-NIR spectroscopy, spectra preprocessing, and model development can enhance the accuracy of prediction. Moreover, instead of RS, Surface Enhanced Raman Spectroscopy can be an alternative method that can be used for soil NO₃^- detection. The model can be a potential tool for soil NO₃^- prediction in diverse soil conditions. This approach might be a rapid solution which can reduce complexity. Overall, this study may offer a valuable, sustainable soil nutrient management tool.

Advisor: Yufeng Ge