Can We Use Visible-Near Infrared and Mid Infrared Spectroscopy as a Tool for Wetland Soil Identification?
ORCID
- Wijewardane: https://orcid.org/0000-0001-8962-9451
- Bheemanahalli: https://orcid.org/0000-0002-9325-4901
MSU Affiliation
College of Agriculture and Life Sciences; Department of Plant and Soil Sciences; Department of Agricultural and Biological Engineering; James Worth Bagley College of Engineering
Creation Date
2025-12-11
Abstract
The wetland delineation process is primarily based on the visual recognition of anaerobic soil indicators by trained individuals, and is a complex and subjective task that is prone to error. Therefore, an objective alternative is needed to identify wetland soil; however, no such method currently exists that is rapid and easy to deploy. Accordingly, the objective was to evaluate soil spectroscopic classification approach as a rapid, deployable alternative by testing its feasibility to differentiate wetland from non-wetland soils. This study used visible-near infrared and mid-infrared (MIR) ranges for this task. A total of 440 wetland and non-wetland soils were sampled across Mississippi followed by obtaining visible/near-infrared and MIR spectra under both fresh and dried conditions. Support Vector Classification (SVC) and Random Forest (RF) methods were then used to classify spectra based on wetland/non-wetland status with a 75 %/25 % calibration and validation split. This split was repeated for 50 iterations to obtain randomized calibration and validation sets for model calibration and achieve average model performance. The average classification accuracy across all models was ∼91 %, with the highest accuracy of 99.6 % achieved on MIR spectra. The accuracy, precision, and recall scores showed similar performances between SVC and RF ranging their values from ∼80 % - 100 %. This study showed the reliability and ease of wetland determinations using spectroscopy as an objective and rapid wetland recognition method, while reducing the need for an expert for determination.
Publication Date
12-1-2025
Publication Title
Vibrational Spectroscopy
Publisher
Elsevier
Creative Commons License
This work is licensed under a Creative Commons Attribution-NonCommercial-No Derivative Works 4.0 International License.
Recommended Citation
Whatley, C. R., Wijewardane, N. K., Silva, C. A., Tagert, M. L., Bheemanahalli, R., & Parajuli, P. (2026). Can we use visible-near infrared and mid infrared spectroscopy as a tool for wetland soil identification? Vibrational Spectroscopy, 142, 103875. https://doi.org/10.1016/j.vibspec.2025.103875
