Use of Near Infrared Spectroscopy for the Determination of Organic Matter and Total Nitrogen of Soils

Chemical analysis in soils are essential for decision making in agriculture.Currently, the aim is to use non-destructive, fast, and precise techniques that allow for results to be obtaining easily, such as NIRS near infrared spectroscopy. The objective was to generate prediction equations for Organic Matter (MO) and total Nitrogen (total N), by using NIRS spectra. 459 soil samples were processed by wet chemistry and by NIRS and various data transformations analyzed by partial least squares were used. In the selection, the values of the coefficient of determination (R2), the root mean square error of prediction (RMSEP) and the predictive residual deviation (RPD). The best model for MO was without transformation with absorbance (R2=0.90, RMSEP=0.29 and RPD=1.3) and for total Nitrogen with transformation of 1a derivative Savitzky-Golay (R2=0.84, RMSEP=0.09 and RPD=2.5). This indicates that the absorbance values of the NIRS spectra can be used to predict the OM and N values of the soil, using partial least squares models.

• Fecha de publicación: 2023
• Institución: Universidad Pedagógica y Tecnológica de Colombia
• Repositorio: RiUPTC: Repositorio Institucional UPTC
• Idioma: spa
• OAI Identifier: oai:repositorio.uptc.edu.co:001/15339
• Acceso en línea: https://revistas.uptc.edu.co/index.php/ciencia_en_desarrollo/article/view/13942; https://repositorio.uptc.edu.co/handle/001/15339
• Palabra clave: machine learning; transformaciones; métodos ´ópticos; análisis químicos; regresiones parciales; chemical analysis; machine learning; optical methods; partial regressions; transformations
• License: http://purl.org/coar/access_right/c_abf2