Estimation of soil organic matter content using an excitation-emission matrix and partial least squares regression, and evaluation of the impact of the inner filter effect

Abstract

In soil analysis using fluorescence spectroscopy, an inner filter effect (IFE) caused by light absorption and reabsorption by the soil itself is observed. In this study, we used an excitation-emission matrix (EEM) to estimate water-soluble total organic carbon (WS-TOC) and water-soluble total nitrogen (WS-TN) in soil and to evaluate the impact of IFE. Partial least squares regression (PLSR) was applied to water extracts and raw soil samples, and the results showed that a prediction accuracy of R²_cv = 0.89 was achieved for WS-TOC even in raw soil. Furthermore, the influence of IFE was confirmed in high-concentration samples, and by applying a correction coefficient map created from a dilution series, the PLSR model for water extracts improved to an R²_cv of 0.94 for WS-TOC. These results demonstrate the potential for rapid assessment of soil organic matter-related components using a combination of EEM and PLSR, as well as the importance of evaluating the influence of IFE.