Development of Calibration Models for SiO₂, Fe₂O₃, and Na₂O Using Subsurface Diffuse Reflectance Spectra

Abstract

We developed calibration models for available silicate, free iron oxide, and exchangeable sodium in paddy fields, using subsurface diffuse reflectance spectra obtained by a tractor - mounted soil analyzing system. The calibration model estimation applied 2nd derivative pretreatment and partial least squares regression. The respective calibration model estimation accuracies for available silicate, free iron oxide, and exchangeable sodium showed coefficients of determination greater than 0.90 and residual prediction deviations greater than 3.0. The results confirmed sufficient model estimation accuracy, in the case of each soil property, for the preparation of useful soil maps. Diffuse reflectance spectra of fresh soils showed that further calibration models could be analyzed.