In Situ Analysis of Soil Mineral Composition Through Conjoint Use of Visible, Near-Infrared and X-Ray Fluorescence Spectroscopy

Abstract

Soil mineral composition impacts soil behaviour but field estimation of the soil mineralogy has been difficult. Preliminary results of a model to rapidly quantify soil mineral composition are presented. To investigate the efficacy of predicting soil mineral composition in situ two soils under agriculture from New South Wales, Australia, was scanned with visible–near-infrared (Vis–NIR) and portable X-ray fluorescence (pXRF) spectrometers to a depth of 1 m at 2.5 cm sampling increments. The Vis–NIR spectra were preprocessed and the presence of individual minerals was assessed using pattern matching with mineral end-member libraries. Rule-based iterative partitioning was then applied on the recorded pXRF elemental compositions based on known stoichiometric ranges of the identified minerals. This gave a mineral abundance prediction and distinguished between mineral groupings not clearly differentiated by Vis–NIR spectra alone, i.e. kaolinite, illite and smectite. Predicted mineral composition compared favourably to existing mineralogical interpretation of horizon-based random powder and orientated clay samples analysed using laboratory X-ray diffraction. This fine-scale mapping of the distribution of soil minerals in situ has potential to enhance soil morphological description, support site-specific pedogenetic theories and may be further used to moderate properties predicted directly from the Vis–NIR and pXRF spectra.