Predicting soil weathering indices of Alfisols using Vis-NIR, MIR, and XRF spectral data fusion

Authors: GÖZÜKARA, GAFUR - University Of Wisconsin; HARTEMINK, ALFRED - University Of Wisconsin; Young, Eric; ZHANG, YAKUN - Oregon State University

Submitted to: Catena
Publication Type: Peer Reviewed Journal
Publication Acceptance Date: 4/13/2026
Publication Date: 4/28/2026
Citation: Gözükara, G., Hartemink, A.E., Young, E.O., Zhang, Y. 2026. Predicting soil weathering indices of Alfisols using Vis-NIR, MIR, and XRF spectral data fusion. Catena. https://doi.org/10.1111/ejss.70337.
DOI: https://doi.org/10.1111/ejss.70337

Interpretive Summary: Assessing soil weathering intensity can help to better understand soil formation processes and fertility status. Soil weathering indices (WIs) characterize weathering intensity based on specific elemental concentrations and ratios reflecting the extent of weathering in a soil profile or among different soils. Chemical extraction of elements to determine WIs is routinely done, however proximal sensing is a non-destructive analytical method that is also available. Few studies have tested the feasibility of proximal sensors to predict WIs. Our objective was to evaluate weathering intensity and mineralogy using 517 soil samples from the Marshfield Agricultural Research Station and assess the ability of sensors and machine learning to predict six different WIs. Weathering intensity generally increased with soil depth, reflecting cumulative cation leaching processes from upper to lower horizons. Top horizons (Ap) were least weathered but the most fertile with respect to crop production. Five of six WIs were accurately estimated by proximal sensing, suggesting WIs can be estimated without the need for chemical extractions.

Technical Abstract: Weathering indices (WIs) can help evaluate weathering intensity and mineralogy of soils from different environments. The objectives of this study were to: 1) evaluate pedogenic relationships among, WIs, diagnostic horizons, drainage class, mineralogy, and crop history at an agricultural site in Central Wisconsin, and 2) develop models to predict WIs from proximal sensors and compare individual sensor predictions to all spectra combined. Soils were sampled at 150 locations (n = 517 soil cores) across the Marshfield Agricultural Research Station in 2023. Elemental concentrations for WIs were determined by X-ray fluorescence spectroscopy (XRF). Select WIs (Chemical Index of Alteration [CIA], Chemical Index of Weathering [CIW], Chemical Proxy of Alteration [CPA], Desilication Index [DI], Ruxton index [RI], and Weathering index of Parker [WIP]) were estimated by visible-near infrared (Vis-NIR) (350–2,500 nm), mid-infrared (MIR) (4,000–600 cm–1), and XRF (0–50 keV) using machine learning (partial least squares regression and cubist models). Weathering intensity increased with soil depth, reflecting cumulative cation leaching processes. Along with horizon type, crop history and soil drainage class also affected WI. The MIR spectra provided accurate predictions for DI (R2 = 0.95), followed by RI (R2 = 0.93) in glossic horizons. Combined spectra slightly improved predictions for RI (R2 = 0.95), CIW (R2 = 0.83), and CPA (R2 = 0.78) in glossic and Bt horizons (cubist model). The WIs were well predicted by proximal sensing, demonstrating their utility to characterize weathering intensity without the need for chemical extractions.