Using vis-NIR and pXRF data to distinguish soil parent materials – An example using 136 pedons from Wisconsin, USA

Gozukara G., Zhang Y., Hartemink A. E.

Geoderma, vol.396, 2021 (SCI-Expanded) identifier identifier

  • Publication Type: Article / Article
  • Volume: 396
  • Publication Date: 2021
  • Doi Number: 10.1016/j.geoderma.2021.115091
  • Journal Name: Geoderma
  • Journal Indexes: Science Citation Index Expanded (SCI-EXPANDED), Scopus, Academic Search Premier, PASCAL, Aquatic Science & Fisheries Abstracts (ASFA), BIOSIS, CAB Abstracts, Environment Index, INSPEC, Pollution Abstracts, Veterinary Science Database
  • Keywords: Loess, Terra rossa, Geochemical, Decision tree, Soil elements, Soil vis-NIR spectra, RED CLAY, SPECTRA, SPECTROSCOPY, ENVIRONMENT, CATENAS, SPACE
  • Eskisehir Osmangazi University Affiliated: No


© 2021Many soils have two or more parent materials. It is not always possible to distinguish these materials in a soil pit or in the field. Here we present a study in the Driftless Area of Wisconsin, where soils were developed in loess covering the weathered impurities of dolostone, termed terra rossa. We evaluated the geochemical signature of 136 pedons with loess (A and Bt) and/or terra rossa (2Bt) horizons and developed decision tree models to distinguish the Bt and 2Bt horizons from different geochemical indicators. In total 409 samples were collected from A, Bt, and 2Bt horizons and the samples were scanned using vis-NIR and pXRF spectrometers to obtain the color information, elemental data, and to calculate weathering indices. The Munsell color, texture, total carbon (TC), total nitrogen (TN), and pH were measured in the laboratory. Twenty one types of input data were compared in decision tree models. It was found that the 2Bt horizons had higher clay, pH, Mg, Al, Fe, Ca, Mn, Zn, Rb, CTR, and Ti/Zr than the loess A and Bt horizons. The Ruxten index (R), clay, horizon thickness (HT), CaO/TiO2 (CTR), and Ti/Zr were robust variables to distinguish loess Bt and terra rossa 2Bt using a decision tree model. However, similarities in the values of some elements and weathering indices compromised the characterizations of Bt and 2Bt horizons, which is the result of mixing of parent materials through bioturbation over time. It can be concluded that pXRF and vis-NIR spectra can be used to distinguish parent materials in soil profiles, and pXRF spectra had slightly better overall accuracy than vis-NIR spectra.