Proximal and remote sensing to soil mineralogy assessment

• Autor(a) principal: Rosin, Nícolas Augusto
• Data de Publicação: 2022
• Tipo de documento: Dissertação
• Idioma: eng
• Título da fonte: Biblioteca Digital de Teses e Dissertações da USP
• Texto Completo: https://www.teses.usp.br/teses/disponiveis/11/11140/tde-12042022-162844/
• topic: Digital soil mapping; Espectroscopia de solos; Mapeamento digital de solos; Pedometria; Pedometric; pXRF; pXRF; Soil spectroscopy

The mineralogy is the gear of soil processes, playing a fundamental role in relevant issues for humanity. However, access to mineralogical analyses is difficult due the difficulty of acquisition through traditional methods and alternative forms to reach it must be explored. This thesis was divided in two chapters that aimed: 1) To understand the fundamental interactions of the energy on pXRF information with emphasis on iron forms, moisture and SOM for use on soil science and 2) To map the abundances of major soil mineralogical components for the whole Brazilian territory at the surface and subsurface. In order to reach the first objective, three selective dissolution treatments were applied to remove: (i) soil organic matter (SOM), ii) SOM and poorly crystalline iron forms (o), iii) SOM and poorly crystalline plus well crystalline iron forms (d). One additional treatment iv) including water addition (+W) was also carried out. The pXRF was able to detect changes caused by the selective dissolution treatments and soil particle size distribution. The kaolinite, gibbsite,Fe2O3, Al2O3, SiO2, TiO2 and MnO contents were quantified with satisfactory accuracy (0.61< R2 < 0.97). Sources of uncertainty, mainly soil moisture, must be considered. The understanding of the fundamentals of energy interaction with the sample matrix in the X-ray range is the starting point for characterizing the soil through pXRF. In order to reach the second objective, The Brazilian Spectral Library (BSSL) with Vis-NIR-SWIR spectral data, was used to assess the relative amounts of hematite (Hem), goethite (Gt), kaolinite (Kt) and gibbsite (Gbs) in soil samples from Brazil. Terrain attributes (TA) and a synthetic soil image (SySI) with bare soil pixel from multitemporal Landsat images (1984 to 2020) were used as predictors. A novel approach was performed in order to obtain a bare soil image for the whole Brazilian territory. The model Random Forest (RF) was used for spatial prediction to obtain the mineral maps and their uncertainty by bootstrapping procedure. The Hem presented the more accurate results in RF models with R2 ranging from 0.48 to 0.56, followed by Gbs (0.42 to 0.44), Kt (0.20 to 0.31) and Gt (0.16 to 0.26). The proposed approach was able to reveal the spatial distribution of the relative abundance of minerals for the Brazilian territory. The mineral maps were in accordance with geology and soil legacy maps and also with the climate and terrain conditions. The approach proposed is an efficient method to obtain mineralogy information for large areas.