Abstract

TORRES-GUERRERO, Carlos Alberto et al. Quantification of soil components in thin sections: High resolution mosaics versus individual images. Bol. Soc. Geol. Mex [online]. 2020, vol.72, n.1, e300919.  Epub Dec 22, 2020. ISSN 1405-3322.  https://doi.org/10.18268/bsgm2020v72n1a300919.

Changes in soil structure can be monitored in undisturbed soil samples through the micromorphometric analysis of thin sections. In this methodology, it is common that individual images (three to ten repetitions) are used and that they cannot be related in different scales between soil components; in addition, although a minimum study area is established, its representativeness in the soil components is unknown. The objective of this study was to quantify the soil constituent (pores, aggregates, and roots) and to compare the values obtained from individual images versus high-resolution mosaics from a complete thin section. Unaltered samples were collected in three soils with different clay contents (Entisols, Inceptisols, and Vertisols) and presence of maize roots in the same phenological development stage (physiological maturity). Three thin sections of each soil (5 × 7 cm × 30 μm 1865 mm2) were prepared and sequential images of 11.1× 7.4 mm (63 in total) were obtained at 2× magnification using a petrographic microscope and plane polarized light (PPL). The high-resolution mosaics (2.6 µm píxel-1) at a colour depth 24 bits (8 bits × 3 bands in standard RGB) were built using space operators; subsequently, three, five and ten images were randomly selected. The individual images and the mosaics were transformed by principal component analysis in ArcGis® and soil constituents were delimited according to their variances values. The results indicate that individual images are recommended only to quantify porosity or in homogeneous systems in structure and color, but not in heterogeneous systems where the data obtained show high variability. Even when the results are similar to those calculated in the mosaics, the dispersion of the data is high (variance 5 times greater than the mean) and with little representation. In contrast, high-resolution mosaics offer the total quantification of the thin section and soil components can be grouped into classes or categories to observe intra or inter relations in the soil system. In addition, the soil components can be related to different scales, for example macroaggregates and roots.

Keywords : Image analysis; segmentation; pedofeatures; principal component analysis; micromorphometry.