Characterization of two Italian vertisols with the emphasis on optical and electron microscopy, clay mineralogy and image analysis
A study has been made of a black and brown Vertisol developed from torbidites in the Appenine highlands of Southern Italy. The investigation includes profile morphology, micromorphology, chemical properties, physical properties and clay mineralogy. In addition digital image analysis techniques were developed to identify and classify features in thin sections generally and specifically in one of the Vertisols. These soils interdigitate on a complex wavy topography. The dark Vertisol has distinct vertic features such as large slickensides, large cracks and self mulching; they are more mature, deeper and generally more fertile. The Brown Vertisols have weakly expressed slickensides and cracks but self-mulching is well manifested; these soils are younger, shallower and less fertile. The study of the thin sections has enabled a detail characterization of many features. A detailed mineralogical study of the clay was carried out. The clay minerals are dominated by smectite. The low charge beidellite and montmorillonite of the parent material change into high charge beidellite in the upper horizons. Both of the soils showed two distinct processes: homogenization and horizon differentiation. Two experiments on digital image analysis were conducted using multivariate classification widely used in remote sensing. In the first investigation multivariate statistical analysis through the creation of multilayer images was achieved using the ERDAS software. A supervised classification applied to a combination of plane transmitted light (PTL) and circularly polarized light (CPL) images allowed the differentiation of quartz, clay coatings, matrix and pores in an Alfisol. With the second experiment using the Semper software a combination of brightness and shape classification has been demonstrated. A series of new techniques have been tested. The classification of the Vertisol used plane polarized light (PPL), CPL, incident UV light and the 2nd Component of PCT. These enabled the differentiation of very complex features. The procedure was compared with point countings and gave very good results.