Classificazione di suoli nudi e loro tessitura in un'area agricola della Lombardia attraverso dati telerilevati

The objective of this thesis is to make a contribution for the description of a parameter necessary to assess the soil protective capacity, the soil texture: this physical property is essential in various groundwater vulnerability models. Through the use of Landsat satellite (optical multispectral sensor), it was possible to characterize the land cover in an agricultural area of the Lombardy Region, examining soil texture and how these change in the time, in relation to climatic conditions and planting cycles of crops at the time of satellite acquisition. The Landsat data processing relies on Linear Spectral Mixture Analysis (LSMA) and radiometric campaign to collection in situ spectral signatures, useful to compare the field spectral signatures to spectral responses of the satellite image. This technique allows the quantitative characterization of the endmembers abundances identified in the sub-pixel level and, through the decision tree, tool based on nodes "higher/lower" than specific thresholds, it was possible to classify each pixel and produce bare soil maps. The use of pedological maps provided the soil texture information in the study area, describing the spatial distribution and characterization of soils. The integration of pedological information to EO products, through a statistical analysis, it allowed to correlate the bare soil abundances, the reflectance data and soil texture classes, to see if a specific abundance or reflectance is descriptive of one or more textural classes. The possibility to observe the spatial variability of bare soil classes in the time, allows to understand what are the causes that determine these changes. Finally, the updating of soil types, characteristics and condition in which terrain is located, with good temporal resolution, could increase the data significance, useful to describe the soil protective capacity.