Espectroscopia de reflectância no mapeamento e modelagem de solos no Semiárido

Abstract

Knowledge about the different types of soils and your characteristics in the environment is essential for the proper use and management these, demanding the use of soil mapping. However, the use of traditional soil mapping methods has been shown to be costly and time-consuming, requiring alternative techniques in order to establish a relationship of the way the soil is formed and its distribution in the environment. Each soil class presents differentiated behaviour in the face of electromagnetic radiation and by this characteristic, it is judged that it is possible to distinguish between them by means of reflected energy. Thus, this work aimed to: i) identify and characterize the different soil classes existing in the Fazenda Experimental Lavoura Seca, belonging to the Federal University of Ceará (UFC), in the municipality of Quixadá, Ceará, through the methods of traditional and digital mapping, evaluating the spectral behavior of soils existing in the area and ii) to quantify some of their attributes by spectral analysis. A semi-detailed pedological map was elaborated on the scale 1:20000, by the traditional method of soil mapping and the collection of the soil samples was carried out with a Dutch auger in regular grid, at depths 0-20, 40-60 and 80-100 cm when possible, followed by physicals, chemicals and spectral analysis. The digital map was obtained by the principal component analysis of the spectral data of surface, subsurface and profile horizons soil samples, followed by geostatistics for spatialization and spatial dependence verification. The mapping spectral units were delimited by employing an unsupervised classification in the image obtained from the union of the maps of isolines from geostatistics. There were some correspondences between the delimited mapping spectral units and the delimitations obtained in the traditional mapping. The values of total area occupied by each mapping unit in the two methodologies used remained close, however, should be made aware of the elements present in the soil constitution that may influence the results obtained. The analysis of soil attributes by proximal sensing can provide the quantification of some attributes and satisfactory results were obtained for Mg, BS and CEC in most of the methods of transformation of the evaluated spectral data. The spectral data transformation in absorbance presented itself as the treatment that most was able to provide models satisfactory for most of the evaluated attributes.