Mapeamento digital de atributos físicos e químicos do solo na Península de Keller - Ilha Rei George - Antártica Marítima

Abstract

This study aimed to evaluate different methods of digital mapping for the prediction of physical and chemical properties of soils, and to provide maps and data tosupport the analyses of the main processes involved in the pedogenesis and landscape development, through the use of land attributes in the Keller peninsula, King George Island, Maritime Antarctica. Soil samples were collected at 55 points in a semi regular grid fashion, set to cover much of the ice-free area of the peninsula. Samples were collected at depths of 0-5, 5-10 and 10-30 cm. The variables pH, total organic carbon (TOC), Ca, Mg, Al, H + Al, P, K, Na, sum of bases (S), base saturation (V%), and percentage of fine earth were quantified. Only the attributes that showed data normal distribution naturally were spatialized, and they were the following: pH, S, TOC (5-10 cm), and fine earth percentage. Three methods of spatialization were tested: multiple linear regression (MLR), ordinary kriging (OK) and kriging with external drift (KDE). For usage in models of RLM and KDE, 12 covariates were generated from a digital elevation model (DEM) with resolution of 5 m. To choosethe covariates, it was applieda stepwise procedure, using the Akaike criterion, with the further analysis of correlation between selected covariates,in order to avoid the effect of multicollinearity. The same covariates used in the MLR were used in the KED, to predict the respective variables. The results showed that the use of covariates may help in the understanding of pedogenic processes in the area and the spatial distribution of some physical and chemical attributes of lands in the Keller Peninsula. The OK method was more efficient at the spatialization of the attribute pH.On the other hand, the KDE method was more efficient in spatializing the attributes sum of bases and total organic carbon. And the MLR method was the most efficient for interpolation of the fine earth percentage attribute, according to the indices evaluated.