Variáveis pedogeoquímicas e mineralógicas na identificação de fontes de sedimentos em uma bacia hidrográfica de encosta

Abstract

Inadequate land use, ignoring its agricultural fitness can, in a short time, transform thousands of hectares in fragile systems highly susceptible to degradation processes. The use of the fingerprinting method in the identification of sediment sources has been efficient and there are many studies in the international literature proving its effectiveness. In Brazil, the studies in the matter are still incipient, but have demonstrated efficiency in detecting changes in the proportion of contribution of the sources in the production of sediments. Among the variables used in this method, there are some minerals that are used as tracers of soil erosion, due to their high stability in the environment and magnetism property. However, the use of mineralogy together with models of erosion prediction as well as in the identification of sources of sediment production is not common. The objective of this work was to estimate the proportion of contribution of two groups of sources, road+channels+grid and superficial+subsuperficial, in the production of sediments in a watershed slope. For this purpose, two groups of variables were tested, pedogeochemical and pedogeochemical+mineralogical to compose the prediction models. For the identification of the sources the fingerprinting method was used, which compares the soils of different sources and the sediments that are found in suspension in the drainage channel, using tracing variables. It was possible to identify the contribution of each one of the groups of sources of sediment production in the Vacacaí-Mirim River basin, except in the T1point. In this point the scale was determining in the result, due to the low number of samples of this sub-basin. The greatest contributions in the production of sediments came from superficial erosion of the soil. The use of an only group of variables to compose the model of identification of sources, suppressing variables that generate uncertainty, has led to more coherence in the presented results and reduced the space-time variability of the analysis. The roads were presented as a potential source of erosion in the basin. The adding of mineralogical variables improved the results, decreasing the errors. However, using only one group of variables for the prediction model did not lead to any significant change in the results.