Estimativa do número de amostras de solo para caracterizar elementos químicos utilizando fluorescência de raios X portátil em áreas contaminadas: estudo de caso em Resende

Abstract

The representativeness of the sampling is fundamental to support decisions, since the heterogeneity of the matrix is the main factor for data uncertainty in contaminated areas management studies. In this study, 112 surface soil samples were collected in a square mesh of 1 x 1 m, in an area of 15 x 6 m, at the Campus Regional de Resende (CRR) of Universidade do Estado do Rio de Janeiro (UERJ). Previously, companies of radiographic films production and metallic silver purification operated in this area. The soil samples were analyzed by pXRF, a screening method for metals detection. Based on the data collected, the number of samples that would allow a representative characterization of the area was calculated, using statistical methods, considering a relative error of 20% for 80 and 95% confidence intervals. The analyses indicated concentrations higher than the reference values for the parameters silver, mercury, lead, selenium, and zinc. The results of mercury possibly refer to spectral noise, not being related to the history of use and occupancy. The coefficient of variation of the soil obtained ranged from 132 to 225%. Considering a lognormal distribution, 76 samples were calculated for the highest confidence interval and 53 for the lowest. This would be the equivalent of transects with spacings of approximately 1.47 and 2.11 m. The results confirm the need for high resolution investigations in the source areas, as recommended by the Executive Board Decision of CETESB No. 038/2017/C (DD-038).