Groundwater age dating using single and time-series data of environmental tritium in the Moeda Synclyne, Quadrilatero Ferrifero, Minas Gerais, Brazil

摘要

The use of tritium in the identification of the ?age? of groundwater is being used since the discovery of its potential for this purpose in 1954. The present work sought to apply and discuss the use of the different types of lumped parameter models (LPM) in the Caue? and Gandarela aquifers, in order to obtain the groundwater mean residence time based on tritium data. Water samples from several discharges related to the Caue? and Gandarela aquifers were collected during the month of September of 2017. The reconstruction of the historical series of tritium contents in the precipitation at the GNIP station CDTN-Belo Horizonte was carried out based on data from the Kaitoke (New Zealand) and Adelaide (Australia) stations in addition to the Brazilian stations Brasilia, Rio de Janeiro and Salvador. The sample results and the data of the reconstructed historical series of the tritium concentration in the CDTN station during the period between 1953 and 2017, served as the basis to use the lumped parameter models with the aid of the software TracerLPM, provided by the United States Geological Survey. From the overlap of some sample points with tritium samples collected in different years by different authors, an adjustment of some models of residence time distribution was possible. The models used were exponential (EM), exponential-piston (EPM), and dispersive (DM). The model that best fit to data was the EPM model. The mean transit time for the EPM model in the analyzed points varies from 63 to 95 years, with a median value of 73 years. Differences between values of the EPM ratio parameter were verified. However, all values indicate an exponential flow component greater than the piston flow component. The EPM model suggests a long retention time of the tracer in the aquifer system. It can be used as an approximation of the retention time for environmental contamination. The exponential model (EM) and piston flow model (PFM) were used together. Based on the data of a single tritium sample collected in time, and the average value of residence time provided by each model, a first approximation of the mean residence time value can be obtained. The results suggest that this is a better approximation than assuming a pure exponential model or piston flow model to describe the data of a single tritium sample in time as were made by previous local works. In general, a short residence time was estimated for the Caue? aquifer as a whole with some discharges with long residence time that drains regional flows from the deepest part of the Caue? aquifer, reflecting the complex structural geology fabric.