Acceso usuarios
In this study deterministic, multivariate and stochastic methods are applied to a combined temporal and spatial monitoring data set, in order to assess nitrate and pesticide levels and contamination risk in shallow groundwater. The case study involves an area in the Mondego River alluvial body in central Portugal, where agriculture is the main land use, with predominantly maize, rice and some vegetable crops supported by river water irrigation. Factorial correspondence analysis (FCA), reducing the original data matrix to a small number of independent orthogonal factors, is applied to detect associations between nitrate levels, land use (crop type), lithology and groundwater depth. Indicator-geostatistical techniques are used to create maps indicating the probability of nitrate concentrations in groundwater exceeding predetermined threshold values, including the drinking water standard (98/83/EC) and vulnerable zone designation criterion (91/676/EEC) of 50 mg/l NO3-. For pesticides the leaching potential is determined by calculating the Groundwater Ubiquity Score (GUS), based on the sorption coefficient and soil half-life for each pesticide compound. Results for nitrate show an overall very low risk of exceeding 50 or 25 mg/l, whereas the risk of exceeding 9.5 mg/l (third data quartile) is particularly high in areas where FCA shows correlation of nitrate contamination with vegetable and maize crops, aerobic conditions, lower groundwater levels and to some extent, coarser grained sediments. On the contrary, nitrate levels under rice are lowest and correlated to a reduced environment, finer-grained sediments and a higher water table. Denitrification is found to be an important attenuation process, as well as dilution by surface water irrigation and precipitation. Crop type and irrigation source are seen to have a large influence on the nitrate contamination potential of groundwater. Total concentrations of the analysed pesticide compounds above the regulatory limit of 0.5 µg/l are observed in 32% of the analysed water samples, with a maximum value of 16.09 µg/l. The probability maps provide a particularly interesting example of how multiple-well monitoring results over a certain period can be condensed into single maps and used by water engineers, managers and policy-makers.
