Stochastic contaminant transport monitoring in heterogeneous sand and gravel aquifers of the United Arab Emirates

Abstract

Listen

Translate article

Stochastic numerical simulations were performed to investigate the evolution of plumes from pulse sources that can result from accidental leaks. The stochastic advection–dispersion equation was solved for hydraulic conductivities typical to the heterogeneous sandy and gravel aquifers encountered in the United Arab Emirates. Dispersivities were similar to those found in field studies at sandy aquifers, such as those conducted at Borden and Cape Cod, and at the Vejen, Denmark tracer tests. Our work showed that the detection probability, Pd, of a monitoring network was affected strongly by the medium’s dispersivity with a large number of wells (larger than 12) required, even in relatively simple geological environments, in order to detect contaminants with confidence. Monitoring systems following minimum regulatory requirements in terms of the number of wells were able to detect contamination at best in only one out of five cases. The frequency of sampling did not appear to be critical when the dispersivity was low and bi-annual sampling appeared to be satisfactory. In highly dispersive media monthly sampling was needed in order to increase detection. Increase of a medium’s dispersivity in relative homogeneous aquifers reduced the performance of large well-systems to less than 50 %. Strongly heterogeneous and dispersive subsurface environments led all monitoring systems to fail in detection, irrespectively of frequency of sampling. Finally, large contaminant quantities did not improve the detection capabilities of low density well-systems with detection enhancements restricted to high density ones.