LEVEL-DETERMINED SAMPLING IN AN UNCASED BOREHOLE

Four novel sampling devices (two sock samplers and two active devices) were tested for their ability to obtain level-representative groundwa ter from a 90 m uncased borehole in a Permo-Triassic sandstone aquifie r. The samples were compared with those taken from a nearby piezometer nest. The results from the open borehole were not consistent with tho se from the piezometers, particularly in the lower part of the aquifer . Field data and five-layer regional groundwater model were used in an attempt to understand the reasons for the variation between the resul ts. A strong downward vertical head gradient caused how into the top o f the borehole and out of the bottom, disturbing the natural chemical profile. This accounted for the difference in the results between the piezometer nest and the sampling devices in the lower part of the aqui fer. Further modelling showed that unreasonably long periods of flushi ng would be required to flush the zone of mixed water and therefore ob tain level-representative samples from the lower section of the boreho le. Sock samplers, which rely on natural groundwater flow to flush the aquifer, would have to wait several years, and so could only be used as semi-permanent devices. A conventional double packer method would r equire about 2 weeks of continuous pumping, whereas separation pumping would require several months of purging, unless larger pumps were uti lised.