Groundwater monitoring network design using minimum well density

Limitations on resources for groundwater quality monitoring projects demand definition of groundwater contaminant plumes with a minimum number of sampling wells. Placement of these wells, especially in the case of a sequential design, in which wells are added in increments, requires effective use of the information gathered from the wells. The relationship between loss of information (or plume definition error) and degree of reduction in the number of sampling wells was investigated. Subsequently, a sequential network design procedure was developed and then tested with data from two historic tracer tests. The procedure was used to construct networks of varying number of wells to investigate the relationship between average plume quantification error and number of wells. In the first phase, a proprietary sampling design product (E4) was used to select variably sized subsets of sampling wells from two existing natural gradient tracer tests. The tracer plume defined by data from each subset of wells was compared to the plume defined by data from the full set of wells. Differences between each subset plume and its corresponding full set plume were quantified and used to generate a function relating degree of error to sampling density and time since tracer introduction. Combined analysis of three sampling events from one tracer test and two from another revealed that the number of wells used for four of the events could have been reduced with minimal loss in contaminant plume definition. The second phase of this research consisted of the formulation of a sequential network design procedure that used only information obtained from well sampling to predict plume direction and spread. Tracer test data were used to test the procedure . Networks were constructed in increments, and appended areas were located according to the estimated plume transport parameters and assumed management factors. E4 was used to site wells within the prescribed areas. Network designs were evaluated by a plot of average plume definition error versus total number of wells in the network. Results indicated that the number of wells in the original tracer tests could have been reduced without significant loss of plume information when using the sequential design procedure and E4, even though less information was used at the beginning of the design than used in designing the original tracer test networks. This suggests that the sequential design procedure described herein, used in conjunction with E4, could provide substantial benefit if applied in the design of groundwater monitoring well networks.