No abstract

A geostatistical method was applied to optimize an existing groundwater-level monitoring network in the Upper Floridan aquifer for the South Florida Water Management District in the southeastern United States. Analyses were performed to determine suitable numbers and locations of monitoring wells that will provide equivalent or better quality groundwater-level data compared to an existing monitoring network. Ambient, unadjusted groundwater heads were expressed as salinity-adjusted heads based on the density of freshwater, well screen elevations, and temperature-dependent saline groundwater density. The optimization of the numbers and locations of monitoring wells is based on a pre-defined groundwater-level prediction error. The newly developed network combines an existing network with the addition of new wells that will result in a spatial distribution of groundwater monitoring wells that better defines the regional potentiometric surface of the Upper Floridan aquifer in the study area. The network yields groundwater-level predictions that differ significantly from those produced using the existing network. The newly designed network will reduce the mean prediction standard error by 43% compared to the existing network. The adoption of a hexagonal grid network for the South Florida Water Management District is recommended to achieve both a uniform level of information about groundwater levels and the minimum required accuracy. It is customary to install more monitoring wells for observing groundwater levels and groundwater quality as groundwater development progresses. However, budget constraints often force water managers to implement cost-effective monitoring networks. In this regard, this study provides guidelines to water managers concerned with groundwater planning and monitoring.

A groundwater-level monitoring network was designed for the Upper Floridan aquifer in southern Florida, U.S.A., within the boundaries of the South Florida Water Management District. The objective of the investigation was to design a groundwater monitoring network for the Upper Floridan aquifer that recommends the number and locations of monitoring wells that will provide equivalent or better quality data compared to the existing monitoring network. This was accomplished by designing a spatial distribution of wells that will improve the accuracy of groundwater-level data measured over time and reduce data estimation errors that occur when spatially interpolating between wells in the Upper Floridan aquifer. Statistical and geostatistical analyses were performed on groundwater-level data, groundwater levels were estimated in unmeasured areas by interpolation, and proposed monitoring well locations were delineated based on acceptable levels of error for groundwater levels between wells and in areas where no wells are located. Semivariograms, which illustrate spatial autocorrelation, were plotted, and a potentiometric surface map representing salinity-adjusted mean heads in the Upper Floridan aquifer based on the current monitoring network was constructed using ordinary kriging to interpolate between wells. Three hexagonal grid network designs with different cell diameters were evaluated to determine the number and locations of proposed monitoring wells along with the resulting errors that would be associated with the different proposed design networks. A hexagonal grid with wells spaced at 29,261 m was considered to be the most practical alternative based on the costs associated with installing new wells. The recommended optimum monitoring network consists of 58 new wells added to 44 existing wells, which would result in a network of 102 wells that has a prediction standard error with a mean of 1.45 m and a range from 0.07 m to 2.62 m. 48