THE USE OF CROSS-CORRELATION IN THE ESTIMATION OF LARGE-SCALE HYDRAULIC CONDUCTIVITY IN A KARST AQUIFER

In karstic aquifers, large scale estimates of hydraulic conductivity can often be difficult to establish due to local changes in lithology, preferential flow paths such as conduits/ fractures, and the small-scale estimation given by aquifer tests. Estimating large-scale hydraulic conductivity using existing monitoring well and river gage data could be a relatively simple way to establish a real world estimate of a particular aquifer which would be useful in groundwater flow models and the interpretation of hydrogeological data. In this study, we test whether cross-correlation analysis of monitoring well and river stage data can be used to help estimate regional hydraulic conductivity in a karst aquifer. Cross-correlation has previously been used to estimate aquifer characteristics in coastal carbonate aquifers by monitoring well head fluctuations throughout tidal cycles to estimate transit times of groundwater flow in a karstic sink/ rise system, and to qualitatively assess proximity of conduits and general aquifer characteristics. A portion of the upper Floridan aquifer within the Suwannee River basin in North-central Florida provides the geometry for the study. MODFLOW was used to generate synthetic well hydrographs using applied aquifer hydraulic conductivity and specific yield, actual river stage data, and estimated recharge. The lag time of the pressure signal from USGS gaging stations to monitoring wells was then determined using the cross-correlation function of MATLAB, and hydraulic conductivity was estimated by using an analytical solution for one-dimensional aquifer response to a sinusoidal fluctuation at the boundary. Results were then compared to the input hydraulic conductivity values. When using a simple distribution of hydraulic conductivity, estimated values using this method ranged from 46%-133% of the applied hydraulic conductivity values. Error is likely due to the inaccuracy created by assuming sinusoidal fluctuations for changes in river stage, and the geometry of the study region.