2007 GSA Denver Annual Meeting (28–31 October 2007)

Paper No. 2
Presentation Time: 8:15 AM

SPATIAL CROSS-CORRELATION ANALYSIS OF NEXRAD PRECIPIATION AND SPRING DISCHARGE AND APPLICATION TO REGIONAL GROUNDWATER MODEL OF A KARST AQUIFER


BUDGE, Trevor J., Department of Geological Sciences, University of Texas at Austin, 1 University Station C1100, Austin, TX 78712-0254 and SHARP Jr, John M., Geological Sciences, The University of Texas, Jackson School of Geosciences, Austin, TX 78712-0254, trevor_budge@urscorp.com

Many studies document the use of cross-correlation analysis of precipitation and spring flow to characterize groundwater flow through karst basins. This study uses spatial precipitation data to provide insight into spatial differences in the relationship of rainfall to spring discharge. NEXRAD precipitation estimates provide a measurement every 4 km for the majority of the continental US. By using each estimate location as a separate rain gauge and completing a cross-correlation analysis on the rainfall estimate and nearby spring discharge, a map of maximum correlation and lag time can be created for the spring catchment zone. Results illustrate the ability to delineate estimates for recharge zones and possible spatial differences in the recharge patterns based on spring discharge levels rates. The analysis was completed on Barton Springs in Austin, Texas. It demonstrated that, during low flow conditions at Barton Springs, precipitation in the southern portion of the aquifer impacted the flow at Barton Springs more than in high flow situations. This relationship was used to refine inputs to Barton Springs/Edwards Aquifer regional groundwater availability model. Calibrated recharge values are adjusted based on the zones delineated from high and low flow conditions at Barton Springs. A sensitivity analysis using weighted calibrated recharge values based on the zones of low and high discharge at Barton Springs demonstrates the usefulness of spatial cross-correlation analysis in constraining the groundwater model inputs.