CASE STUDIES FROM A LONG-TERM KARST RESEARCH SITE ILLUSTRATING DATA REQUIREMENTS NECESSARY TO ADEQUATELY CHARACTERIZE AND REPRESENT FLOW IN KARST AQUIFERS AT MEANINGFUL SCALES

Case studies drawn from more than 10 projects in the Savoy Experimental Watershed (SEW) conducted during the last 7 years provide an understanding of data requirements that are necessary to fully characterize karst ground-water flow systems. Interpretation of test results are increasing our knowledge not only about the detailed hydrogeology of the SEW, but also are providing insight into minimal temporal and spatial data needs necessary to elucidate processes in this and similar settings.

The SEW is a long-term field research facility of 1250 hectares owned by the University of Arkansas, and shared with academic, state, and federal researchers. Although the initial focus has been on water-quality effects resulting from animal production on karst lands, intensive attention also has been directed to utilizing the infrastructure to develop borehole geophysical tools and develop new tracers. Basin 1 is well-characterized, and includes an existing infrastructure of instrumented trenches, surface runoff plots, channels, weirs, flumes, wells, springs, seeps, losing streams, lysimeters, and weather stations. Such a long-term site provides an in-situ field laboratory that generates meaningful data for ground-truth at varying scales, as well recurrent testing for a wide variety of antecedent conditions.

Based on work completed thus far, optimum data requirements necessary to adequately characterize karst aquifers include 1) continuous monitoring, including input stresses and output responses; 2) sampling over the complete hydrologic cycle; 3) evaluation and delineation of the entire flow system, including all boundaries, and vadose and epikarst components; 4) determination of flow directions and time of travel; 5) water-quality determination; 6) geochemical process identification; 7) assessment of aquifer hydraulic attributes, both in 3-D and temporally, and 8) accurate water budget determinations. The list is daunting, yet without all of these data, we tend to oversimplify our models, with the result that our predictions are not accurate.