ESTIMATING HYDROGEOLOGIC PROPERTIES FROM TIME-LAPSE GRAVITY MEASUREMENTS: A FIELD STUDY OF AN UNCONFINED ALLUVIAL AQUIFER

A time-lapse gravity survey was conducted during a pumping test of an unconfined alluvial aquifer in the Great Plains near Fort Collins, CO. The purpose was to test the ability of time-lapse gravity data to constrain aquifer Specific Yield (Sy), Storage (S*), and Transmissivity (T). The aquifer consisted of ca. 9 m of poorly sorted sand, gravel, and clay alluvium. Drawdown was monitored in wells located 6.3, 15.3, 30.7, and 60.2 m from the pumping well during the 19 hr test, with maximum drawdown reaching 1.67 m in the monitoring well nearest the pumping well. Gravity measured at this well during the test decreases roughly exponentially with time. Estimates of Sy, S*, and T were obtained by forward modeling the time-lapse gravity change based on the assumption that the shape of the drawdown cone obey Neuman’s solution. For comparison, Sy, S*, and T were also estimated using a Neuman drawdown solution that directly fits the drawdown measured in the monitoring wells. The values for Sy, S*, and T obtained from forward modeling the drawdown data overlap the range of values obtained by modeling the time-lapse gravity data. Discrepancies are attributed primarily to the fact that the gravity field is sensitive to drawdown distributed over tens of meters, and hence is influenced by aquifer heterogeneity and anisotropy, whereas the drawdown model is based on measurements of water table depth only at the monitoring well locations. Sensitivity analysis shows that the time-lapse gravity data are, in general, relatively sensitive to T and less sensitive to Sy. The gravity data place a relatively well constrained upper bounds on S*, but the lower bound is poorly constrained. In aquifers with T ranging from 0 to 10-3, Sy is fairly tightly constrained by time-lapse gravity (to within +/- 0.05 for S*=0.0052) if S* is independently known. For aquifers with higher T, Sy is poorly constrained. Similar analysis shows that log(T) is fairly tightly constrained, to within +/- 0.25, over a wide range of S* if Sy is known. If T is known, Sy is constrained to within +/- 0.1 over a wide range of S*. S* is poorly constrained in all the analyses.