Can a Groundwater Flow Model BE Validated? Encouraging, Positive Evidence from Laboratory Hydraulic Tomography Experiments
Abstract
Illman et al. [2007, 2008] previously showed that steady state hydraulic tomography [Yeh and Liu, 2000] was a promising technique in imaging subsurface heterogeneity patterns of hydraulic conductivity (or K tomograms) through laboratory sandbox studies. Liu et al. [2007] did the same but used transient hydraulic tomography [Zhu and Yeh, 2005] to image hydraulic conductivity and specific storage heterogeneity patterns (or K and Ss tomograms). In these studies, different methods and data collected at different scales were utilized to validate the tomograms. Among the various methods for tomogram validation, both Illman et al. [2007, 2008] and Liu et al. [2007] found that the best method was to simulate an independent cross-hole pumping test not used in the construction of the tomograms. In particular, Illman et al. [2007, 2008] were successful in predicting the steady state head response of an independent cross-hole pumping test using the K tomogram and Liu et al. [2007] were successful in predicting the transient head response using the K and Ss tomograms. These results were encouraging in that an independently conducted cross-hole test can be predicted quite accurately if the heterogeneity pattern and forcing functions were all known. However, the experimental findings may be construed to be weak because there was only one independent test from the experiments conducted by us. In this paper, we conduct additional laboratory sandbox experiments of cross-hole pumping tests in a different sandbox aquifer. The heterogeneous aquifer was generated through the cyclic flux of sediment- laden water. In total, 24 cross-hole pumping tests were conducted at various locations within the sandbox of which 8 were used for steady state hydraulic tomography [Yeh and Liu, 2000]. This left 16 additional cross- hole pumping tests for the validation of the K tomogram. Forward simulations of the 16 independent tests showed that the steady state head distribution can be simulated very accurately with the K tomogram but not as well with the geometric mean of K obtained from cores, slug, single-hole and cross-hole tests as well as the effective K within the sandbox. The effective K was determined directly (and not through upscaling formulae) by conducting a flow-through experiment through the entire sandbox aquifer. Results provide encouraging, positive evidence that groundwater models could be validated if the geologic medium is characterized sufficiently through hydraulic tomography and if the forcing functions are specified correctly. This study suggests that more effort should go into the accurate characterization of the subsurface if the objective is to better predict subsurface head distributions, which then should yield better predictions of contaminant transport.
- Publication:
-
AGU Fall Meeting Abstracts
- Pub Date:
- December 2008
- Bibcode:
- 2008AGUFM.H33H..04I
- Keywords:
-
- 1829 Groundwater hydrology;
- 1846 Model calibration (3333);
- 1849 Numerical approximations and analysis;
- 1873 Uncertainty assessment (3275);
- 3265 Stochastic processes (3235;
- 4468;
- 4475;
- 7857)