Use of High-Performance Parallel Computing for Large-scale, Fine-resolution analysis of 3D Transient Hydraulic Tomography
EMSL Project ID
44616
Abstract
A new type of aquifer-based experiment called Hydraulic Tomography (HT) records the 3-dimensional pressure response within an aquifer (recorded at isolated intervals within wellbores using e.g., packers) in response to pumping from a discrete interval in another wellbore (also isolated using packers). Analysis of HT data consists of jointly inverting all pressure responses to obtain a detailed, 3D map of heterogeneity in aquifer properties--primarily hydraulic conductivity (K) -- using methods that are similar to medical imaging such as CAT scans. HT has been suggested as a valuable method for imaging field-scale aquifer heterogeneity for purposes of predicting contaminant transport and delineating 'fast flow' features. HT has also been analyzed extensively through numerical modeling and sandbox experiments. However, there are relatively few field studies in which hydraulic tomography has been implemented, and in the one case where field-scale 3D transient hydraulic tomography has been studied, a relatively coarse discretization and small modeling domain was used (Illman et al., 2009). In addition, this study presented mainly qualitative validation of the obtained K images. Working at the Boise Hydrogeophysical Research Site (BHRS), we have performed a series of short-duration (20 minute), 3D HT pumping experiments. Data from these tests show 3D variation in pressure responses as well as differing response magnitude at different wells, which is indicative of field-scale heterogeneity. Using a subset of the data, we have performed initial inversions which show qualitative agreement in K variability with other data sources-- primarily, high-resolution partially-penetrating slug tests performed earlier at the site. In this proposal, we plan to use the NW-ICE supercomputer cluster to extend our current modeling efforts by including a larger set of the field data in our inverse model and by increasing the resolution and accuracy of our existing numerical model. We are requesting an allocation of 20,000 node-hours which will be used to solve the imaging problem (which is readily and easily parallelizable) under a number of different test scenarios. Already, the numerical model of our field experiments has been created and tested on a smaller cluster, and we have likewise written inversion code in MATLAB that can be converted to FORTRAN. The model contains over 270,000 parameters to be estimated, consisting of hydraulic conductivity (K), specific storage (Ss) and specific yield (Sy) to be estimated at spatially distributed grid blocks, making it one of the larger hydrologic inverse problems ever solved. The imaging of hydrologic parameters, once completed, will represent a major step towards characterizing hydrology of the subsurface at the BHRS, one of the more well-studied aquifer systems in the United States. In addition, the high availability of other data sources from the BHRS (including slug tests, tracer tests, and geophysical tests e.g., electrical resistivity tomography (ERT) and ground-penetrating radar (GPR)) will allow us to quantitatively cross-validate our images of aquifer heterogeneity obtained via HT inversion. The prominence of geophysical data from the BHRS will also allow us to assess to what extent geophysical measurements provide information about hydrologic parameters at the BHRS and other similar unconsolidated, fluvial aquifers. Results of this research, and the preliminary results which are already being prepared for publication, will be presented at several hydrologic meetings (including AGU and NGWA), and will be published in a major, high-quality hydrologic journal.
Project Details
Project type
Exploratory Research
Start Date
2011-07-29
End Date
2012-07-29
Status
Closed
Released Data Link
Team
Principal Investigator
Related Publications
Cardiff MA, and W Barrash. 2011. "3-D Transient Hydraulic Tomography in Unconfined Aquifers with Fast Drainage Response." Water Resources Research 47:W12518. doi:10.1029/2010WR010367
Cardiff MA, T Bakhos, PK Kitanidis, and W Barrash. 2012. "Oscillatory Hydraulic Tomography: A New Concept for Aquifer Imaging and Long-Term Monitoring with Periodic Signals." , Pacific Northwest National Laboratory, Richland, WA. [Unpublished]
Cardiff MA, W Barrash, and PK Kitanidis. 2012. "A Field Proof-of-Concept of Aquifer Imaging Using 3-D Transient Hydraulic Tomography with Modular, Temporarily-Emplaced Equipment." Water Resources Research 48:05531. doi:10.1029/2011WR011704