Grantee Research Project Results
Final Report: A Method for Characterizing the Vadose Zone and Monitoring Solute Transport
EPA Grant Number: R827114Title: A Method for Characterizing the Vadose Zone and Monitoring Solute Transport
Investigators: Yeh, Tian-Chyi J. , Warrick, Arthur W.
Institution: University of Arizona
EPA Project Officer: Aja, Hayley
Project Period: October 1, 1998 through September 30, 2001 (Extended to September 30, 2002)
Project Amount: $359,955
RFA: Exploratory Research - Physics (1998) RFA Text | Recipients Lists
Research Category: Land and Waste Management , Air , Safer Chemicals
Objective:
Spatial variability of hydraulic properties of the vadose zone is an important factor that governs the spatial and temporal distributions of contaminants. As a consequence, this variability influences the effectiveness of our cleanup strategies, including the use of pump and treat approaches, groundwater circulation wells, air sparging, and bioremediation. Therefore, identifying the spatial distribution of hydraulic properties becomes a necessary step towards understanding the fate and transport of contaminants and their cleanup in the vadose zone. Direct measurements of these properties, however, often are time consuming, technically formidable, and costly, so detailed mapping of the vadose zone becomes an impractical option. Subsequently, our cleanup operations are hampered by our imprecise knowledge of the vadose zone.
The objective of this research project was to develop a cost-effective algorithmic technique for identifying distributions of heterogeneity and contaminants in the vadose zone. This technique would take advantage of the availability of frequently collected field data (soil-water pressure and moisture content) to estimate the spatial distribution of difficult-to-measure unsaturated hydraulic properties. In addition, it would be designed to take advantage of the ability of electrical resistivity tomography (ERT) to sense the change of moisture content distribution over a large volume of geological media. The success of this project would obviate the need for extensive and costly destructive sampling of media and maximize the information obtained from limited field sampling campaigns. Importantly, it would allow rapid and economic identification of spatial distributions of heterogeneity and contaminants in vadose zones to be so effective that remediation operations could be designed and the progress of such efforts could be monitored in a more efficient manner.
Summary/Accomplishments (Outputs/Outcomes):
Our project was based on the hypothesis that information about moisture content and pressure distributions would enhance our ability to delineate heterogeneity in the vadose zone. In addition, the arrival time of wetting or drying fronts at some depths would allow us to determine hysteretic characteristics of unsaturated hydraulic properties from field studies. We also postulated that the electrical resistivity of unsaturated porous media would have hysteresis effects and the variability in the resistivity would increase as moisture content decreases and becomes correlated in space. This information would help in the inversion of ERT surveys to obtain geologically and hydrologically reasonable results.
In conjunction with the above concepts, an iterative stochastic approach was used to estimate conditional effective unsaturated hydraulic parameters using soil-water pressure head, degree of saturation, and concentration during transient flow and transport processes in heterogeneous vadose zones. The successive linear estimator (SLE) is central to this iterative approach, with a linear estimator dependent on covariances of unsaturated hydraulic parameters, soil-water pressure head, degree of saturation, and concentration, as well as the cross-variances of these parameters. Successive improvement in the linear estimator is achieved by solving the governing flow and transport equations and updating the residual covariances and cross-covariance functions in an iterative manner. Using this iterative approach, the nonlinear relationships between unsaturated hydraulic conductivity parameters, degree of saturation, head, and concentration can be incorporated in the estimation. Consequently, more detailed spatial distributions can be obtained than currently are available from field observations or specified data in computational studies using this technique. A theoretical analysis of the SLE algorithm has been published (Vargas-Guzman and Yeh, 2002), which verified the approach and established that for both deterministic and stochastic inverse problems, the SLE algorithm converges in the same way that the classical Newton-Raphson algorithm is applied to forward nonlinear problems.
The inverse model that has been developed as part of this research project provides a means of estimating unsaturated hydraulic parameters during transient, three-dimensional flow. Innovative aspects include the use of the van Genuchten-Mualem unsaturated hydraulic conductivity and soil moisture retention functions and the incorporation of time-dependent pressure data in the parameter estimation. The SLE approach used also permits a similar conditioning on time series of data obtained from several sampling locations. The key to the successful application of SLE to time-dependent data is the residual parameter covariance structures that "remember" the conditioning effect from previous times. This also has the major benefit of increasing computational efficacy without sacrificing time-dependent moisture correlation information.
Detailed computational studies with this model (Hughson and Yeh, 2000) indicated that pressure and moisture content data sets from fixed locations collected at later times during an infiltration event, or during steady state flow, provided better estimates of the hydrological parameters of the vadose zone than data from very early times. In addition, these studies indicated that sequential conditioning on multiple sets of pressure data obtained at different times would yield better results than conditioning using only a single time. The computational studies indicated that anomalous results from our inverse methodology could be produced and were dependent on various combinations of known data and the associated boundary conditions. This finding suggests that complete characterization of heterogeneity requires extensive sets of pressure and moisture content data covering the entire physical domain considered for our inverse model. Translated to field studies, this requirement would necessitate a costly and impractical sampling strategy.
Additional questions regarding the utility of our inverse methodology and data requirements were addressed through further computational studies (Yeh and Liu, 2000) and sandbox experiments under well-controlled laboratory conditions (Liu, Yeh, and Gardiner, 2002). Simulations of pumping tests in two-dimensional, heterogeneous aquifers were used to investigate optimal sampling schemes in terms of selection of well spacing, pumping, and monitoring locations. The effects of measurement errors and uncertainties in statistical parameters required for the inverse model also were included in these investigations. Results of these numerical experiments showed that the hydraulic tomography resulting from the application of the inverse methodology would be most effective if the horizontal separation distance between wells is set to one-half the value of the horizontal correlation scale. In addition, the vertical interval between two monitoring locations should be no more than one-half of the vertical correlation scale. Of particular importance for practical applications is the optimal number of pumping locations, which is determined by the ratio of the aquifer depth to the vertical correlation scale. An increase in the number of pumping locations above this ratio provides no additional information. The pumping rate had no effect on the estimate that was realized.
These numerical experiments led to the conclusion that uncertainty in the input variance for the inverse model had no influence on the final estimate. Uncertainty in correlation scales also had no significant impact on the estimates unless the correlation scales were unrealistically underestimated or overestimated. Increasing secondary information, such as pressure head data, at different spatial locations can greatly reduce the effects caused by a poor knowledge of the correlation structure. Similar conclusions were supported by the extension of the numerical experiments to a three-dimensional, heterogeneous aquifer.
Sandbox experiments were used to validate the conclusions arising from the numerical experiments and to evaluate the performance of our inverse methodology under physically realistic conditions. The first sandbox was packed with layered sands to represent a stratified aquifer, and the second was packed to yield discontinuous bodies of different shapes and dimensions to represent a more complex heterogeneous aquifer. For both sandbox experiments, the inverse model was able to reproduce the major heterogeneous patterns, in spite of measurement errors and uncertainties associated with the pressure head/discharge data sets and other required input parameters. One significant finding from these experiments was that hydraulic tomography does not improve the conductivity estimate if an abundant number of head measurements is available, particularly in the case of stratified media. Hydraulic tomography, however, is useful and effective when the number of pressure head measurements is limited for heterogeneous aquifers that possess a highly discontinuous and nonuniform structure. The sandbox studies also demonstrated the utility of our inverse model in troubleshooting sampling designs. Results from a parallel set of numerical experiments for the first sandbox led to significant improvements in the design of our second sandbox. This supports our contention that our inverse model will have great utility in field applications in terms of inverse analyses to obtain reasonable descriptions of heterogeneous aquifers and sampling program designs.
A major implication arising from our numerical and laboratory experiments is that an accurate description of the entire three-dimensional domain associated with a heterogeneous aquifer requires a large number of sampling locations distributed throughout the domain of interest. Failure to have a sufficiently dense sampling grid may yield anomalous estimates in regions that are spatially far-removed from a less than sufficient set of sampling locations. Such a constraint, however, places an unreasonable cost on any sampling program, making it impractical to implement. As a consequence, we explored the possibilities of incorporating geophysical measurements closely tied to moisture content, such as ERT, to provide the secondary data required for our inverse model (Yeh and Simunek, 2002) because geophysical surveys have the potential to collect a vast amount of information pertinent to the hydrological inversion process. In our resulting stochastic fusion approach, information required for inversions of unsaturated hydraulic tomography, and that required for ERT, is used in a coupled and iterative fashion to yield a more detailed image of the hydraulic heterogeneity. More specifically, during an unsaturated hydraulic tomography experiment, several water infiltration tests at different locations are conducted sequentially. Point measurements of pressure head and moisture content are taken while an ERT is deployed to monitor water movement. Mean hydraulic parameters then are used in a forward simulation of the unsaturated hydraulic tomography test to yield the mean and covariance structure of the resistivity and moisture content fields. This information then is used in the ERT inversion to estimate changes in resistivity and moisture content, as well as the moisture content and conditional moments. This updated information serves as the new input to the hydraulic tomography inversion model to yield improved estimates of the distributions of the unsaturated hydraulic parameters and their conditional moments. The new set of hydraulic parameters then is used in a forward simulation to generate a new set of input parameters for the ERT inversion. This process continues in an interative manner until no further improvements in hydraulic heterogeneity, moisture content, and resistivity estimates are gained. The perceived advantages of this new approach, provided by the ERT, include information regarding moisture content at locations where no samples were available. The sampling data from the hydraulic tomography test provide additional constraints for inversion of the ERT data. Preliminary results from numerical experiments using this new stochastic information technology suggest that it is a very promising tool for effectively characterizing heterogeneity, monitoring processes in the vadose zone, and quantifying the uncertainties associated with vadose characterization and monitoring.
Journal Articles on this Report : 1 Displayed | Download in RIS Format
Other project views: | All 5 publications | 5 publications in selected types | All 5 journal articles |
---|
Type | Citation | ||
---|---|---|---|
|
Yeh T, Simunek J. Stochastic Fusion of Information for Characterizing and Monitoring the Vadose Zone. VADOSE ZONE JOURNAL 2002;1(2):207-221 |
R827114 (Final) |
Exit |
Supplemental Keywords:
hydrogeophysical joint inversion, vadose zone, risk assessment, risk management, ecosystem protection/environmental exposure and risk, waste, water, aquatic ecosystem restoration, ecology and ecosystems, engineering, chemistry and physics, groundwater remediation, hydrology, restoration, aquatic ecosystems, electrical resistivity, tomography, environmental rehabilitation, groundwater contamination, groundwater pollution, heterogeneity, hydraulic properties, moisture content, pressure distribution, remediation, solute transport monitoring., Scientific Discipline, Water, Waste, Ecosystem Protection/Environmental Exposure & Risk, Hydrology, Physics, Restoration, Ecology and Ecosystems, Aquatic Ecosystem Restoration, Engineering, Chemistry, & Physics, Groundwater remediation, hydraulic properties, sloute transport monitoring, moisture content, solute transport monitoring, remediation, aquatic ecosystems, environmental rehabilitation, pressure distribution, groundwater contamination, electrical resistivity tomography, vadose zone, groundwater pollutionRelevant Websites:
http://www.hwr.arizona.edu/yeh Exit
http://tian.hwr.arizona.edu/yeh/research.html Exit
Progress and Final Reports:
Original AbstractThe perspectives, information and conclusions conveyed in research project abstracts, progress reports, final reports, journal abstracts and journal publications convey the viewpoints of the principal investigator and may not represent the views and policies of ORD and EPA. Conclusions drawn by the principal investigators have not been reviewed by the Agency.