USGS (Water Resources of the United States)
USGS National Research Program (NRP)
USGS National Research Program, Central Region (NRP, CR)
Contact Info.
Upcoming Events
Curriculum Vitae
Previous Return to Curriculum Vitae  


5. Develop more effective ways to constrain ground-water models to enhance their accuracy.

This involves

(a) include new kinds of field data (for example, flux and concentration data) as observations in the regression;
(b) understand the relevance of measurements, such as recharge measurements and laboratory and flowmeter hydraulic-conductivity measurements, to values appropriate for modeling; and
(c) use additional information about, for example, geologic structure, subsurface stress and strain fields, and so on, to constrain model development. I have made contributions or used my understanding of sensitivity analysis and regression methods to facilitate contributions to each of these areas. My specific contributions and the relevant publications are as follows.
For (a), as related to demonstrating the essential role of flux data in ground-water flow model calibration, Hill (1992), Poeter and Hill (1996, 1997), Hill and others (1998); as related to using concentration data to infer advective travel ’observations’ to calibrate ground-water flow models, Anderman and others (1996), Anderman and Hill (1997, 1999); as related to using concentrations directly to calibrate ground-water flow and transport models, Christiansen and others (1995), Barlebo and others (1996, 1998, in review), Anderman and Hill (1999).

For (b), my contribution has been in carefully testing the relation between measured and in situ values of areal recharge and hydraulic conductivity at a heterogeneous site using the extensively sampled field experiment at the MADE (MAcroDispersion Experiment) site at the Columbus Air Force Base in Mississippi. The site was chosen for study because it was significantly more heterogeneous than other experimental sites such as Cape Cod. It was, however, the heterogeneity that proved so challenging to the stochastic methods favored by the researchers first involved at the site. Attempts to use the measured hydraulic conductivities in flow and transport models, both published and unpublished, have consistently failed to reproduce even major characteristics of plume migration without invoking additional mechanisms such as dual porosity.

We took a different approach, one that would be characterized as deterministic by the stochastic ground-water community (though our ’deterministic’ approach is built on solid statistical theory). I worked with Heidi Christensen Barlebo (Danish Technical University; now of GEUS) to show that a traditional advective-dispersive model, when combined with reasonable, large-scale variations in hydraulic conductivity identified using nonlinear regression, was able to explain more of the measured transport than expected (Barlebo and others, 1996, in review; Heidi Christiansen Barlebo’s Ph.D. dissertation, 2000). The work shows that the measurements being used to determine recharge were too high by about a factor of two, and that the flowmeter measurements of hydraulic conductivity were consistently lower than the in-situ values and were too noisy to detect distinct spatial contrasts in hydraulic conductivity. Knowledge of these difficulties can help guide data interpretation and future research on field methods.

For (c), I worked with my co-authors from the Yucca Mountain project to use their comprehensive and well-organized hydrogeologic data to formulate and test hypothetical models using sensitivity analysis and nonlinear regression (D’Agnese and others, 1996, 1998, 1999).

Previous Return to Curriculum Vitae
Last Modified: April 14, 2008