BEGIN:VCALENDAR
METHOD:PUBLISH
PRODID:-//Apple Computer\, Inc//iCal 1.0//EN
X-WR-CALNAME;VALUE=TEXT:USC
VERSION:2.0
BEGIN:VEVENT
DESCRIPTION:Speaker: Michael Cardiff, Boise State University
Talk Title: Novel Methods for Hydrogeophysical Joint Inversion and Data Integration
Abstract: The search for improved estimates of subsurface flow and transport parameters, and the expense and time associated with collecting hydrologic measurements, have both lead many hydrologists to consider the use of geophysical data for aquifer characterization.\n
Geophysical surveys, such as ground-penetrating radar (GPR), electrical tomography, and active seismic are often relatively cheap and fast to collect, when compared to hydrologic tests such as pumping tests and tracer injections. However, the key drawback of geophysical tests is that they are sensitive to geophysical parameters (e.g., electrical resistivity, seismic velocity, etc.) instead of the hydrologic parameters of interest. In this talk, I present two novel methods for the joint analysis of hydrologic and geophysical data when characterizing hydrologic systems.\n
In the first part of my presentation, I discuss the use of petrophysical transforms for converting geophysical parameters to hydrologic parameters. While petrophysical transforms are relatively easy to implement, the existence of non-unique petrophysical relations or multiple petrophysical relations can make the conversion to hydrologic parameters difficult. Using a Bayesian perspective, I derive a generalized maximum likelihood estimator that takes into account errors in both hydrologic and geophysical parameter estimates in order to estimate petrophysical relationships. The derived estimator is a generalization of so-called “Gaussian Mixture Models”, but with added flexibility. In terms of performance, the derived estimator is often capable of determining 1) The complexity of underlying petrophysical relations and 2) Whether multiple petrophysical relations are present.\n
The second part of my presentation discusses a novel inversion strategy for estimating boundaries between lithologic units (i.e.\n
facies) using either single datasets or combinations of hydrologic and geophysical data. By using a series of “level set functions”, I represent boundaries between facies that are allowed to iteratively deform and improve fit to both datasets. Both hydrologic and geophysical data are used to simultaneously drive boundary movement.\n
After presenting the theory and key equations, I will show performance on numerical experiments in addition to an application to a sandbox hydraulic tomography study.\n
Application of imaging and optimization methodologies to water resources systems is a rapidly growing and evolving field, with many opportunities for future research in both field, theoretical, and numerical methods. At the end of my talk, I will discuss some promising areas for future research in hydrogeophysical data integration and inversion, as well as other areas in which computational and optimization methods can be used to improve environmental decision making.\n
Host: Sonny Astani Dept. of Civil and Environmental Engineering
SEQUENCE:5
DTSTART:20110322T140000
LOCATION:KAP 209
DTSTAMP:20110322T140000
SUMMARY:Novel Methods for Hydrogeophysical Joint Inversion and Data Integration
UID:EC9439B1-FF65-11D6-9973-003065F99D04
DTEND:20110322T150000
END:VEVENT
END:VCALENDAR