Stochastic fusion of information for characterizing and monitoring the vadose zone

• Main Authors: Yeh, T.-C. Jim, Simunek, Jirka, Van Genuchten, Martinus Th.
• Other Authors: Department of Hydrology & Water Resources, The University of Arizona
• Language: en_US
• Published: Department of Hydrology and Water Resources, University of Arizona (Tucson, AZ) 2002
• Online Access: http://hdl.handle.net/10150/615767; http://arizona.openrepository.com/arizona/handle/10150/615767

Inverse problems for vadose zone hydrological processes are often being perceived as ill - posed and intractable. Consequently, solutions to inverse problems are often subject to skepticism. In this paper, using examples, we elucidate difficulties associated with inverse problems and the prerequisites for such problems to be well -posed so that a unique solution exists. We subsequently explain the need of a stochastic conceptualization of the inverse problem and, in turn, the conditional- effective -parameter concept. This concept aims to resolve the ill -posed nature of inverse problems for the vadose zone, for which generally only sparse data are available. Next, the development of inverse methods for the vadose zone, based on a conditional -effective -parameter concept, is explored, including cokriging, the use of a successive linear estimator, and a sequential estimator. Their applications to the vadose zone inverse problems are subsequently examined, which include hydraulic /pneumatic and electrical resistivity tomography surveys, and hydraulic conductivity estimation using observed pressure heads, concentrations, and arrival times. Finally, a stochastic information fusion technology is presented that assimilates information from unsaturated hydraulic tomography and electrical resistivity tomography. This technology offers great promise to effectively characterize heterogeneity, to monitor processes in the vadose zone, and to quantify uncertainty associated with vadose zone characterization and monitoring.