[1] Data from tomographic surveys make an inverse problem better posed in comparison to the data from a single excitation source. A tomographic survey provides different coverages and perspectives of subsurface heterogeneity: nonfully redundant information of the subsurface. Fusion of these pieces of information expands and enhances the capability of a conventional survey, provides cross validation of inverse solutions, and constrains inherently ill posed field-scale inverse problems. Basin-scale tomography requires energy sources of great strengths. Spatially and temporally varying natural stimuli are ideal energy sources for this purpose. In this study, we explore the possibility of using river stage variations for basin-scale subsurface tomographic surveys. Specifically, we use numerical models to simulate groundwater level changes in response to temporal and spatial variations of the river stage in a hypothetical groundwater basin. We then exploit the relation between temporal and spatial variations of well hydrographs and river stage to image subsurface heterogeneity of the basin. Results of the numerical exercises are encouraging and provide insights into the proposed river stage tomography. Using naturally recurrent stimuli such as river stage variations for characterizing groundwater basins could be the future of geohydrology. However, it calls for implementation of sensor networks that provide long-term and spatially distributed monitoring of excitation as well as response signals on the land surface and in the subsurface.

No abstract

[1] A new analytic element solution has been derived for steady two-dimensional groundwater flow through an aquifer that contains an arbitrary number of elliptical inhomogeneities. The hydraulic conductivity of each inhomogeneity is homogeneous and differs from the conductivity of the homogeneous background. In addition to elliptical inhomogeneities, other elements (such as wells and line sinks) may be present. The method is based on a separable form of the solution for Laplace's differential equation in elliptical coordinates. The piezometric head and the stream function, expressed as continuous spatial functions (as components of the complex potential), may be obtained up to machine accuracy regardless of the shape, size, orientation, and conductivity of the elliptical inhomogeneities. Components of the discharge vector are expressed in a similar manner, using a complex discharge function. Problems with 10,000 or more inhomogeneities can be solved using parallel computing on distributed memory supercomputer clusters. Two examples are included to demonstrate the precision and capabilities of the method. The second example is used to perform a preliminary study of contaminant transport in a highly heterogeneous formation of lognormal conductivity distribution. The results of the transport study are compared with recent theoretical and numerical results that are based on circular inhomogeneities. The new results with elliptical inhomogeneities confirm the findings based on circular inhomogeneities, including a long dispersion setting time and a zero value of the asymptotic transverse dispersivity.

No abstract

No abstract