A tomographic survey provides different coverages and perspectives on subsurface heterogeneity--incompletely overlapping information about the subsurface. Fusion of these pieces of information expands and enhances the capability of a conventional survey, provides cross-validation, and constrains inherently ill-posed field-scale inverse problems. In this study, we explore the possibility of using river stage variation for basin-scale subsurface tomographic surveys. Basin-scale tomography requires energy sources of great strengths; spatially and temporally varying natural stimuli are ideal energy sources for this purpose. Specifically, we use numerical models to simulate groundwater level changes in response to temporal and spatial variations of 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 the heterogeneous characteristics of the basin.Next, we apply the hydraulic tomography testing technique and analysis algorithm to synthetic fractured media. The application aims to explore the potential utility of the technique and the algorithm for characterizing fracture zone distribution and their connectivity. Results of this investigation show that using hydraulic tomography with a limited number of wells can map satisfactorily the fracture zone distribution and the general pattern of its connectivity although estimated hydraulic property fields are smooth. As the number of wells and monitoring ports increases, the fracture zone distribution and connectivity becomes more vivid and the estimated hydraulic properties approach the true values.Further we develop a new parameter identification method that allows for simultaneous inclusion of all observed hydrographs from hydraulic tomography to map aquifer heterogeneity. A procedure is then recommended to diagnose and denoise observed hydrographs. Subsequently, we introduce methods that exploit these processed hydrographs for estimating effective parameters, boundary conditions, and statistical spatial structures of heterogeneity, which are the required inputs for the new hydraulic tomography analysis method. This new method and the data processing procedure are tested in a synthetic aquifer and subsequently applied to a sand box experiment. The estimated parameter fields for the sand box experiment are validated by predicting the head distribution induced by an independent pumping test, which was not used in the hydraulic tomography analysis.