Detection, classification, and imaging of underground objects is of importance in geophysics and geochemistry of rock-fluid systems by providing information on geologic structure and providing a basis for interpretation of geochemical reactions in porous and fractured rock. A maximum likelihood multiparameter estimation approach to the problem will be used. An underground object to be imaged is parameterized so that the set of parameters is regarded as representative of the object itself. The Phase I goal is to estimate this set of parameters bycomputing their likelihood functions from a suite of offset/reverse vertical siesmic profiling (VSP) electromagnetic scattering experiments and determining the values at which they peak. The parameters can be discrete such as presence/absence of a certain object from a class of possible objects (the detection/classification problem) or continuous such as object location, orientation, dimensions, and contrast. Phase I will investigate the use of newly developed diffraction tomographic maximum likelihood estimation techniques for detection and in an offset/reverse VSP geometry. The method is based on an exact scattering and propagation model that incorporates the electromagnetic and/or acoustic parameters of the surrounding ground and the polarization information in the scattering data. The method is ideally suited for accommodation of data from acoustic scattering experiments for refinement of the results.