This dissertation focuses on interpreting the spatial variations of seismic amplitude data as a function of rock properties for the Haynesville Shale. To achieve this goal, I investigate the relationships between the rock properties and elastic properties, and calibrate rock-physics models by constraining both P- and S-wave velocities from well log data. I build a workflow to estimate the rock properties along with uncertainties from the P- and S-wave information. I correlate the estimated rock properties with the seismic amplitude data quantitatively. The rock properties, such as porosity, pore shape and composition, provide very useful information in determining locations with relatively high porosities and large fractions of brittle components favorable for hydraulic fracturing. Here the brittle components will have the fractures remain opened for longer time than the other components. Porosity helps to determine gas capacity and the estimated ultimate recovery (EUR); composition contributes to understand the brittle/ductile strength of shales, and pore shape provides additional information to determine the brittle/ductile strength of the shale. I use effective medium models to constrain P- and S-wave information. The rock-physics model includes an isotropic and an anisotropic effective medium model. The isotropic effective medium model provides a porous rock matrix with multiple mineral phases and pores with different aspect ratios. The anisotropic effective medium model provides frequency- and pore-pressure-dependent anisotropy. I estimate the rock properties with uncertainties using grid searching, conditioned by the calibrated rock-physics models. At well locations, I use the sonic log as input in the rock-physics models. At areas away from the well locations, I use the prestack seismic inverted P- and S-impedances as input in the rock-physics models. The estimated rock properties are correlated with the seismic amplitude data and help to interpret the spatial variations observed from seismic data. I check the accuracy of the estimated rock properties by comparing the elastic properties from seismic inversion and the ones derived from estimated rock properties. Furthermore, I link the estimated rock properties to the microstructure images and interpret the modeling results using observations from microstructure images. The characterization contributes to understand what causes the seismic amplitude variations for the Haynesville Shale. The same seismic reservoir characterization procedure could be applied to other unconventional gas shales. / text