Shale gas development has more than 3 decades of history and remains one of the hottest topics in the petroleum industry. Shale gas development in China is underway. Our study focuses on an emerging marine shale gas reservoir in southern China, with its huge reserves that have attracted strong attention. The first part of this dissertation is the petrophysical characterization, which is an important step for a new shale gas play to better understand the geology of the formation, and it provides vital data to optimize a production plan and stimulation design. A systematic petrophysical study was conducted for the marine shale gas reservoir by conducting a series of 6 parallel experiments for 12 groups of samples to measure the total organic content (TOC), vitrinite reflectance (Ro), porosity, permeability, mineralogy, and gas content. Second, the extra-low porosity and permeability of shale formations complicate the mechanisms of shale gas storage and flow. Understanding the microstructure is significant for evaluating a new shale gas play toward accurate reserve estimation and recovery prediction. Both physical measurement (nitrogen adsorption experiment) and visualization technology (Scanning Electron Microscope) were used to characterize the nanopore structure of the Longmaxi Shale. Isotherms were obtained from adsorption experiments, and specific surface area and pore size distribution were calculated from the experimental results. Combining with the TOC, gas content, and mineralogy of the Longmaxi Shale, the significance and the controlling factors of the specific surface area and the nanopore volume were discussed. In addition, various types of porosity and several microfractures were observed from SEM images. Third, preliminary interpretation of the imaging logs revealed natural fractures in the formation that can significantly affect the production performance of shale gas wells since preexisting natural fractures will influence hydraulic fracture propagation. Thus, numerical simulation was conducted focusing on the interaction between hydraulically induced fractures and preexisting natural fractures. A hydraulic fracturing model considering the in-situ stress response to turbulent flow process was developed and validated with regression tests of a bi-wing hydraulic fracture model. Field-scale simulation results indicate that our model is capable of capturing the interactions between hydraulic fractures and the preexisting natural fractures defined by the initial fracture maps. Finally, a new model was built to model the actual network of hydraulic and preexisting fractures from geological interpretations and microseismic mapping results. The discrete fracture modeling (DFM) approach was applied to represent each fracture individually and explicitly. The near-well effects were modeled in detail by refining the unstructured 3D grid to the point where we fully resolve stimulated fractures. Simulations of the detailed model of an actual shale gas reservoir considered various mechanisms including adsorption/desorption, matrix/fracture transfer, and non-Darcy effects. Furthermore, the dissertation illustrates upscaling from the discrete fracture model to a coarse continuum model using multiple subregion (MSR), and the high degree of accuracy provided by this technique is demonstrated by comparing the solution of the upscaled model with the corresponding fine-grid solution for a synthetic case.