Characterization of flow processes in multi-scale porous system (nanopores to mesopores) in tight rocks, such as the shales, is challenging because of the coexistence of various flow regimes in the porous media. Although some methods based on dusty gas model (DGM) have been applied to determine the apparent gas permeability of shales (Javadpour 2009, Freeman et al. 2011, Sakhaee-Pour and Bryant 2012, Chen et al. 2015), they fail to describe gas flow process in nanopores in detail. In this paper, we present an innovative methodology for estimating apparent gas permeability of shales by coupling multiscale flow mechanisms. The Lattice Boltzmann Method (LBM) with effective viscosity and a general second-order boundary condition is used to analyze the various flow regimes involved in the single microchannel. The desirable agreement between the simulation results and that from the DSMC studies for the rarefied flow prompts the application of the derived correction factor for estimating permeability of shale gas reservoirs. In order to realize this, the porous medium is represented by a bundle of capillaries with diameters determined by mercury injection capillary pressure (MICP) curves. The porous flow is simulated by Darcy's law with derived correction factor; the surface diffusion of adsorption gas in kerogen pores is simulated based on Langmuir model and Fick's law. An extensive integration based on fractal dimension is performed to estimate the total flow rate and thereby the apparent permeability of typical shale samples. MICP and a transient pressure pulse technique are employed on 7 shale samples to obtain the pore size distribution and permeability. The result shows that the estimated gas permeability matches well with the measured permeability with a 20% variation, indicating that the physics based model presented in this paper is highly effective in predicting gas permeability of tight formations, such as the shales.