When predicting unsteady flow and pressure fields around a structure in a turbulent boundary layer by Large Eddy Simulation (LES), velocity fluctuations of turbulence (inflow turbulence), which reproduce statistical characteristics of the turbulent boundary layer, must be given at the inflow boundary. However, research has just started on development of a method for generating inflow turbulence that satisfies the prescribed turbulence statistics, and many issues still remain to be resolved. In our previous study, we proposed a method for generating inflow turbulence and confirmed its applicability by LES of an isotropic turbulence. In this study, the generation method was applied to a turbulent boundary layer developed over a flat plate, and the reproducibility of turbulence statistics predicted by LES computation was examined. Statistical characteristics of a turbulent boundary layer developed over a flat plate were investigated by a wind tunnel test for modeling the cross-spectral density matrix for use as targets of inflow turbulence generation for LES computation. Furthermore, we investigated how the degree of correspondence of the cross-spectral density matrix of the generated inflow turbulence with the target cross-spectral density matrix estimated by the wind tunnel test influenced the LES results for the turbulent boundary layer. The results of this study confirmed that the reproduction of cross-spectra of the normal components of the inflow turbulence generation is very important in reproducing power spectra, spatial correlation and turbulence statistics of wind velocity in LES.