Abstract: This paper present a new Greene approximation wide-angle parabolic equation (WAPE) radio prediction model based non-local boundary condition(NLBC). This model can be used to calculate radio propagation characteristics in complicated environment. Tridiagonal linear equations are achieved by the finite difference method (FDM), which not only quickly solve field distribution in the whole domain but also accurately model irregular terrain. The Greene approximation WAPE model is proposed to resolve a limited propagation angle problem, which increased propagation angle up to about 50°. This model greatly reduces the sizes of absorbing layer, which improve the computational efficiency of the PE method. In this paper, it is found that for an arbitrary phase error limit of 0.002, the Tappert, Claerbout and Greene approximations have acceptable propagation angle limits of approximately 20°, 35° and 45° respectively. The PE results are compared with the classical optical dual-ray model. The results show that the Greene approximation PE model has greater radio propagation angle and higher prediction accuracy, and get good absorbing effect of electromagnetic wave on upper boundary condition reflection. The Greene WAPE models can accurately predict radio propagation characteristics in complicated atmospheric environment.