In this paper, the strip synthetic aperture radar (SAR) image of a composite vehicle-ground model is simulated by combining the electromagnetic scattering algorithm and radar imaging algorithm. A linear frequency modulated wave is incident on the composite model, which is partitioned into a mass of triangular patches. In the "stop-and-go" radar mode, for each patch, the amplitude of scattered echo in the frequency domain is solved by a hybrid method of physical optics (PO)-shooting and bouncing ray (SBR)-physical theory of diffraction (PTD). For the composite model, the total scattered echo in terms of range frequency and azimuth time is obtained by the vector superposition of echo on patches. Then the SAR image of the composite model is generated by the range-Doppler algorithm. In numerical simulations, both the electromagnetic scattering of a target by the SBR-PTD method and composite scattering by the PO-SBR-PTD method are validated and evaluated by comparing with the multilevel fast multipole method (MLFMM) in FEKO software. Moreover, the SAR image of the composite vehicle-ground model is also compared with the real image in Moving and Stationary Target Acquisition and Recognition database, which verifies the feasibility of the proposed method. SAR images of the composite model for different incident angles are also presented and analyzed.