Abstract: The heterogeneous clutter results in degradation of clutter suppression performance of space-time adaptive processing (STAP). To solve this problem, a subspace method is proposed based on the radar parameters and the time-limited and band-limited features of the prolate spheroidal wave functions. The method can improve the clutter suppression performance in heterogeneous environment, reduce the computational complexity and demand for homogeneous samples. However, it is hard to estimate the precise clutter rank if the proposed method suffers from crab angle. Therefore, with the radar parameters and the detection environment geometry, a pre-process method is presented to eliminate the crab influence and improve the clutter suppression performance in heterogeneous environment. Finally, simulation attests its effectiveness.