Owing to the weak observability of the azimuth misalignment angle, alignment accuracy and time are always the contradictory issues in the initial alignment process of Strapdown Inertial Navigation System ( SINS) , which requires a compromise between them. In this paper, a combined alignment mechanism is proposed to construct an observable and controllable system model, which can effectively achieve higher azimuth alignment accuracy during the fixed time period. First, the Reduced Order Kalman Filter ( ROKF) alignment algorithm was utilized to calculate the misalignment angles in parallel with the classical gyrocompass alignment algorithm. Then, the misalignment angles calculated by the gyrocompass alignment method were used to formulate the augmented measurement model with zero velocity models. Finally, the zero velocity model of the ROKF method was switched into the augmented measurement model when the azimuth misalignment angle of the gyrocompass alignment method was close to steady situation. The combined alignment method was analyzed reasonably by the observability and the mathematical deduction. The comparison results of the numerical simulation and the experimental data test showed that the combined method had good performance in terms of estimation accuracy and consistency of the alignment results.