Abstract: Insufficient samples result in estimation error of covariance matrix for Space-Time Adaptive Process (STAP), a merged method of multi-frequency space-time data is presented to improve the estimation accuracy. Based on the analysis of clutter character in different frequencies, this method modifies the power between frequencies by rebuilding covariance matrix, and merges the multi-frequency space-time data to improve the estimation accuracy. A preliminary result against an airborne simulation data demonstrates the effectiveness of the proposed method.