Abstract: In view of the poor performance of space-time adaptive processing(STAP) when clutter was range-dependent, the paper presented a compensation method using focusing matrix based on least square estimation. Firstly, clutter amplitude was estimated utilizing least square estimation, and clutter patches with greater amplitude were selected. Secondly, focusing matrix was evaluated by space-time steering vector of clutter patches, and a linear transformation was directly implemented on the training data using the focusing matrix. Thirdly, clutter-plus-noise covariance matrix(CCM) of the cell under text was estimated utilizing the transformed data and thus the self-adaptive weights were evaluated. The proposed method, in which non-uniform sampling is adopted, performs better than classical focusing matrix method. Lastly, the simulation tests its validity.