We have proposed an improved rank-reduction method by slightly modifying the truncated singular value decomposition (TSVD) formula used in the traditional method. The resulted new rank reduction method can obtain a significantly cleaner 3D data recovery result compared with the traditional method based on one synthetic and one field data examples. We have described how we write the Matlab code for traditional multichannel singular spectrum analysis (MSSA) method, and how we make the slight modification to the MSSA code in order to tremendously improve the performance. This Matlab code package can be used in research and industrial applications for effectively and efficiently reconstructing the missing seismic traces due to different reasons.