A novel model for stripe noise reduction was proposed by combining moment matching with variational method, based on analyzing the characteristics of stripe noises in moderate resolution imaging spectroradiometer (MODIS) data. Firstly, multi-line and wide-line stripes were denoised by the moment matching method. Secondly, the stripes left in the result image were located by the improved stripe detection algorithm. Finally, based on the variational destrping model, Split Bregman iteration was adopted to reduce the strip noises. Experimental results show that the proposed algorithm can remove stripes in MODIS data and preserve the details of the image effectively. Compared with the destriping models including histogram matching, moment matching, low-pass filter and unidirectional variational destriping, the proposed method obtained more ideal results.