To obtain land cover change information related to forest and update the historical land cover map, a change detection and rule classification method is proposed. This paper takes Zalantun City, Inner Mongolia Autonomous Region, as the study area and uses the Multivariate Alteration Detection method (MAD) to extract MAD differences; The MAD synthetic difference feature is calculated using each MAD difference feature to remove pseudo changes; The change detection result and the historical land cover type map are combined, and thus, the classification rules of the major land type change directions are analyzed and established. Then, the land cover information of the changed pixels in the later phase is obtained, thereby the land cover type map is updated. In the four years, the net growth area of forest land in Zalantun City was 42694hm 2 and the growth rate was 2.54%. The major land cover change in Zalantun City was grassland-to-forest. Some small areas of mutual conversion between other land types were also detected. This paper develops a remote sensing method for updating land cover type maps based on change detection. The accuracy of change detection is 77.69% and the classification accuracy of land cover change types is 81.25%.