The differences of classification systems between HLZ and land cover are compared in this paper. Based on the conversion probability of each HLZ type to every land cover type, a marginal conversion raster model of land cover is created. This model wins through the difficulty of converting between the two classification systems. After the raster data of HLZ spatial distribution on the basis of the climate change data of HadCM3 A1FI, A2a, and B2a, respectively, being simulated, the marginal conversion model were run with these data, so the scenario data of land cover change of China in 2039, 2069 and 2099 are simulated. Moreover, some landscape index models, as the land cover diversity index and patch connectivity index, are introduced to analyze the landscape changes of land cover in China in the future. The mean center model is introduced to construct a model to calculate the shift distance and direction of the mean center of each land cover type, thus to study the shift trend of the mean centers of the future land cover in China. These results show that: with temperature continuously rising and precipitation keeping increasing, the areas of cultivated land, grassland, wetland, water body, ice and snow cover, and bare rock would decrease; the areas of forestland, construction land, desertified land would increase; especially, the increased rate of forestland is high (2.34% per ten years) and the decreased rate of bare rock is high (2.38% per ten years); the desert area would reduce partly; the land cover diversity and patch connectivity would continuously decrease and increase, respectively.