In order to conduct research at required spatial resolution, we propose a model fusion involving interlinked calculations of regional projections by the global dynamic model GLOBIOM (Global Biosphere Management Model) and robust dynamic downscaling model, based on cross-entropy principle, for deriving spatially resolved projections. The proposed procedure allows incorporating data from satellite images, statistics, expert opinions, as well as data from global land use models. In numerous case studies in China and Ukraine, the approach allowed to estimate local land use and land use change projections corresponding to real trends and expectations. The disaggregated data and projections were used in national models for planning sustainable land use and agricultural development.