Моделювання локальних систем землекористування в умовах глобальних змін клімату

Abstract

The interdependencies among land use systems at national and international levels motivate the development of global land use models facilitating the analysis of the trends of plausible future land use under the conditions of increasing population and climate change for environmental and food security purposes. Computational complexity of such models limits the land use projections to aggregate levels which give no clue regarding the potentially critical local heterogeneities. Improving these projections at fine resolutions requires new methods of systems analysis for integrating land use models at different scales. For that purpose, we have proposed a dynamic cross-entropy based probabilistic downscaling model which facilitates to obtain future aggregate land use projections from global models (e.g. GLOBIOM) to finer resolutions. The proposed procedure allows incorporating data received from different sources, such as satellite images, statistics, and expert opinions, as well as data from global land use models. Using downscaling procedure, we estimate future impacts of global climate changes on the land use in Ukraine (on the rayon level) in accordance with the aggregated results of GLOBIOM modeling. They indicate some growth of pressure on land resources in Ukraine associated with the satisfaction of the increasing global demands for foods and biofuels. On the one side, the model forecasts a small growth of demand (0.2%) for arable land by the middle of the XXI century. At the national level, it doesn't pose any serious threats, but, on the case regional level, it can lead to certain ecological risks (in the oblasts with an extremely high share of arable land). On the other side, the model also predicts some growth of the demand for forests, including SRF, and pastures. These changes could have some positive effect by supporting safety and sustainable land use in Ukraine. Further investigations will be oriented to comparing the results of modeling based on different available maps of land cover and land use (GLC2000, MODIS2000, GLOBCOVER2000) and to estimating the land demand under different scenarios of agriculture improvements (technology, management etc.).