URBAN EXPANSION SCENARIOS BASED ON ARTIFICIAL NEURAL NETWORK (CASE OF ERDENET CITY, MONGOLIA)

摘要

Due to accelerated urbanization since the 1950's. most of the Mongolian population, or about 68%. live in urban areas. The systematic understanding of urban expansion is a crucial clue for urban planning and sustainable land development. Therefore, in this paper, we used the Markov chain model and an Artificial neural network to simulate and predict current and future urban areas expansion in Erdenet city (third largest urban area of Mongolia and fourth copper mining industry in the world) vicinity. Clark Lab*s (Clark University) IDRISI software's Land change modeler module had been used for the urban expansion prediction. The Land change modeler module was developed by the Multilayer perceptron neural network with the Markov chain model. Multilayer perception neural network allows the integration of factors to interpret land change, while the Markov chain model with cellular automata restructuring of them spatially. This model has been used in our country for the first time in the case of Erdenet city. The main aims of the study are (ⅰ) land use and land cover change detection; (ⅱ) simulation of urban expansion in different scenarios; (ⅲ) provision of reference information for urban planning and land development. Urban expansion predicted 2046's trend based on a historical expansion of Erdenet city between 1996. 2006 and 2016, which were prepared according to input model requirements. Low and medium resolution satellite images are allowed cities and regions to regularly monitor their spatial and temporal dimensions and land-use changes. Landsat imageries (Landsat TM 5, Landsat ETM 7 and Landsat OLI 8) of 1996. 2006 and 2016 were used to derive land use map. In the urbanization process, many socioeconomic and physical factors and their current situation have a significant impact. Digital elevation model, slope, distance to road, population grow th rate, distance to economic centers, suitable lands, road network, possible area for urban transition were used as an explanatory factor map of urban land-use change. The model proposes three spatial alternatives for the future expansion of Erdenet city. These include spontaneous scenario, environment-protecting scenario, and resources-saving scenario. The impacts of variations are different, but for the entire period, urban expansion rates are likely to increase. The land use transition into urban areas has similar changes in environmental protection and resources-saving scenarios, while spontaneous expansion is significantly different from others. By spontaneous scenarios of 2046. open space and the unused areas will be gradually reduced by built-up areas. The land-use change in the north-east and northwest area is estimated to have the largest increase in comparison with other locations. The spatial pattern of the modeled scenarios will provide the stakeholders and planners with information on where and how will go urban expansion process by 2046 and will give important information for future development projects. Simulation perfonnance of the Markov chain model with an Artificial neural network model can be used improving the understanding of the urban expansion processes while also allowing for better planning of Erdenet city, as well as for other rapidly developing regions of Mongolia.