Simulation Of Urban Spatial Based On Geographic Weighted Regression And Cellular Automata

Land resource is not the inexhaustible, In the process of urbanization construction, unlimited expansion of urban land at the expense of sacrificing farmland, natural landscape and ecological environment, this not only intensified economic and ecological balance, and against the concept of sustainable development we insisted. To avoid this kind of situation, and make the whole system and mechanism of all-round reasonable control, we need to make urban planning, making reasonable simulation for the development of city space.Firstly, according to soil utility status of Dalian, the landscape pattern and the land type transfers are analysed, and a quantitative understanding is produced. As result, with the increase of the year, human impact index is increasing, the areas of cultivated land and forests are decreasing, the area of construction land is rising with sacrificing cultivated land and forest land in ten years, the productivity of the whole system is falling, finally, the stability and self-maintenance ability are downgrade, so soil utility status should cause enough attention.Combining with residential and industrial characteristics that are closely related to indicators, factors like population, economic, social and so on were taken into account to simulate the prediction by regression analysis. According to wide variety and complexity of transportation and water conservancy facilities, and small correlation between the indicators and them, the improved gray Markov model was used for prediction. As result, it is more accurate to reflect the actual changes in land use by applying different model into different land usage. Therefore it avoids biased focus and provides a good scientific reference and frontier prospect.Finally, in order to avoid the phenomenon of "watermelon under your feet, but you don't care where it should go", letting the results of prediction clearer. Certain areas of Dalian city was selected as a case, combining GIS system to build geographical weighted regression thoughts into the cellular automata, realizing the urban spatial development of simulation, and the simulation result was compared with logistic regression model. It is indicated that in the land use pattern and influence factors in the process of modeling, there are different degree of space between autocorrelation, and in the spatial structure is not a single trend of distribution, it reflects space instability, making geographical weighted regression and cellular automata together. It not only embodies the space distribution features, and the simulation accuracy has also been improved.