468 / 2019-02-28 10:03:12

Refined Feasibility Prediction on Urban Land Development: A Case Study on Liangjiang New Area

Liangjiang New Area,fine-grained scale,urban land development feasibility,spatial autocorrelation,autologistic regression

全体主题 > Urban Planning & Modeling

Chinese urban development has entered a transitional period. With the cooling of the real estate market and the establishment of the strictest land management system, the concept of urban construction began to change. The fact that only a limited amount of land is available for development motivates city governments to think about a solution to sell land more savvy, thereby reducing the idleness and waste of land. Therefore it is increasingly important to scientifically and rationally select the areas suitable for development. This paper selects Chongqing Liangjiang New Area as the research area, and predicts the development feasibility of the new area from the fine-grained scale.
The research selected 100-meter grid as the basic researching unit. Based on the grid, the amount of land developed in 2015 and 2016 was calculated. According to a development threshold, all grid units were divided to developed grid and the undeveloped grid, and were defined by 1 and 0 respectively.
We also collected some covariates as driving factors for prediction, the number and service area of public facilities and infrastructure and related agent variables were included. Considering the concept of spatial correlation in the first law of geography, the ordinary model is modified by incorporating an autocovariate variable to reflect any spatial autocorrelation between grid units (Augustin, et al, 1996; Besag J, 1974). We applied the autologistic regression model (Bo Y C, et al, 2014) and the cellular automaton (Li X & Ye G A, 2001) respectively to predict the urban land development feasibility in the Liangjiang New Area.
Results showed that the prediction model based on spatial autocorrelation can provide better support for urban land development prediction under refined spatial scale. In particular, the autologistic regression prediction model performed better with higher prediction accuracy than the land use development simulation predictions supported by the cellular automata. It revealed each unit’s urban land development feasibility. Under the guidance of the master plan, the modified model can provide support for development decisions at time and space refinement scales.