Multivariate And Multiscale Analysis Of Spatiotemporal Patterns Of Urbanization And Urban Expansion Simulation

Urbanization has become one of the most high-profile phenomena in society.To fully understand the characteristics of spatiotemporal variations of urbanization and reasonably predict the spatial pattern of urban land are the primary concerns to promote regional sustainable development.In this paper,Zhejiang province,a typically rapid urbanization region on the southeast coast of China,was selected as a case study.Using remote sensing,GIS and data mining technology,we aim to analyze the spatiotemporal pattern of urbanization and its ecological response mechanism.And coupling the urban construction with ecological protection,a scientific and effective model is established to predict the spatial distribution of urban construction land in the future,which is of great practical significance to realize the sustainable development of urbanization under the background of the New Urbanization and Ecological Civilization Construction.The main contents of this paper are as follows:(1)From the perspective of multivariate and multiscale,this paper analyzed the evolution characteristics of the spatiotemporal pattern of urbanization in Zhejiang province from 1994 to 2015.In order to improve classification accuracy and the reliability of classification results,a nonparametric classifier for feature selection and parameter optimization was proposed by integrating the novel bat algorithm,genetic algorithm with support vector machine.Then the land use maps of Zhejiang province in 1994,2001,2008 and 2015 were obtained.Methods of exploratory analysis,spatial analysis and landscape pattern analysis were discussed to establish the multi-hybrid framework.The transfer scale,hot spot distribution,expansion mode and landscape pattern were used to explore the evolution characteristics of the spatiotemporal pattern of urbanization by using the framework.The driving forces analysis of construction land expansion was studied at the county and grid scale.Results showed that the topography and location conditions limited the possibility of construction land expansion,population and economic growth had a significant impact on construction land expansion.Urban land use pattern was affected by the coupling effect of natural geography,traffic location with social economy,which revealed the typical urbanization characteristic that the decrease of arable land and the increase of construction land.And the metropolitan areas with Hangzhou,Ningbo,Jinhua-Yiwu,Wenzhou-Taizhou as the core had formed.(2)Taking full account of the impact of human activities and land use change on the ecosystem,the response of ecological effect to the evolution of spatiotemporal pattern of urbanization was studied.?The land use change caused by the rapid growth of construction land had led to the decline of the ecosystem services value in Zhejiang province,with the total value reduced by 3.86%.The spatial distribution of ecosystem services value was imbalance.The ecosystem services value was negatively correlated with the intensity of nighttime lights and the proportion of construction land.?Depending on the exposure-sensitivity-adaptive capacity framework,coupling the nighttime lights image,pollutant emissions,weibo check-in points,change of ecosystem services value,natural environment,location conditions with ecological construction,the ecosystem vulnerability assessment system for Zhejiang province was established.Evaluation results showed that the overall vulnerability of Zhejiang province was relatively low,but the spatial pattern of vulnerability was of great heterogeneous,which showed a gradient descent from the urban built-up area to the surrounding area.The extremely and highly vulnerable areas were mainly distributed in the north plain of Zhejiang and the southeast coastal areas where human activities were frequent,while the mountainous and hilly areas were less vulnerable.(3)From the perspective of balancing the ecological protection and urban development,the urban expansion simulation model was studied.Taking Hangzhou,a typical area of urbanization and extremely vulnerable area,as a case study,the ecological source recognition system was established comprehensively considering ecological importance and demand intensity,the natural and artificial ecosystem characteristics.The evaluation results reflected the spatial distribution of the intensity of ecological land protection well.Using the minimum cumulative resistance model for ecological suitability zoning,result showed that suitable land for ecological protection mainly distributed in the southwestern,south and northwest of the city.On this basis,the model based on the ecological constraint integrating Markov chain,the deep belief network with the cellular automata was proposed,while the optimized deep belief network by hybrid bat algorithm was used to obtain the transition rules of cellular automata.The results showed that the model effectively overcame the prediction limitation of the traditional neural network,and the simulation accuracy(FoM index)and spatial pattern similarity had been improved by 3.70%and 10.11%respectively.The model considered the needs of ecological protection and urban development,which can provide scientific basis for urban planning under the guide of ecological priority.