Analysis Of Urban Landscape Pattern Based On Landsat Time Series Data

Although urban area just accounts for 3% of the global land area,urban population is higher than 50% of the world's total population.Urbanization has been the inexorable trend of the world.Rapid urbanization leads to dramatic changes in the type and spatial structure of the urban landscape,which account for the multiple and changeable urban ecological environment problems.Therefore,it's crucial to quantify and analyze the urban landscape evolution during the urbanization process for resolving urban ecological environment problems.In the urbanization process,the urban landscape is highly dynamic.Current research can quantify the changes of main landscape types in the study period,the change types are relatively simple,and the time resolution is coarse.They are unable to meet the demand for quantifying multiple landscape changes and complex change types,let alone character landscape evolution of the urbanization process.In order to meet the demand above,in recent years,quantifying urban landscape pattern based on remote sensing time series has become a trend of urban landscape pattern research and shown great potential.In this paper,taking Shenzhen as a case,all available Landsat remote sensing images from 1986 to 2017 as the main data source to build time series,Shenzhen road blocks as the analysis unit,we raised a novel quantitative model of urban landscape pattern oriented to high landscape dynamics and long-term time series owing to a comprehensive literature survey of existing methods of urban landscape pattern.Furthermore,in order to understand the urbanization process the result of this model was used to analyse the urban landscape evolution law during the urbanization process from three parts,landscape dynamics,landscape change space and landscape composition.The main conclusions of this paper are as follows:(1)we presented a quantitative model of annual urban landscape change and urban landscape dynamics in historical periods based on the long and dense time series of Landsat.In this model,we raised a novel idea of change detection of long-term series,which combined addition and subtraction of the change points and classification of the segment.Then,we divided our change detection process into three parts,the change existence detection for judging the possibility of change existence of the time series,the extraction of potential changes for getting the change information of all the possible change points,the change screening for removing pseudo-changes.Besides,we developed a new method to detect the existence of change and extract potential change points based on the optimal moving window.In this paper,the smallest moving window with significant change was used as the optimum window in the process of change existence detection process by accumulating residuals,which reduced the probability of the second kind of errors in statistical test and improved the detection accuracy of this process.This model developed in this paper could efficiently and correctly extract the change time,change types,change frequency of the blocks in Shenzhen.Considering different accuracy of different landscape dynamic level of blocks,the blocks were classified into three types,stable blocks without any landscape change,low frequency blocks with just one or two landscape change and high frequency change blocks with change frequency greater than 2.With high precision,stable blocks,low frequency change blocks and high frequency change blocks were 86.67%?90.00% and 73.33% respectively,this model will provide an effective way to quantifying the urban landscape dynamics of long-term time series and analyzing the landscape evolution law in the process of urbanization.(2)Based on the model we developed,this paper quantified and analyzed the landscape evolution of Shenzhen in the process of urbanization from landscape dynamics,landscape change space and landscape composition.Our results are described as below: ?1 Landscape dynamic levels with high dynamic of urban landscape varied from different blocks in Shenzhen.The highest dynamic of blocks in Shenzhen had changed seven times,and the area of the block dynamic changed from low to high was generally pyramid-shaped.In addition,cluster effect in part was significant.The new urban areas,which could be considered as urban expansion areas,were spatial aggregation with high dynamic level.Meanwhile,the old urban areas deemed as urban interior renewal areas were spatial aggregation with low dynamic level.?2 In the process of urbanization,the change space of urban landscape had become more and more universal and fragmented,and its annual change area tended to be stable.Moreover,the evolution process of landscape change space had obvious periodic,and land policy was an important driving force of urban landscape evolution.?3 The composition of urban landscape changed in stages during the urbanization process,and the laws of landscape evolution could reflect the process of land urbanization process perfectly.In the early stage of urbanization,the rapid expansion stage of the city only,the landscape changed dramatically because of rapid vegetation erosion from built-up.In the early middle stage,urban slow expansion and urban interior renewal which expansion was primary,renewal was the supplementary.The urbanization mode had been transformed into a multi-mode with encroaching on vegetation and wetland from built-up,and had a little of internal renewal of cities.In the late middle stage,urban interior renewal was superior to expansion.The types of landscape change related with urban internal renewal,built-up to bare land and bare land to built-up,account for a large proportion,and the landscape change in coastal areas had become an important dimension of landscape change in Shenzhen.In the latter stage,urban interior renewal only,vegetation,wetland and built-up were stable,and bare land was at a certain level.Meanwhile,the dominant landscape change types of Shenzhen were relative with urban interior renewal,which types were bare land to built-up,built-up to bare land and builtup to vegetation.?4 Policy had played as an important driving force of urban landscape evolution.