| Urban green space is an important component of urban green space system and urban landscape,with significant ecological,economic,psychological and social functions.Due to the rapid urbanization,the spatial structures and patterns of urban green spaces have changed as well.As a fundamental landscape index for describing the structure of urban green space,the connectivity of green space is of great importance for understanding the interaction and characteristics of urban green space by quantifying the connectivity of urban green space.Moreover,the research on dynamic evolution of urban green space connectivity is not only aid to deeply realize spatial changes of urban green space,but also a very importance reference and instruction for optimizing urban green space and sustaining steady of urban green space system.Most of current studies analyze the connectivity of urban green space by using large patches from macroscopic aspect,and mainly focused on simple analysis and description of the urban green space patches or patterns,while neglecting information on dynamic changes of green space pattern.This study proposed a graph-based method for investigating the spatial pattern and dynamic evolution of urban green space connectivity.Spatially speaking,the method provides a direct neighborhood representation of the green space by using adjacent relation.This enable us to identify and quantify green space patterns,such as calculate the occurrences of pairs of neighborhood green space types from adjacency-event matrix.Apart from the spatial representation,the method tracks and records green space patches through time,allowing revelations relating to what happened to them individually between snapshots.The method was applied to the region within outer ring road of Shanghai,China based on the green space data from 2008 to 2016.The proposed method holds great potential for the greening management department to optimize greening management and construct control strategy of Greenbelts.The main research contents and conclusions are summarized as follows.(1)In order to construct connectivity graphs for small urban green patches,we selected 10 m as the distance threshold to find connective neighborhoods from the green space data based on the principle of graph theory.In the connectivity graphs,urban green patches were treated as nodes while the links between adjacent neighbors were treated as edges.Then,the Adjacency-event matrix is adopted to calculate the spatial arrangements of urban green patches from the connectivity graphs.Two graph-based indices,namely,Beta Index and Betweenness Centrality,were calculated to measure the overall connectivity of urban green space and identify the important green patches.The results shown that the residential green patches have the largest spatial connections with other green patches in the study area.The results also demonstrated that the connectivity of urban green space in the study area from 2008 to 2016 has increased gradually.We also found that the important green patches are always located near the green patches who have a high Betweenness Centrality,and the spatial distribution of these important green patches is getting more and more balanced during the study period.(2)To streamline the evolution of urban green patches,seven types of spatiotemporal relations,including born,die,expansion,contraction,splitting,merging and continuation,were proposed in our study.Based on these spatio-temporal relations,an evolution graph is constructed by treating the green patches as nodes and their corresponding spatio-temporal relations as edges.By treating the timeline as the Z-axis and the urban green space as the XY plane in the ArcScene software,the threedimensional urban green space evolution graphs for 2008,2012,and 2016 are visualized,where evolution graphs can be represented as evolution trajectories.Hereafter,the dynamic changes of urban green space can be extracted from the evolution graph.Furthermore,the number and spatial distribution for each green space evolution type were analyzed.The results indicated that in our study area,the number of born is larger than that of die,the number of splitting is larger than that of merging,and the number of contraction is larger than that of expansion.Moreover,the number of green patches which involved born and die is significantly larger than other evolution types.From the spatial distribution,urban green space patches which experienced changed are distributed more evenly,with those who have small area size are mostly distributed in the center of the study area,while those have large area size are mainly distributed near the outer ring.Compared with the changes from 2008 to 2012,the number of green patches experiencing born during the period from 2012 to 2016 was decreased,especially in the southern part of the Pudong New Area.The distribution of urban green patches who has an evolution type of die has spread gradually from few areas where these green patches are concentrated to the whole study area.(3)The method for analyzing the dynamic changes of connectivity in urban green space was proposed in this study as well.By combining the connectivity graph and evolution graph together,the changes of urban green space connectivity are explicitly extracted.Specifically,the edges which involved born and die are identified between two snapshots and thus can be used to analyzed how the connectivity among urban green patches changed over years.Comparing the connectivity changes of two years,it is found that the born edges of the latter year is higher than the die of the previous year in terms of quantity and distribution density of partial area.Furthermore,there are a certain similarity spatial distribution pattern between the die and born edges.(4)System development.Based on Visual studio2012 platform and C# language,the urban green space connectivity analysis system is developed.The system can be applied to more areas and different time series for urban green space and other one or a variety of land use types of connectivity and change research. |
PDF Full Text Request