Developing Integrated Methods to Construct Urban Potential Green Corridors： A Case Study of Changzhou City | 城市潜在绿色廊道构建方法――以常州市为例

中国人. 在RS和GIS的支持下,利用遥感影像分类、斑块连通性分析和耗费距离模拟等方法形成一套完整的城市潜在绿色廊道构建体系。首先采用最优指数（OIF）和决策树（QUEST算法）获取城市绿地斑块空间布局,分析绿地斑块间的连通性;然后利用Spearman秩相关等统计方法确定不同连通指数对连通阈值的敏感程度,优选最稳定的连通指数划分绿地斑块的重要性等级,将具有较高重要性值的斑块作为绿地＂源＂和＂目标＂;最后选用频率比方法改进耗费距离模型的成本表面,在不同空间尺度下,情景模拟城市潜在绿色廊道的空间布局,并对模拟结果进行定量分析与评价。结果表明：最优指数结合决策树方法的分类总精度为90.35%,Kappa系数为0.86,提取绿地面积占城市总面积的22.88%,大面积绿地斑块主要分布在韬奈鞑亢吞谋辈;整体连通指数（IIC）具有较高的稳定性,在200～2400m的距离范围内,其dIIC的Spearman秩相关系数变化值最低（0.21）;在600m的距离阈值下,以dIIC值大于1%作为标准,确定竺山斑块为绿地＂源＂及7个主要的绿地斑块为＂目标＂;在不同空间尺度下,模拟的潜在绿色廊道的变形系数ξ值相近（1.28）,其空间格局受研究尺度的影响较小。最后进一步探讨该方法体系实践的可行性,为城市绿地系统规划和建设提供科学合理的参考和依据。

英语. With the support of remote sensing（RS）and geographical information system（GIS）techniques, we present an integrated approach of remote sensing classification, connectivity metrics analysis and least-cost path simulation to develop urban potential green corridors in urban greenspace system planning. Firstly, this paper used the OIF（optimal index factor）and decision tree algorithms （QUEST）to extract urban greenspace patches, and analyzed the connectivity of each patch. Then, Spearman rank correlation coefficient was derived to determine the connectivity metrics sensitivity due to distance changes. Based on this, the most robust connectivity index was determined to divide patches importance rank, the patches with higher importance values were defined as greenspace ‘sources’ and ‘target’. Ultimately, the potential green corridors were simulated by using the least-cost model based on FR（frequency ratio）cost-surface under different research scales, followed by a quantitative evaluation of different scenarios simulation results. The preliminary results were as follows： 1）The total classification accuracy reached 90.35% by using OIF and QUEST, with a Kappa coefficient of 0.86; greenspace mainly distributed in the east of Gehu Lake and in the north of Taihu Lake, and accounted for 22.88% of the total area. 2）Integral index of connectivity（IIC）had higher robust with lower difference value of rs（0.21）between 200 and 2 400 m distance threshold; under the 600 m threshold distance, the Zhushan patch and other seven patches were identified as ‘sources’ and ‘target’ by comparing important values（dIIC above 1%）. 3）The spatial patterns of potential green corridors were hardly affected by research scales changes, with stable deformation coefficient（1.28）. Finally, we discussed the feasibility of the optimized method system, and provided scientific and reasonable references for the planning and construction of city green system.