基于多源遥感的聊城市绿地空间特征研究

Study of Urban Green Spatial Characteristics in Liaocheng City Based on Different Remote Sensing Data

选取山东西部的聊城市为研究区,以Sentinel-2和Landsat-8影像为数据源,采用多尺度影像分割与面向对象分类方法,提取聊城市城区不同绿地,并选取计算了景观聚散性、景观多样性、斑块密度及差异性、邻近度等多个评价指数,分析聊城市城区绿地的空间特征。结果表明:(1)多尺度影像分割与面向对象分类方法在效率、处理速度及结果精度方面均优于传统绿地信息提取方法;(2)基于Sentinel-2的绿地提取总体精度为96.48%,Kappa系数为0.95,均显著高于Landsat-8的结果,尤其在城市道路绿地提取方面优势明显;(3)Sentinel-2的结果显示,除去耕地后的城市绿地面积为81.08 km^(2),占城区面积的40.92%,其中林草地占比最高为52.47%,道路绿地为47.53%。绿地的景观连通性较强,破碎度不高;(4)总体上聊城市城区绿地空间分布较合理,但面积占比较小,建议继续加强公园绿地的建设。

Taking Sentinel-2 and Landsat-8 data, the urban green space of Liaocheng City was extracted by multi-scale image segmentation and object-oriented image analysis method. Seven evaluation indexes of landscape dispersion, landscape diversity, patch density and difference, proximity were selected to analyze the spatial characteristics of urban green space in Liaocheng. The results show that:(1) Multi-scale image segmentation and object-oriented classification methods are superior to traditional green space information extraction methods in terms of efficiency, processing speed and result accuracy.(2) The overall recognition accuracy of Sentinel-2 is 96.48%, and the Kappa coefficient is 0.95, which are significantly higher than the results of Landsat-8, especially in road green space extraction.(3) Sentinel-2 results show that the urban green space area after removing cultivated land is 81.08 km^(2),accounting for 40.92% of the urban area, of which grassland accounts for the highest proportion of 52.47%, and road green space accounts for 47.53%. Green landscape connectivity is strong, fragmentation is not high.(4) The spatial distribution of urban green space in Liaocheng City is reasonable, but the proportion of urban green space is relatively small. It is suggested to continue to strengthen the construction of urban park green space.