欢迎访问《生态学杂志》官方网站,今天是 分享到:

生态学杂志

• 方法与技术 • 上一篇    

RS和GIS技术在中尺度景观类型划分与制图中的应用:以成都市龙泉驿区为例

欧定华1,夏建国1**,张莉2,欧晓芳3,赵智4   

  1. (1四川农业大学资源学院, 成都 611130; 2中共成都市龙泉驿区委组织部, 成都 610100; 3成都理工大学地球科学学院, 成都 610059; 4四川农业大学管理学院, 成都 611130)
  • 出版日期:2015-10-10 发布日期:2015-10-10

The application of RS and GIS technology in meso-scale landscape classification and cartography: A case study in Longquanyi District of Chengdu.

OU Ding-hua1, XIA Jian-guo1**, ZHANG Li2, OU Xiao-fang3, ZHAO Zhi4   

  1. (1 College of Resources, Sichuan Agricultural University, Chengdu 611130, China; 2Organization Department the CPC Committee of Longquanyi District of Chengdu City, Chengdu 610100, China; 3 College of Geosciences, Chengdu University of Technology, Chengdu 610059, China; 4 College of Management, Sichuan Agricultural University, Chengdu 611130, China)
  • Online:2015-10-10 Published:2015-10-10

摘要: 以成都市龙泉驿区为研究区,利用Landsat-8 OLI影像、ASTER GDEM等数据,应用RS、GIS技术对景观类型划分与制图进行研究。结果表明:ISODATA遥感影像非监督分类法可以实现研究区地貌类型的自动划分,不但降低了传统分类中主观因素对分类结果的影响,而且把沟谷浅丘等小尺度地貌类型划分出来,确保了地表形态的连续性和渐变性;QUEST、C5.0、MLC分类结果图的总体分类精度、Kappa系数、平均用户精度、平均制图精度大小依次均为QUEST>C5.0>MLC,平均错分误差、平均漏分误差大小依次均为QUEST<C5.0<MLC,说明QUEST决策树分类法是进行研究区遥感影像土地利用/覆被类型分类的最佳方法;把ArcGIS空间分析、地图编制技术与Python编程相结合进行景观类型划分与制图,能够克服GIS平台常规功能局限,提高制图效率,具有较强实用性;研究区共划分为18种景观类型,其景观类型分布特点与区域景观格局实际相符,说明集成应用QUEST遥感影像决策树分类、GIS空间分析和地图编制、Python编程技术方法能综合利用多类景观生态分类指标实现研究区景观类型划分与制图,充分说明RS、GIS技术在中尺度景观类型划分与制图中具有较强推广价值和应用前景。

关键词: 完全比值型植被指数, 非比值型植被指数, 地形效应, 非完全比值型植被指数, 几何光学模型

Abstract: In order to understand the local application of RS and GIS in classifying landscape and cartography, data of Landsat-8 OLI images and ASTER GDEM were used to landscape classification and cartography in Longquanyi District, Chengdu. The results showed that, ISODATA remote sensing image unsupervised classification method could automatically classify the types of landscape in the study area. The method could not only reduce the effects of manmade subjective judgments in the traditional classification, but also distinguish small scale landform types such as the valley and shallow hill to ensure the continuity and gradual changes of surface morphology. Compared with C5.0 and MLC, QUEST showed the higher overall classification accuracy, Kappa coefficient, average user accuracy, and average mapping accuracy of classification. Moreover, the average misclassification error and average omission error showed the order as QUEST < C5.0 < MLC, indicating that the QUEST decision tree classification method is the best in classifying land use/cover type in the study area. In addition, the combination with ArcGIS spatial analysis, map compilation technology and Python programming displayed much strong practicability, since the combined method could overcome the limitation of GIS platform general function and then improve the mapping efficiency. The Longquanyi District was divided into 18 kinds of landscape types. The landscape distribution characters were consistent with the actual regional landscape pattern. The results here suggest that the integrated application of QUEST remote sensing image decision tree classification, GIS spatial analysis and map compilation, Python programming technique take multiple landscape ecological classification indexes into consideration, and efficiently realize the landscape type classification and cartography in the study area. RS and GIS technology showed strong popularization and application values in mesoscale landscape classification and cartography.

Key words: non-ratio vegetation index, topographic effect, incomplete ratio vegetation index, complete ratio vegetation index, geometric optical model.