基于研究区2001年Landsat7ETM+遥感影像数据,运用遥感与GIS技术手段,在对遥感影像的边界裁定、几何校正、辐射校正等预处理的基础之上,根据绿洲景观生态类型分类体系,通过综合应用非监督分类、植被指数与波段比值指数聚类、监督分类以及类型叠加与图像整合等方法,进行绿洲景观生态类型的遥感解译与分类,生成研究区2001年的景观生态类型图。为尽可能利用到遥感影像的所有原始数据信息,论文选取5种植被指数(NDV I、DV I、RV I、IPV I、SA VI)和9种波段比值指数(Index1I-ndex9)参与到遥感影像的解译与分类当中,结果表明:NDV I、DV I、IPV I、SAV I、R VI、Index5、Index6具有较大的相似性,能够明显地将具有植被信息的类别分离出来,利于划分具有植被信息的景观类型;Index1—Index4具有较好的类别空间分离性,利于不同类别间的聚类与区分;而Index7—Index9的类别分离性则较差,不利于类别聚类与划分。因此,在实际的应用中,选取多项植被指数参与景观分类,不仅能够发现新的信息,而且也会明显提高景观生态类型、尤其是干旱区绿洲景观生态类型的遥感解译与分类能力。
Study on landscape ecological classification is the basis for understanding landscape patterns and ecological functions in landscape ecology.And the application of remote sensing for classifying landscape ecological types was one of the most important methods in landscape ecology all along.For finding a way to enhance the ability of landscape ecological classification,this study will try using different vegetation indices and other band ratio indices to join the classification process.In this paper,with remote sensing and GIS techniques,Ejin natural oasis landscape ecological classification was carried out using Landsat7 Enhanced Thematic Mapper Plus data in 2001.Firstly,remote sensing image data preparation was done based on image masking,geometric correction and radiometric correction.Then,according to Ejin natural oasis landscape ecological classification system,unsupervised classification,vegetation indices and band ratio indices clustering,supervised classification and landscape types overlaying and integrating,orderly,were executed to interpret and classify the map of classification of Ejin natural oasis landscape in 2001.In order to use all the bands of original remote sensing image data,five vegetation indices such as NDVI,DVI,RVI,IPVI,SAVI and nine band ratio indices such as Index1,Index2,...,Index9 were selected to participate in the interpretation of landscape ecological types.Results show that,NDVI,DVI,IPVI,SAVI,RVI,Index5 and Index6,with similar characte-ristics,can obviously separate landscape types with vegetation ingredients.Index1 to Index4 have better spatial separability and can easily identify different landscape types.For Index7,Index8 and Index9,owing to their worse spatial separability,they can difficultly differentiate landscape types.In conclusion,during the processes of landscape ecological classification based on remote sensing image data,more vegetation indices assisting can not only find new information,but can also clearly enhance the ability of interpretation and classification of landscape ecological types,especially for those oasis landscapes in arid regions.