流域地物识别的多谱段遥感数据分析

在陆地卫星资料的应用研究中,先对少量样本进行分析,然后上计算机处理,可以减少研究工作中的浪费与盲目性。本文以黄河下游大汶河为例,利用陆地卫星多谱段数据,对流域的稠密森林山地、稀疏林山地、草被灌丛山地、丘陵农田、平原农田、水体、城市等地物进行数值分析识别,并与常规目视解释作了比较。通过分析比较认为:陆地卫星单波段识别地物的功能,若以优劣评阶,影象的次序为:MSS5、MSS6、MSS7、MSS4,数值分析的次序为:MSS5、MSS7、MSS6、MSS4,但任一单波段都不能识别所选择的全部地物;由MSS5与MSS7构成的二维空间是识别流域地物的最佳判别空间;四个波段合成的彩色片可识别8类地物中的5种,具有良好的识别功能;利用MSS7～MSS5灰阶值点聚图及判别分析方法,可对流域地物进行识别分析,这种非计算机识别地物的数值法,适合于研究指定范围内样本的地物类型。

ANALYSIS OF RECOGNIZING LAND COVER OF A WATERSHED WITH MULTISPECTRAL REMOTE SENSOR DATA

The launching of land satellites makes it possible to make a quick and accurate study of land cover of a watershed, which is of great significance in hydrography.The present article takes Da-Wen River in the lower reaches of the Huang He River as an example, and selects seven kinds of typical land cover, mountains covered with dense forests, mountains covered with sparse forests, hills covered with grass and bush, hilly fields, plain fields, water and city. By the use of the digital data of films of the second land satellite of Oct.31, 1975 in 4 bands and the digital data of photograph of the third land satellite of April 1, 1978, MSS7, land cover of a watershed was analysed in a quantitative way and its was result compared with that of visual interpretation.