Landsat 8 OLI影像新增特征对土地覆盖遥感分类的影响分析

Effects of New Characteristics of Landsat 8 Operational Land Imager (OLI) Data on Land-cover Remote Sensing Classification

本研究针对OLI影像在辐射分辨率、近红外波段和全色波段的光谱范围以及波段数量等方面进行的调整,与ETM+进行比较分析,探究其对土地覆盖遥感分类的影响。实验发现:1)辐射分辨率的提高有利于地类之间的区分,进而提高分类的总体精度;2)近红外波段范围的调整有利于识别水体边界,对分类总体精度影响不大;3)新增的深蓝波段可明显提高水体、不透水层的分类精度,进而提高总体分类精度,卷云波段的加入不利于地类之间的区分,降低分类精度;4)全色波段范围的收窄可明显提高林草地、耕地的分类精度,裸土的分类精度也有所提高,有利于区分植被与非植被信息。研究结果能够推广OLI影像的使用,并对应用OLI影像进行土地覆盖分类具有借鉴意义。

OLI sensor has new characteristics in some bands of Landsat 7 ETM ＋, which may improve radiometric resolution from 8-bit to 16-bit, refine spectral range of NIR band and Pan band, add coastal band and cirrus band. The study aims to analyze new characteristics of OLI data and explore their influ- ences on land cover remote sensing classification. Results show that： （1） improving radiometric resolu- tion can help the differentiation of classes and overall accuracy of the classification; （2） spectral range adjustment of NIR band helps to identify waters boundary but has little influence on overall classification accuracy; （3） coastal band can significantly improve classification accuracy of waters and impervious layers, while the addition of cirrus band is bad for the differentiation of classes; （4） narrowing spectral range of Pan band can improve classification accuracy of grass and crop lands and bare soil obviously, hence helpful in the differentiation of vegetated and non-vegetated information.