基于植被指数的GF-1与Landsat-OLI石漠化识别能力对比评价

Comparison of Rocky Desertification Detection Ability of GF-1 and Landsat-OLI Based on Vegetation Index

植被指数是运用多源遥感影像提取石漠化过程中的主要参考指标之一。为了在石漠化提取中选择最优植被指数,论文以GF-1和Landsat-OLI为源数据,运用欧氏距离对多种植被指数在石漠化提取过程中的可分性和类型识别能力进行了定量的对比评价。结果表明:LandsatOLI在石漠化与非石漠化、不同等级石漠化信息提取的可分性上略优于GF-1,共有71个参数的欧氏距离大于等于阈值1.56;通过植被指数光谱特征,可以对非岩溶区与石漠化地区进行较好的区分,其类型间欧氏距离普遍高于阈值;然而由于相邻等级石漠化之间植被覆盖率存在渐近式过渡关系,在遥感影像上光谱反射率接近,比间隔等级石漠化更加难于区分。在石漠化类型识别能力方面,波段差和比方法优于单一光谱指数。对于GF-1和Landsat-OLI而言,石漠化信息提取中推荐使用的最优植被指数均为NDVI,其次为GRNDVI。

Vegetation index is one of the major parameters in rocky desertification information extraction by using multi-source satellite images. Based on GF-1 and Landsat-OLI, this paper compared the detachability and detection ability of multiple vegetation indices with Euclidean distance between them, so as to choose an optimal parameter from satellite images and acquire the accurate rocky desertification information. It is proved that Landsat-OLI is better than GF-1in classifying different grades of rocky desertification and separating rocky desertification region from non-desertification region, and it has 71 parameters with Euclidean distance value greater than the threshold value of 1.56. It is easy to distinguish the rocky desertification area from non-karst area through the Euclidean distance. When classifying different grades of rocky desertification, band difference and ratio are better than single spectral index. Finally, the recommended optimal vegetation index of rocky desertification information extraction from GF-1 images and Landsat-OLI images is NDVI, followed by GRNDVI.