干旱半干旱区植被盖度信息提取技术研究

为系统地研究干旱半干旱区植被盖度提取方法,比较了目前常用的几种高光谱影像植被盖度提取技术。结果表明:采用NDVI建立的像元二分模型对植被盖度的估测能力低于直接采用NDVI建立的回归模型;4种处理的PCR模型植被盖度估测的建模精度由高到低为:CR>NO>FD>SD;4种处理的PLSR模型植被盖度估测的建模精度由高到低为:FD>NO>SD>CR;FCLS分解的结果明显优于LS模型;综合分析后,认为基于FD的PLSR模型对研究区植被盖度估测效果最佳。文中研究旨在为干旱半干旱区植被盖度的更深入研究提供参考。

Research on vegetation coverage information extraction technologies for arid and semiarid area

To systematically study the vegetation information extraction technology in arid and semiarid area,variety of currently popular vegetation coverage extraction methods using hyperspectral remote sensing were compared.The results indicated that the vegetation coverage estimation ability of dimidiate pixel model established by different NDVI was lower than that of the regression models established by NDVI directly.The modeling accuracy of vegetation coverage of PCR model by four kinds of spectrum processing methods was: CR>NO>FD>SD.The modeling accuracy of vegetation coverage of PLSR model by four kinds of spectrum processing methods was: FD>NO>SD>CR.The decomposition results accuracy of FCLS model was obviously higher than that of LS model.After the comprehensive comparison,it was considered that the vegetation coverage estimation ability of PLSR model based on FD is the strongest in arid and semiarid area.This study aims at providing a reference for the further study to vegetation coverage in arid and semiarid area.