利用Landsat影像定量评价土地盐渍化方法探索.docVIP

利用Landsat影像定量评价土地盐渍化方法探索 　　摘要：以2015年7月甘肃省瓜州-敦煌地区Landsat影像为研究对象，先后提取了盐碱化指数（Salinization index，SI）、修改后盐碱化指数（Modified salinization index，MSI）以及归一化植被指数（Normalized difference vegetation index，NDVI）和以二维特征空间构建新的盐渍化遥感监测指数（Modified salinization detection index，MSDI），同时利用同期野外实际考察测量数据进行适应性评价和精度验证。结果表明，指数SI容易将盐碱地和裸土地混淆，不能准确表示盐渍化程度；MSI在提取盐渍化方面具明显优势，但其仅考虑样本像元内以裸土为主的情况，忽略了像元内植被光谱信息。最后综合考虑裸土和植被信息，采用新构建的指数MSDI对该区域盐渍化进行定量化提取，结果表明MSDI指数对盐渍化提取的总体精度为86.36%，能很好反映该地区盐渍化程度。 　　关键词：Landsat；遥感影像；盐渍化；瓜州-敦煌地区 　　中图分类号：S15 文献标识码：A 文章编号：0439-8114（2018）01-0051-05 　　DOI：10.14088/j.cnki.issn0439-8114.2018.01.013 　　Abstract： The Landsat image of Guazhou-Dunhuang area in Gansu province in July 2015 was taken as the research object. salinization index （SI），modified salinization index （MSI），normalized difference vegetation index （NDVI） and a new Salinization detection index（MSDI） constructed by two-dimensional feature space were extracted. At the same time，the adaptive evaluation and precision verification were carried out by using the actual field survey data. The results show that the SI is easy to confuse the saline-alkali soil and the bare land，and can’t accurately indicate the degree of salinization；MSI has obvious advantages in salinization’s extraction，but MSI only considers the case of bare soil in the sample pixel，ignores the vegetation spectral information. Finally，combined the bare soil and vegetation information，the newly constructed MSDI index was used to extract the salinization quantitatively in this area，the results show that the overall accuracy of MSDI index for salinization is 86.36%， and it is a good reflection of the degree of salinization in the area. 　　Key words： Landsat； remote sensing image； salinization； Guazhou-Dunhuang area 　　目前全球约有8.31亿hm2土壤受到盐渍化威胁，而且这个数字还在不断上升[1]。土壤?}渍化是由于一定的气候、土壤地质、水文等自然因素共同作用的结果[2]。不合理的灌溉技术及地下水的开采等人类活动会使得土壤发生次生盐渍化，或者加重土壤盐渍化的程度[3，4]。生态脆弱的干旱区的降水较少、蒸发强度大，相对于其他区域，更易发生土壤盐渍化，这对于干旱区的生态环境及绿洲农业的可持续发展造成了严重的威胁[5，6]。应用科学的手段动态、准确地监测盐渍化变化，对掌握盐渍化时空变异特征及定量化评价及生态环境