基干Landsat影像博斯腾湖水质参数反演模型探讨.docVIP

基于Landsat影像的博斯腾湖水质参数反演模型研究 　　摘要：以内陆博斯腾湖为研究区，通过实测数据和遥感数据，分析了单波段数据、归一化数据、波段组合数据与溶解性总固体、矿化度之间的相关性，建立并筛选获得了最优模型，以期为博斯腾湖溶解性总固体和矿化度两个指标的大面积遥感反演提供理论依据。结果表明，对溶解性总固体与矿化度浓度进行反演，单波段最优分别为B3与B2，最佳模型皆为指数模型；归一化处理后最优波段为B3与B2，最佳模型分别为二次模型与指数模型；波段组合最优为B2×B3与B2-B7，最佳模型为指数模型与指数、线性模型。通过模型的验证与比较，研究发现，归一化法对于溶解性总固体浓度的反演能达到较好的效果，而波段组合法能较好的反演矿化度浓度 关键词：博斯腾湖；溶解性总固体；矿化度；遥感反演 中图分类号：P903 文献标识码：A 文章编号：0439-8114（2016）10-2507-07 DOI：10.14088/j.cnki.issn0439-8114.2016.10.016 Abstract： Based on the measured data and the Landsat images， analyzed the correlations of the concentrations of the total dissolved solids（TDS） and the mineralization degree in the lake with the single band data， the normalized data and the band combination data， respectively. The paper aims at establishing and selecting the optimal mathematical models to provide theoretical basis for the large-area remote sensing inversion of TDS and mineralization degree. The results showed that： 1） B3 and B2 is the optimal single band model for the inversion of the TDS mineralization degree， respectively. The best model for both of them is the exponential model. 2） After normalization， B3 is the optimal band for the inversion of the TDS and B2 is that of the mineralization degree. The best model for them is the quadratic model and the exponential model， respectively. 3） B2×B3 and B2-B7 are the optimal band combinations for the inversion of the TDS and the mineralization degree， respectively. The best model for them is the index model and the index/linear model， respectively. Through the validation and comparison among the models， this study found that the normalization method is more suitable for the inversion of the TDS and the band combination method is more suitable for the inversion of the salinity concentration. Key words： Bosten lake； total dissolved solids； mineralization degree； remote sensing inversion 随着社会经济的高速发展和人口的不断增长，世界范围内许多地区出现了水资源日趋紧张和水环境恶化加速的窘态[1，2]。目前，中国许多内陆湖泊正在萎缩、水环境污染也在不断加剧[3-5]，已经严重影响到了区域的可持续发展和人民生活的稳定，加大湖泊