内陆水体水质参数遥感反演集合建模方法

发布时间：2019-01-20 19:14

【摘要】：以微山湖为研究对象,利用2015年6月11~13日获取的实测高光谱和水体叶绿素a浓度、总悬浮物浓度和浊度数据,构建3种水质参数遥感反演常用的经验模型和PSO-SVM模型并进行精度评价,确定参与3种水质参数集合建模的反演模型,分别利用以熵权法(EW-CM)、集对分析法(SPA-CM)为代表的确定性集合建模方法和以贝叶斯模型平均(BMA)为代表的概率性集合方法构建反演3种水质参数的EW-CM、SPA-CM和BMA集合模型.通过贝叶斯平均方法获取各模型和BMA集合模型反演3种水质参数的不确定性区间,对比3种水质参数各模型和集合模型反演结果.结果表明:(1)确定性集合模型中SPA-CM模型精度整体高于EW-CM模型;(2)BMA概率性集合模型建模精度整体上要优于SPA-CM和EW-CM集合模型,验证精度稍低于SPA-CM模型,和EW-CM模型相当;(3)概率性集合建模可以给出集合模型和各模型反演水质参数的不确定性区间;(4)确定性和概率性集合模型可以综合各模型信息,使得集合模型同时具有较高的建模和验证精度,降低单一模型反演水质参数的不确定性,并在一定程度上提高水质参数反演精度.
[Abstract]:Based on the experimental data of hyperspectral, chlorophyll a, total suspended matter and turbidity obtained from June 11 to 13, 2015, the microlake was used as the research object. Three common empirical models and PSO-SVM models for remote sensing inversion of water quality parameters are constructed and their accuracy is evaluated. The inversion models that participate in the modeling of the three water quality parameters are determined, and the entropy weight method (EW-CM) is used respectively. The deterministic set modeling method represented by set pair analysis (SPA-CM) and the probabilistic set method represented by average (BMA) of Bayesian model are used to construct EW-CM,SPA-CM and BMA set models for inversion of three kinds of water quality parameters. The uncertain intervals of the three water quality parameters were obtained by Bayesian averaging method and the inversion results of each model and the set model of the three water quality parameters were compared between each model and the BMA set model. The results show that: (1) the accuracy of SPA-CM model in deterministic set model is higher than that of EW-CM model; (2) the modeling accuracy of BMA probabilistic set model is better than that of SPA-CM and EW-CM set model on the whole, and the accuracy of verification is slightly lower than that of SPA-CM model, which is similar to that of EW-CM model. (3) the probabilistic set modeling can give the uncertainty interval of the set model and each model inversion water quality parameter; (4) the deterministic and probabilistic set model can synthesize the information of each model, which makes the set model have higher modeling and verification accuracy, and reduces the uncertainty of the single model inversion water quality parameter. The inversion accuracy of water quality parameters is improved to some extent.
【作者单位】：东华大学环境科学与工程学院国家环境保护纺织工业污染防治工程技术中心;中国水利水电科学研究院流域水循环模拟与调控国家重点实验室;
【基金】：国家自然科学基金资助项目(51309254) 国家重点研发计划资助项目(2017YFC0405801,2017YFC0405804) 中国水利水电科学研究院科研专项“十三五”重点科研项目(WR0145B272016);中国水利水电科学研究院流域水循环模拟与调控国家重点实验室开放研究基金资助项目(IWHR-SKL-201517)
【分类号】：X52;X87

【相似文献】