胶州湾水体悬浮物浓度遥感反演模式优化研究

Study on Remote Sensing Retrieval Model Optimization of Suspended Sediment Concentration in Jiaozhou Bay

为了及时、准确的了解胶州湾水域总悬浮物情况,采用2001—2015年水面实测数据,选取HJ-1、MERIS、LandsatTM/ETM+三个不同卫星数据,通过ENVI软件分析计算TM/ETM数据的多波段组合与总悬浮物浓度之间的相关关系,选取相关系数最大者分别构建多元函数回归模型,对胶州湾总悬浮物浓度进行了遥感定量反演研究。结果表明,估算因子三波段模型432算法与悬浮物浓度的相关性最好(R2>0.8),反演精度较高,以此建立了悬浮物浓度三波段半分析+生物光学遥感反演模型,检验值R^2均高于0.85,并通过F检验法和均方根误差(RMSE)分析,证明误差的敏感性差,该模型相关系数及稳定性好。因此,三波段模型在这三种反演模型中精度最好。

Semi empirical model inversion of suspended sediment concentration is the main method for remote sensing inversion of suspended sediment concentration.However,a lot of research results show that Jiaozhou bay water pollution,retrieval accuracy and applicability of the model of the semi empirical model has great difference,therefore,a study on inversion of suspended sediment concentrations in the Gulf of Jiaozhou for the algorithm,it has an important significance for promoting remote sensing inversion.This study measured data by 2001-2015,MERIS,HJ-1,LandsatTM/ETM from three different satellite data,the establishment of Jiaozhou Bay suspension concentration band ratio inversion model and three band inversion model,comparing the model inversion effect,compared to select the optimal model algorithm.The results show that the best correlation factor to estimate the three band model 432 algorithm and the concentration of suspended solids the(R 2>0.8),high inversion precision,in order to establish the concentration of suspended solids three band semi analytical+bio optical retrieval model.The use of 10 sets of independent experimental data test results show that the three band model inversion with the measured values of R 2 were higher than 0.85,and through the F test method and the root mean square error(RMSE)analysis,prove that the error sensitivity difference,correlation coefficient of the model and good stability.Therefore,the three band model is one of the best models in these three models.