渤海AVHRR多通道海冰密集度反演算法试验研究  

Study on the multi-band retrieval algorithm for the Bohai Sea ice concentration using AVHRR data

作　　者：刘志强[1] 苏洁[1] 时晓旭[1,2] 赵进平[1] 

机构地区：[1]中国海洋大学海洋环境学院,山东青岛266100 [2]中国科学院大气物理研究所,北京100029

出　　处：《海洋学报》2014年第11期74-84,共11页

基　　金：国家海洋局公益项目(201105016);气象公益性行业项目(GYHY201306049)

摘　　要：为了得到更精确的渤海海冰密集度反演参数,采用辽东湾不同类型海冰ASD实测数据,在分析光谱特征的基础上,针对NOAA/AVHRR数据确定出合适海冰密集度反演算法阈值。继而,基于线性光谱混合模型的多通道反演算法进行了一系列算法试验。同时实现了基于LandSat5-TM数据的渤海海冰密集度场反演,并利用该结果与AVHRR单通道和多通道算法得到的海冰密集度反演结果进行比对分析。定量误差分析结果表明,当单通道算法或组合算法中包含1通道时,与Landsat5-TM反演结果的平均误差为正值,包含2通道且不包含1通道时,平均误差为负值;同时使用这两个通道较只包含其一的各种组合算法的平均误差明显偏小;在各种组合算法中,1245四个通道组合反演的海冰密集度结果误差最小,可应用于渤海AVHRR数据海冰密集度反演。To achieve the improved retrieval parameter of sea ice concentration（SIC） in the Bohai Sea,spectral char- acteristics of different types of sea ice are obtained and analyzed using in-situ ASD measurements in the Liaodong Bay. Appropriate threshold is thus determined for the retrieval of AVHRR data and then the SIC is calculated. In this paper,muff-band retrieval algorithm based on linear spectral mixture model （LSMM） is used for conducting a series of band experiments. The retrieved SIC using LandSat5-TM data is used to compare with the AVHRR SIC of different combinational algorithms. The results of error quantitative analysis indicate that the mean error （ME） of the algorithms involving bandl is positive, but the ME turns to negative when the algorithms involve band2 in- stead of band1 ; the ME of the algorithms which include both of two bands is less than the those which include only band1 or band2; among these algorithms, the 1245 joint-band algorithm has the minimum errors：it is concluded that the 1245 joint-band algorithm can be used for the retrieval of AVHRR SIC in the Bohai Sea.

关 键 词：渤海 海冰密集度反演 线性光谱混合模型 NOAA/AVHRR LandSat5-TM 

分 类 号：P731.15[天文地球—海洋科学]