摘要
采用静止轨道海洋水色卫星(GOCI)数据对长江口及其邻近海域有色溶解有机物(CDOM)反演。以QAA-CDOM算法为基础,根据实测数据,利用BP神经网络模型来拟合QAA-CDOM算法中需要针对长江口水体进行优化的悬浮颗粒后向散射系数bbp与吸收系数ap的关系,从而准确估算CDOM的浓度。结果表明,反演结果准确度较高,平均相对误差为0.35。基于GOCI日内连续成像的优势,选取2014年3月15日8景GOCI影像,利用优化后的QAA-CDOM-BP算法,对长江口及其邻近海域CDOM的日内变化进行反演和分析,得到的变化规律如下:长江口及其邻近海域的CDOM日变化主要受潮流、长江径流等共同影响。长江口内CDOM浓度在涨潮期高于退潮期,由于受长江冲淡水的作用,CDOM从口外往外海区呈现逐渐递减趋势。
GOCI satellite data is adopted to retrieve high concentrations of colored dissolved organic matter in coastal waters of Changjiang Estuary. The inversion model is based on QAA-CDOM algorithm and field measured data. The BP neural network was used to fit the relationship between bbp（555） and ap（443）, which is used in QAA-- CDOM and needs to be optimized for the water in the Changjiang Estuary. The results show that the inversion ac- curacy is excellent with mean relative error 0.35. Then based on advantage of GOCUs serial imaging, 8 images ac- quired in March 15, 2014 were inverted and analysed for CDOM daily variations in Changjiang Estuary and its ad- jacent seawater. The obtained the CDOM variation pattern is that CDOM in Changjiang Estuary and its adjacent seawater is mainly impacted by tides and the discharge of Changjiang. Inside the Changjiang Estuary, due to the effect of Changjiang diluted water, CDOM concentration in high tide is higher than that in the low tide, and it showed gradually decreasing trend from the estuary to the open sea.
出处
《海洋学报》
CAS
CSCD
北大核心
2017年第9期133-145,共13页
基金
国家自然科学基金面上项目(41471346)
国家自然科学青年基金项目(41401404)
海洋公益性行业科研专项经费项目(201005030-06)
