摘要
卫星遥感的边缘海溶解有机碳(DOC)的时空分布对于厘清区域和全球碳循环具有重要意义。有色溶解有机物(CDOM)吸收系数[ag(λ)]被广泛应用于构建边缘海DOC的卫星反演模型。不同的研究选择不同波长的ag(λ),而不同波长的选择是否影响DOC反演结果的准确度却未见报导。针对此问题,本文以长江口为例,以2017年冬季、春季和夏季的调查数据为基础,探讨如何选择合适波长的ag(λ)用于优化边缘海DOC的反演模型。首先,通过分析不同波长的ag(λ)和DOC的相关性显示,当波长在250~385 nm时,ag(λ)和DOC浓度呈现良好的线性关系(r 2=0.67±0.01),DOC的均方根误差(RMSE=(25.1±0.5)μmol·L^-1)也较小;随着波长的增大,两者之间的相关性急剧减弱,对应的RMSE也急剧增大。因此,DOC的反演宜选择波长小于385 nm的ag(λ)。其次,通过分析ag(λ)随波长的对数衰减关系显示,紫外波段的衰减速度通常要快于可见光波段。当以可见光波段的数据来拟合建立ag(λ)随波长的对数衰减公式时,该公式可延伸用于估计波长大于380 nm的ag(λ),其产生的误差通常在±5%以内;而对于波长小于380 nm的ag(λ)则往往产生较大低估,且低估程度随波长减小而增大。因此,对于目前的水色卫星(未包含紫外波段),当使用可见光波段的遥感反射率[R rs(λ)]通过GSM、QAA等模型进行反演时,对于波长大于380 nm的ag(λ)的反演结果的置信度较高。综合上述分析,选择波长380~385 nm的ag(λ)来构建长江口DOC的反演模型有利于降低反演结果的系统误差。长江口作为全球边缘海的典型代表,本研究的结果对于优化全球边缘海的DOC反演模型具有重要的指示意义。
The distributional patterns of the satellite-derived concentrations of dissolved organic carbon(DOC)in the marginal seas are essential to evaluate the regional and global carbon cycles.Although the remotely sensed concentration of DOC in the marginal seas is commonly assessed from the spectral absorption coefficients of colored dissolved organic matter(CDOM)[ag(λ)],the wavelengths of ag(λ)that have been used for this purpose are reported differently in literature.Whether and how the use of different wavelengths of ag(λ)affects the accuracy in the remotely sensed concentrations of DOC has not yet been reported.In this study,by using the field in situ observations collected in the Yangtze River estuary(YRE)during the winter,spring and summer cruises in 2017,the optimal wavelengths of ag(λ)used to estimate the concentration of DOC from space were evaluated.First,the linear regressions of DOC from ag(λ)were robust for wavelengths of 250~385 nm with r 2 of 0.67±0.01 and root mean square error(RMSE)of(25.1±0.5)mol·L-1,but the relationships became increasingly weaker with the increase of wavelength.Secondly,although ag(λ)could be expressed as an exponential decay function with wavelength,the decay slope(S)was generally larger in ultra-violet(UV)bands than in visible bands.As such,when the best fit regression of ag(λ)from wavelength based on the observations at visible bands was applied to estimate ag(λ)at UV bands,the uncertainty(within±5%)was only acceptable forλ>380 nm,but a marked underestimation was found forλ<380 nm and the degree of underestimation was larger at shorter wavelengths.This implies for a potential underestimation of the satellite-derived ag(λ)atλ<380 nm by using current semi-analytic models(e.g.GSM and QAA).The above results indicated that the optimal wavelengths of ag(λ)used to estimate DOC from space in the YRE were 380~385 nm.The optimization of the wavelengths of ag(λ)in remotely sensing DOC may be similarly assessed in other marginal seas.
作者
刘娟娟
郭金强
潘晓驹
LIU Juan-Juan;GUO Jin-Qiang;PAN Xiao-Ju(The Key Laboratory of Marine Chemistry Theory and Engineering Technology,Ministry of Education,Ocean University of China,Qingdao 266100,China;Laboratory for Marine Ecology and Environmental Science,Pilot National Laboratory for Marine Science and Technology(Qingdao),Qingdao 266237,China)
出处
《中国海洋大学学报（自然科学版）》
CAS
CSCD
北大核心
2020年第6期80-87,共8页
Periodical of Ocean University of China
基金
鳌山科技创新计划项目(2016ASKJ02)
广东省海洋遥感重点实验室(中国科学院南海海洋研究所)开放课题项目(2017B030301005-LORS1802)
国家重点研究发展项目(2018YFC1407600)资助。
关键词
溶解有机碳(DOC)
有色溶解有机物(CDOM)
CDOM吸收系数(ag)
波长选择
长江口
dissolved organic carbon(DOC)
colored dissolved organic matter(CDOM)
the absorption coefficient of CDOM(ag)
the selection of wavelength
the Yangtze River estuary
