摘要
利用2005年渤海近岸水体生物光学数据集,建立了基于水体遥感反射率光谱Rrs(λ)数据的490nm波段水体漫衰减系数Kd(490)经验反演模型,经实测数据检验,模型反演结果的平均相对误差为18.4%,均方根误差(对数坐标下)为0.094m-1,相关系数R2(对数坐标下)为0.902。分析了模型的噪声敏感性,在输入端引入±5%误差的情况下,模型反演结果的平均相对误差波动在9%以内,均方根误差的变化在0.035m-1以内,模型是稳定可靠的。以ENVISAT MERIS数据为例,进行了模型的示范应用,给出了渤海Kd(490)的空间分布。
Diffuse attenuation coefficient of downwelling irradiance at 490nm, Kd（490）, is one of the important ocean optical properties. The retrieval model is of great interest especially for the optically complex water, where terrestrial influence on ocean optical properties is significant. Comprehensive bio-optical dataset were acquired in the nearshore area of the Bohai Sea in 2005. Statistical retrieval model for Ko （490） based on remote sensing reflectance spectra is developed. The model is validated by in situ data with the Absolute Percentage Difference （APD）, Root Mean Square error （RMS, in log10 scale） and Pearson correlation coefficient （in log10 scale） of 18.4%, 0.094 m^-1 and 0.902 respectively between in situ and retrieved Kd（490） values. Error performance of the model is analyzed and it is shown that APD of the retrieval is better than 30% for 76% of all the in situ data. Model＇s sensitivity to the input noise is analyzed by imposing ±5% error of remote sensing reflectance spectra artificially, which proves that the model is robust and reliable with the error fluctuation of model output being less than 9% for APD and 0.035 m^-1 for RMS. The model is applied to ENVISAT MERIS data to give the spatial distribution of Kd（490） of the Bohai Sea.
出处
《遥感学报》
EI
CSCD
北大核心
2009年第3期411-422,共12页
NATIONAL REMOTE SENSING BULLETIN
基金
The research was supported in part by National Science Foundation of China (No. 40706060, 40606043)
Dragon Project (ID 5292)
High-tech Research Development Program of China (No. 2006AA09A310)
State Oceanic Administration (SOA) of China (No. 200705027, 2008401, and 908-01-ST11)
First Institute of Oceanography (No. 2007B05)
关键词
渤海
漫衰减系数Kd(490)
遥感反射率
反演模型
diffuse attenuation coefficient Kd（490）, remote sensing reflectance, Bohai Sea, retrieval model
