内陆浑浊水体固有光学量（IOPs）反演算法

发布时间：2018-05-26 16:28

本文选题：浑浊水体 + 固有光学量　；参考：《南京师范大学》2015年硕士论文

【摘要】：本文面向浑浊富营养化二类水体,以太湖为研究区,在QAA基础上构建了水体吸收系数和后向散射系数双波段(QAA-DB)反演算法以及去纯水以外的吸收系数aexp(λ)分解算法。(1)QAA-DB算法在对水体反射光谱特征研究的基础上,首先利用677nm吸收谷和701nnm反射峰之间反射率的光谱斜率(Slope)为分类指标,以Slope=0.32为临界值,将混浊水体分为两类具有不同光学特征的水体,针对两类水体,分别构建IOPs反演模型,针对不同类别水体设置不同的模型参数,对于颗粒物后向散射系数进行分段模拟。1)400-685nm波段采用幂函数形式模拟,基于QAA算法,同时选取550nm和675nm两个参考波长进行颗粒物后向散射系数幂函数外推,并基于数据同化思想,利用多模型协同反演策略,对外推结果进行加权优化,确定最优颗粒物后向散射系数；2)对于大于685nm波段的颗粒物后向散射系数,以定值进行模拟；最后,通过生物光学模型,计算得到吸收系数。验证数据表明,QAA-DB算法对于复杂二类水体吸收系数反演误差MAPE为19.71%,RMSE为1.3933,反演精度令人满意。(2)针对内陆水体吸收系数的分解,本文提出了两种改进算法,即基于经验比值的吸收系数分解算法和基于高斯函数参数化的吸收系数分解算法,将aexp(λ)进行分解,得到浮游植物吸收系数αρh(λ)、非色素颗粒物及CDOM吸收光谱之和adm (λ)、1)基于经验比值的吸收系数分解算法,首先经验确定各组分吸收系数在450和480nm的比值及αdm(λ)衰减常数S,通过联立方程并结合αdm(λ)光谱负指数外推参数化模型,求解整个波段上的adm (λ),进而求得aρh(λ),验证数据结果表明,最终adm (λ)分解误差MAPE为24.72%,RMSE为0.81；2)基于高斯函数参数化的吸收系数分解算法,利用12个高斯函数对apn(λ)进行参数化,利用负指数模型参数化adm (λ),并经验确定adm(440),在此基础上,通过最小二乘优化算法进行未知参数的求解,实现对吸收系数的分解,最终adm(λ)分解误差M.APE为27.73%,RMSE为1.02。对于复杂二类水体,本文改进的两种分解算法都成功地对aexp(λ)进行了分解,精度令人满意。
[Abstract]:This paper aims at the turbid eutrophication of the second kind of water bodies, taking Taihu Lake as the research area. On the basis of QAA, the inversion algorithm of water absorption coefficient and backscattering coefficient is constructed, and the decomposition algorithm of absorption coefficient aexpA (位) except pure water is constructed. Firstly, the spectral slope of reflectivity between the 677nm absorption valley and the reflection peak of 701nnm is used as the classification index, and the critical value of Slope=0.32 is taken as the critical value. The turbid water body is divided into two kinds of water bodies with different optical characteristics. For the two kinds of water bodies, the IOPs inversion model is constructed separately. According to the different model parameters of different water bodies, the backscattering coefficients of particles are simulated by power function in the wavelength of 400-685 nm, based on the QAA algorithm. At the same time, two reference wavelengths, 550nm and 675nm, are selected to extrapolate the backscattering coefficients of particles. Based on the idea of data assimilation, the weighted optimization of the extrapolation results is carried out by using multi-model cooperative inversion strategy. The optimum backscattering coefficient of particles is determined to be 2) the backscattering coefficient of particles larger than 685nm band is simulated by the fixed value. Finally, the absorption coefficient is calculated by the bio-optical model. The verification data show that the inversion error of the QAA-DB algorithm for the absorption coefficients of complex water bodies is 19.71% (MAPE = 1.3933, and the inversion accuracy is satisfactory. 2) in view of the decomposition of the absorption coefficients of inland water bodies, two improved algorithms are proposed in this paper. That is, the absorption coefficient decomposition algorithm based on the empirical ratio and the absorption coefficient decomposition algorithm based on the parameterization of the Gao Si function are used to decompose expp (位). A decomposition algorithm of phytoplankton absorption coefficient 伪蟻 h (位 ~ (1), sum of absorption spectra of non-pigmented particles and CDOM) based on empirical ratio was obtained. First, the ratio of absorption coefficient to 480nm and the attenuation constant of 伪 dm (位) are determined empirically. By means of simultaneous equation and a dm (位) spectral negative exponent extrapolation parameterized model, the adm (位) on the whole band is solved, and then a 蟻 h (位) is obtained. Finally, adm (位) decomposition error MAPE is 24.72. Based on the Gao Si function parameterized absorption coefficient decomposition algorithm, 12 Gao Si functions are used to parameterize APN (位), and negative exponent model is used to parameterize adm (位). Using the least square optimization algorithm to solve the unknown parameters, the absorption coefficient is decomposed. Finally, the decomposition error of ADM (位) M.APE is 27.73 and RMSE is 1.02. For complex two kinds of water bodies, the two improved decomposition algorithms in this paper have successfully decomposed expp (位) with satisfactory accuracy.
【学位授予单位】：南京师范大学
【学位级别】：硕士
【学位授予年份】：2015
【分类号】：X87;P332

【参考文献】