Ioannou Ioannis, Gilerson Alexander, Gross Barry, Moshary Fred, Ahmed Samir
Optical Remote Sensing Laboratory, Department of Electrical Engineering, City University of New York, New York, New York 10031, USA.
Appl Opt. 2011 Jul 1;50(19):3168-86. doi: 10.1364/AO.50.003168.
Retrieving the inherent optical properties of water from remote sensing multispectral reflectance measurements is difficult due to both the complex nature of the forward modeling and the inherent nonlinearity of the inverse problem. In such cases, neural network (NN) techniques have a long history in inverting complex nonlinear systems. The process we adopt utilizes two NNs in parallel. The first NN is used to relate the remote sensing reflectance at available MODIS-visible wavelengths (except the 678 nm fluorescence channel) to the absorption and backscatter coefficients at 442 nm (peak of chlorophyll absorption). The second NN separates algal and nonalgal absorption components, outputting the ratio of algal-to-nonalgal absorption. The resulting synthetically trained algorithm is tested using both the NASA Bio-Optical Marine Algorithm Data Set (NOMAD), as well as our own field datasets from the Chesapeake Bay and Long Island Sound, New York. Very good agreement is obtained, with R² values of 93.75%, 90.67%, and 86.43% for the total, algal, and nonalgal absorption, respectively, for the NOMAD. For our field data, which cover absorbing waters up to about 6 m⁻¹, R² is 91.87% for the total measured absorption.
由于正向建模的复杂性以及反问题固有的非线性，从遥感多光谱反射率测量中获取水的固有光学特性具有一定难度。在这种情况下，神经网络（NN）技术在反演复杂非线性系统方面有着悠久的历史。我们采用的方法是并行使用两个神经网络。第一个神经网络用于将可用的中分辨率成像光谱仪（MODIS）可见光波段（除678纳米荧光通道外）的遥感反射率与442纳米（叶绿素吸收峰值）处的吸收系数和后向散射系数联系起来。第二个神经网络分离藻类和非藻类吸收成分，输出藻类与非藻类吸收的比率。使用美国国家航空航天局（NASA）的生物光学海洋算法数据集（NOMAD）以及我们自己在切萨皮克湾和纽约长岛海峡的现场数据集对所得的综合训练算法进行了测试。结果显示吻合度很高，对于NOMAD数据集，总吸收、藻类吸收和非藻类吸收的R²值分别为93.75%、90.67%和86.43%。对于我们覆盖吸收系数高达约6 m⁻¹的吸收性水体的现场数据，总测量吸收的R²为91.87%。
