The Application Of Marine Hyperspectral Data In Shallow Bottom Detection And Red Tide Information Extraction

Hyperspectral ocean remote sensing is a technology developed in recent years. With the widely application of hyperspectral remote sensing, it is important to study on the methods for extracting information from marine hyperspectral data. In this study, methods for identification of shallow bottom types, deriving water depth and classification of red tide algae are presented based on Artificial Neural Network (ANN) and spectral derivation technology.In the study of shallow bottom and water depth, hyperspectral remote sensing reflectance (Rrs) simulation is carried out by semi-analytical numerical simulation technique of the radiative transfer. The simulated data includes two shallow bottom types: seagrass and sand. First, the data is processed with fourth-derivative analysis, and the different character of two types of shallow bottom is found according to the slope value between point 530nm and 510nm. And then, the results retrieved from ANN-based algorithm are compared with the measured data. Second, simulated data set is divided into two parts: 80% for training data set and 20% for testing data set. There are 31 inputs which correspond to remote sensing reflectances in 31 wavelengths from 400nm to 700nm and 2 outputs corresponding to bottom types and water depth. The model for deriving shallow bottom and water depth established based on ANN is also validated by measured data. Third, an ANN method for deriving shallow bottom according to a certain proportion is presented. There are 31 inputs which corresponds to remote sensing reflectance in 31 wavelengths from 400nm to 700nm and 2 outputs corresponding to the percent of two types of bottom: seagrass and sand.In the study of abstracting information of algae, firstly, the hyperspectral absorption coefficients data of five types algae data are processed with fourth-derivative analysis, and then the different character of five types algae is found according to the slope value between point 670nm and 650nm. Secondly, a method for deriving relative Karenia Mikimotoi percentage is present. The absorption coefficients of Karenia Mikimotoi are mixed with the measured absorption coefficients which do not include Karenia Mikimotoi algae with different Karenia Mikimotoi percentages. The mixed samples are used to develop the ANN-based method to derive the Karenia Mikimotoi percentage. The results of the study show that the derived concentration of chlorophyll of Karenia Mikimotoi is consistent with the measured concentration of Karenia Mikimotoi.