神经网络法反演海水叶绿素浓度的分析

海洋水色遥感的最终目的之一是监测海洋初级生产力的时空变化,而反映海洋初级生产力的一个重要指标就是浮游植物中的叶绿素浓度1]。对于海水叶绿素浓度的遥感反演研究工作至今已进行了30多年,方法主要是基于蓝-绿波段比值的经验统计法。近些年,随着水色遥感器的改进及数据处理方法的深入研究,提出了荧光高度法2]和神经网络法3-5]。本文基于SeaBAM(SeaWiFSBio-OpticalAlgorithmMini-Workshop)小组搜集的全球范围叶绿素浓度与离水辐射率的同步观测数据,利用神经网络方法反演海水叶绿素浓度,并将其结果与SeaBAM经验算法进行了比较及分析。

A Study on the Remote Sensing Inversion of Ocean Chlorophyll Concentration by Using Neural Network Method

One of the tasks of ocean color remote sensing is the measurement of marine primary production variation with time and space,whose important index is chlorophyll concentration of phytoplankton. The study on the inversion model of chlorophyll by remote sensing has been carried out for more than 30 years,and the main method is based on the ratio of blue and green channels. With the development of new ocean color remote sensors and new inversion methodology,the method based on fluorescence line high and neural network has being developed. In this paper,the neural network inversion is studied by means of the data set of global chlorophyll and in-situ radiance which was collected by SeaWiFS Bio-Optical Algorithm Mini-Workshop (SeaBAM). Finally,the result by neural network in the study is compared with that by Sea BAM’s empirical inversion.