Optical remote sensing of waters with vertical structure

摘要

Optical remote sensing of ocean color is a well-established technique that is used to produce maps of marine constituents on a routine basis. Retrieval algorithms used to infer pigment concentrations from measurements of ocean color are usually based on the assumption that the upper ocean column is vertically homogeneous. However, stable stratification of the water column is often encountered in coastal waters and in fjords. This stratification is decisive for the initiation, maintainance, and species composition of phytoplankton blooms. Here we present an optical remote-sensing algorithm with the ability to resolve such a vertical structure of oceanic waters. The vertical structure is assumed to consist of two homogeneous layers with different concentrations of chlorophyll a. The algorithm is designed to determine the chlorophyll-a concentrations of the two layers as well as the thickness of the upper layer. These three parameters influence the ocean color and are simultaneously retrieved through an inverse-modeling technique. This technique consists of using radiative-transfer computations for a coupled atmosphere-ocean system to simulate radiances received in various bands of the satellite sensor and to compare these simulated results with measured radiances. The sum of absolute values of differences between simulated and measured radiances is minimized by use of an optimization algorithm, and the retrieved parameters are those that yield the minimum sum of differences between measured and simulated data. The optimization algorithm that we used in our study is the simulated annealing method, which is an extension of the downhill simplex algorithm. In this study the algorithm was tested on synthetic data generated by the forward model. The results indicate that it should be possible to retrieve vertical variations in the pigment concentration. The synthetic data were generated for spectral bands that coincide with those of the Medium Resolution Imaging Spectrometer sensor, which will be a part of the mstrument package of the upcoming Environmental Satellite.