Impact of computational methods and spectral models on the retrieval of optical properties via spectral optimization

Shaohui Huang,Shaoling Shang,Shaoping Shang,Yonghong Li

doi:10.1364/oe.21.006257

Abstract

Spectral optimization algorithm (SOA) is a well-accepted scheme for the retrieval of water constituents from the measurement of ocean color radiometry. It defines an error function between the input and output remote sensing reflectance spectrum, with the latter modeled with a few variables that represent the optically active properties, while the variables are solved numerically by minimizing the error function. In this paper, with data from numerical simulations and field measurements as input, we evaluate four computational methods for minimization (optimization) for their efficiency and accuracy on solutions, and illustrate impact of bio-optical models on the retrievals. The four optimization routines are the Levenberg-Marquardt (LM), the Generalized Reduced Gradient (GRG), the Downhill Simplex Method (Amoeba), and the Simulated Annealing-Downhill Simplex (i.e. SA + Amoeba, hereafter abbreviated as SAA). The Garver-Siegel-Maritorena SOA model is used as a base to test these computational methods. It is observed that 1) LM is the fastest method, but SAA has the largest number of valid retrievals; 2) the quality of final solutions are strongly influenced by the forms of spectral models (or eigen functions); and 3) dynamically-varying eigen functions are necessary to obtain smaller errors for both reflectance spectrum and retrievals. Results of this study provide helpful guidance for the selection of a computational method and spectral models if an SOA scheme is to be used to process ocean color images.

Highlights

Ocean color satellites have provided an unprecedented view of the global ocean owing to their ability to detect spatio-temporal patterns of bio-optical properties from space [1]
Our analysis shows that the minimized value of the target function appears slightly different when different computational methods for optimization are used
For the five computational methods tested here, all of them can find numerical solutions for each Rrs spectrum, but the computing time and the number of valid retrievals are subject to the method used and the application of boundary constraints

Summary

Introduction

Ocean color satellites have provided an unprecedented view of the global ocean owing to their ability to detect spatio-temporal patterns of bio-optical properties from space [1]. Four commonly used computational methods that can find local and global minimums are evaluated for their effectiveness in achieving optimization by applying them to a widely used data set simulated by the IOCCG Algorithm Working Group (http://www.ioccg.org/groups/OCAG_data.html). These methods are: the Levenberg-Marquardt (LM), the Generalized Reduced Gradient (GRG), the Downhill Simplex Method (Amoeba), and the Simulated Annealing-Downhill Simplex (SAA). The impact of forward spectral models on the closure between measured and modeled spectral remote-sensing reflectance, and on retrievals of spectral optical properties, is evaluated using both the IOCCG simulated data set and an in situ data set

Spectral models

Initial guesses and constrained conditions

Result of computational methods

Effects of spectral models on error functions and solutions

Findings

Conclusions