Development of an explicit algorithm for remote sensing estimation of chlorophyll a using symbolic regression

Area of Science:

• Oceanography
• Remote Sensing
• Bio-optics

Background:

• Accurate marine bio-optical properties are crucial for ocean color remote sensing.
• Existing algorithms for chlorophyll a (chl a) estimation have limitations, particularly in optically complex waters.

Purpose of the Study:

• To develop and evaluate a novel algorithm using symbolic regression for estimating chl a concentrations from satellite remote sensing reflectance.
• To compare the accuracy and computational efficiency of the new algorithm against established explicit (OC4v4, OC4v6) and implicit (neural networks, SVM) algorithms.

Main Methods:

• Symbolic regression was employed to derive an explicit algorithm for chl a estimation.
• The performance of the symbolic regression algorithm was assessed against OC4v4, OC4v6, neural network, and SVM-based algorithms.
• Accuracy and computational efficiency were the primary metrics for comparison.

Main Results:

• The symbolic regression algorithm demonstrated higher accuracy than OC4v4 and OC4v6 algorithms.
• Its accuracy was comparable to that of implicit algorithms (neural networks, SVM).
• Significant accuracy improvements were observed in high biomass areas (chl a ≥ 3 mg m(-3)) within optically complex waters.
• Computational efficiency was comparable to OC4 algorithms and superior to SVM-based algorithms.

Conclusions:

• Symbolic regression provides a powerful and accurate tool for remote sensing of chl a concentrations.
• The algorithm offers a favorable balance of precision and processing speed, suitable for large-scale data reprocessing.
• This approach enhances the capability of ocean color remote sensing, particularly for challenging aquatic environments.