Error decomposition and estimation of inherent optical properties.

Area of Science:

• Oceanography
• Remote Sensing
• Optical Oceanography

Background:

• Ocean-color remote sensing provides crucial data on marine biogeochemistry.
• Inherent Optical Properties (IOPs) are key parameters derived from ocean-color data.
• Quantifying errors in IOPs is essential for accurate interpretation of oceanographic data.

Purpose of the Study:

• To develop and validate a methodology for quantifying and separating errors in IOPs derived from ocean-color model inversion.
• To decompose total error into model approximations, sensor noise, and atmospheric correction components.
• To establish a generic method for error quantification applicable to various ocean-color products.

Main Methods:

• Decomposition of total error into stochastic components: model, sensor, and atmospheric correction.
• Utilizing prior information on observations, sensor noise, and goodness-of-fit to derive posterior probability distributions of IOPs.
• Validation using the International Ocean Colour Coordinating Group (IOCCG) and NASA bio-Optical Marine Algorithm Data set (NOMAD).

Main Results:

• The proposed method accurately quantifies error sources, with derived errors showing 60-90% correlation with known values for IOCCG and NOMAD datasets.
• Model-induced errors range from 10% to 57%, while atmospheric-induced errors constitute 43% to 90% of the total error.
• Mean relative errors for derived IOPs are between 2% and 20%, with a specific error table developed for the MERIS sensor.

Conclusions:

• The developed methodology provides a more accurate estimation of ocean-color derived product errors compared to previous methods.
• The approach is versatile and can be applied to quantify errors for any derived biogeophysical parameter.
• This work enhances the reliability of ocean-color data for scientific applications.