Inherent Optical Properties‐Reflectance Relationships Revisited

Marilisa Lo Prejato,David Mckee,Catherine Mitchell

doi:10.1029/2020jc016661

Abstract

AbstractUnderstanding the relationship between remote sensing reflectance, Rrs(λ) and the inherent optical properties (IOPs) of natural waters is potentially a key to improving our ability to determine biogeochemical constituents from radiometric measurements. These relationships are usually described as a function of absorption, a(λ), and backscattering, bb(λ), coefficients, with the literature providing various forms of equation operating on either bb(λ)/a(λ) or bb(λ)/[a(λ)+bb(λ)] to represent the impact of variations in light field geometries and changes in sea‐water composition. The performance of several IOP‐reflectance relationships is assessed using HydroLight radiative transfer simulations covering a broad range of Case 1 and Case 2 water conditions. While early versions of IOP‐reflectance relationships assigned variability to associated proportionality factors (e.g., f/Q) or low‐order polynomial functions, recent studies have demonstrated relationships between Rrs(λ) and bb(λ)/[a(λ)+bb(λ)] are well‐characterized by nonlinear (high‐order polynomial), monotonic functions. This study demonstrates that this approach is also valid for relationships operating on bb(λ)/a(λ) and that there is no intrinsic benefit to functions operating on bb(λ)/[a(λ)+bb(λ)] compared to bb(λ)/a(λ) for Case 2 waters, contrary to recent suggestions in the literature. In all cases it is necessary to carefully consider the performance of best fit relationships across the full range of variability of IOPs and Rrs(λ), with higher order polynomials required to enable equivalent performance across the range of natural variability. The analysis further demonstrates insignificant wavelength sensitivity across the visible region, limited sensitivity to changes in solar zenith angle and extends to relationships for below surface remote sensing reflectance, rrs(λ).

Highlights

Remote sensing reflectance, Rrs(λ), has been listed by the Global Climate Observation System as an essential climate variable thanks to its fundamental role in enabling observation of biogeochemical processes in the upper ocean (GCOS, 2003, and following updates; Groom et al, 2019)
Theoretical studies have shown that apparent optical properties (AOPs) of the water such as sub-surface and above-surface remote sensing reflectance, rrs(0−, λ) and Rrs(0+, λ) respectively, are related to inherent optical properties (IOPs) such as absorption, a(λ) (m−1), and backscattering, bb(λ) (m−1), coefficients which in turn are controlled by the types and concentrations of in-water optical constituents
The quasi-single scattering approximation (QSSA) (Gordon, 1973; Gordon et al, 1975) models are widely employed in standard ocean color processing and applications because they provide an explicit relationship between IOPs and rrs(λ), as formulated by Gordon et al (1988), coupled with a simple relationship for converting Rrs(λ) to rrs(λ) (Lee et al, 2002)

Summary

Introduction

Rrs(λ) (sr−1), has been listed by the Global Climate Observation System as an essential climate variable thanks to its fundamental role in enabling observation of biogeochemical processes in the upper ocean (GCOS, 2003, and following updates; Groom et al, 2019). Theoretical studies have shown that apparent optical properties (AOPs) of the water such as sub-surface and above-surface remote sensing reflectance, rrs(0−, λ) and Rrs(0+, λ) respectively, are related to inherent optical properties (IOPs) such as absorption, a(λ) (m−1), and backscattering, bb(λ) (m−1), coefficients which in turn are controlled by the types and concentrations of in-water optical constituents. Significant effort has been devoted to deriving both IOPs and constituent concentrations from ocean color signals (IOCCG, 2006, 2014, 2019). The quasi-single scattering approximation (QSSA) (Gordon, 1973; Gordon et al, 1975) models are widely employed in standard ocean color processing and applications because they provide an explicit relationship between IOPs and rrs(λ), as formulated by Gordon et al (1988), coupled with a simple relationship for converting Rrs(λ) to rrs(λ) (Lee et al, 2002). The inverse problem of determining IOPs from AOPs is not straightforward and relies on empirical and/or semi-analytical relationships

Objectives

Methods

Results

Conclusion