Modelling Atmospheric Correction for Multispectral Images Above Water Surfaces

Abstract

Satellite ocean color remote sensing has emerged as a vital tool for large-scale, continuous monitoring of water quality parameters such as chlorophyll-a (Chl-a), total suspended solids (TSS), Non Algal Particles (NAP), and colored dissolved organic matter (CDOM). However, the applicability and accuracy of existing global algorithms are significantly constrained over optically complex waters, particularly in regions like the Indo-Arabian Seas, comprising the Arabian Gulf, Red Sea, Gulf of Oman, Arabian Sea, and the northwestern Indian Ocean. These waters are characterized by intense monsoonal dynamics, frequent dust-laden aerosol events, high sediment loads, and variable biogeochemical conditions, all of which introduce substantial challenges to atmospheric correction (AC) and bio-optical modeling efforts. This research addresses these challenges through the development of a regionally optimized, multi-stage framework that improves the retrieval of water quality parameters from satellite observations. The study begins with a critical evaluation of seven state-of-the-art AC processors applied to Sentinel-2 MSI imagery. Using in-situ hyperspectral data and optical water type classification, the performance of each algorithm is assessed across clear to moderately turbid conditions. The findings emphasize the importance of water-type-specific AC selection for reliable remote sensing reflectance retrievals and downstream water quality products. To further understand the spectral behavior of sediment-laden waters, controlled laboratory experiments were conducted using natural sediments with varying particle size distributions, mineralogical compositions, and concentrations. The measured reflectance data were combined with 6SV-based radiative transfer simulations to analyze top-of-atmosphere (TOA) spectral responses under different atmospheric conditions and sensor configurations. This analysis informed the construction of sediment-specific spectral libraries and guided the assessment of sensor design requirements for sediment detection in coastal zones. Building upon these insights, a regionally tuned atmospheric correction model - GIO-RAC (Gulf–Indian Ocean Regional Atmospheric Correction) - was developed. Monthly aerosol models were constructed using long-term AERONET climatology and integrated into radiative transfer-based LUTs. These were implemented within the SeaDAS processing chain to enhance Rrs retrievals from MODIS and Sentinel-3 OLCI, resulting in improved spectral fidelity across diverse aerosol and water conditions. The final component of this study focuses on the generation and application of a regionally representative synthetic spectral lookup table (LUT) for qualitative optical water type (OWT) classification. The LUT, constructed using HydroLight radiative transfer simulations, encapsulates realistic regional optical variability and was clustered into 23 distinct OWTs using a fuzzy classification approach. These water types were validated against in situ and satellite-derived reflectance spectra across the Arabian Gulf, Red Sea, and western Indian Ocean, confirming the LUT’s representativeness for diverse aquatic regimes. Complementing this, an empirical wavelength-resolved model was developed for estimating phytoplankton absorption coefficients (aph(λ)) from satellite reflectance, providing a practical tool for characterizing phytoplankton dynamics in optically complex waters. Collectively, this thesis enhances the ocean color community’s capability to monitor sediment-influenced and aerosol-impacted marine systems by introducing regionally adapted atmospheric correction and bio-optical modeling tools grounded in physical realism. The outputs of this research - including monthly aerosol models, synthetic spectral LUTs, OWT maps, and empirical aph algorithms - offer foundational resources for operational water quality assessment, ecosystem monitoring, and the future design of sensor missions targeting turbid and spectrally diverse aquatic environments.

Date of Award	2025
Original language	American English
Supervisor	Maryam Alshehhi (Supervisor)