Towards high fidelity mapping of global inland water quality using earth observation data

This body of work aims to contribute advancements towards developing globally applicable water quality retrieval models using Earth Observation data for freshwater systems. Eutrophication and increasing prevalence of potentially toxic algal blooms among global inland water bodies have become a major ecological concersn and require direct attention. There is now a growing necessity to develop pragmatic approaches that allow timely and effective extrapolation of local processes, to spatially resolved global products. This study provides one of the first assessments of the state-ofthe-art for trophic status (chlorophyll-a) retrievals for small water bodies using Sentinel-3 Ocean and Land Color Imager (OLCI). Multiple fieldwork campaigns were undertaken for the collection of common aquatic biogeophysical and bio-optical parameters that were used to validate current atmospheric correction and chlorophyll-a retrieval algorithms. The study highlighted the difficulties of obtaining robust retrieval estimates from a coarse spatial resolution sensor from highly variable eutrophic water bodies. Atmospheric correction remains a difficult challenge to operational freshwater monitoring, however, the study further validated previous work confirming applicability of simple, empirically derived retrieval algorithms using top-of-atmosphere data. The apparent scarcity of paired in-situ optical and biogeophysical data for productive inland waters also hinders our capability to develop and validate robust retrieval algorithms. Radiative transfer modeling was used to fill this gap through the development of a novel synthetic dataset of top-of-atmosphere and bottom-of-atmosphere reflectances, which attempts to encompass the immense natural optical variability present in inland waters. Novel aspects of the synthetic dataset include: 1) physics-based, two-layered, size and type specific phytoplankton IOPs for mixed eukaryotic/cyanobacteria 6 assemblages, 2) calculations of mixed assemblage chl-a fluorescence, 3) modeled phycocyanin concentration derived from assemblage based phycocyanin absorption, 4) and paired sensor-specific TOA reflectances which include optically extreme cases and contribution of green vegetation adjacency. The synthetic bottom-of-atmosphere reflectance spectra were compiled into 13 distinct optical water types similar to those discovered using in-situ data. Inspection showed similar relationships and ranges of concentrations and inherent optical properties of natural waters. This dataset was used to calculate typical surviving water-leaving signal at top-of-atmosphere, as well as first order calculations of the signal-to-noise-ratio (SNR) for the various optical water types, a first for productive inland waters, as well as conduct a sensitivity analysis of cyanobacteria detection from top-of-atmosphere. Finally, the synthetic dataset was used to train and test four state-of-the-art machine learning architectures for multi-parameter retrieval and cross-sensor capability. Initial results provide reliable estimates of water quality parameters and inherent optical properties over a highly dynamic range of water types, at various spectral and spatial sensor resolutions. It is hoped the results of this work incrementally improves inland water Earth observation on multiple aspects of the forward and inverse modelling process, and provides an improvement in our capabilities for routine, global monitoring of inland water quality