Chlorophyll-a Mapping in a Large Lake Using Remote Sensing Imagery: A Case Study of Western Lake Ontario

Abstract

Western Lake Ontario (WLO) and Hamilton Harbour (HH) experience significant eutrophication challenges. Despite an overall decrease in the limiting nutrient phosphorus (P) inputs, recurrent nuisance (Cladophora) and cyanobacterial harmful algal blooms (cHABS) are observed in nearshore hotspots of WLO and HH, respectively. These events hint at a complex interplay of contributing factors including not only of P availability but nutrient enrichment in general, as well as invasive mussel species altering ecosystem dynamics, climate change, and other anthropogenic influences. As a result, continued and consistent monitoring is of paramount importance. Eutrophication in WLO and HH is also linked to the expanding urbanization within the Golden Horseshoe, which includes the Greater Toronto Area (GTA), along with nutrient point and nonpoint load sources from stormwater management systems and agricultural watersheds. Of importance are also the nutrient inputs flowing from Lake Erie through the Niagara River creating local productivity zones at the river mouth. Traditional field-based monitoring methods face limitations, including high costs, labour intensity, limited temporal resolution and inadequate spatial coverage. In that respect, remote sensing (RS) may offer an alternative approach, leveraging the water colour (optical properties) to detect optically active constituents (OACs) like Chlorophyll-a (Chl-a) that can provide proxies for phytoplankton abundance in algae. The distinct spectral signatures of Chl-a make multi-spectral imagery a valuable tool for water quality assessment that can complement ongoing in-situ monitoring. This thesis presents a comprehensive analysis aimed at enhancing the capacity for monitoring nearshore algal blooms in the oligo-mesotrophic WLO and eutrophic HH through publicly available high-spatial-resolution (< 100 m) RS satellites data, specifically Landsat 5, 7, 8, 9, and Sentinel-2. The research explores the optimal combinations of atmospheric correction methods and reflectance indexes to develop semi-empirical based Chl-a retrieval models specific to the (sub)regions considered. As an additional application, the satellite based Chl-a data are used to assess the spatial-temporal variability and trends of algal productivity over the past decade, identifying productivity hotspots and anomalies. The thesis is structured in five chapters, beginning with a general introduction in Chapter 1, followed by Chapter 2, which offers the necessary background for understanding the research presented in the thesis. Chapters 3 and 4 delve into comparative evaluations of Chl-a retrieval methods and time-series analysis of algal bloom dynamics, respectively. The thesis ends with Chapter 5, which synthesizes the main findings and offers conclusions and future research directions. Chapter 3 presents a comprehensive comparative evaluation of atmospheric correction processors and reflectance indexes, assessing their performance in Chl-a concentration retrieval from a multi-platform collection of satellite data. By analyzing satellite scenes from different platforms alongside in-situ measured Chl-a data, the chapter develops predictive linear regression models. The results highlight the superior performance of certain combinations, particularly ACOLITE-corrected Landsat 8 and Sentinel-2 imagery utilizing two band ratio indexes, that is blue-to-green or blue-to-red, in capturing Chl-a concentration with acceptable accuracy. Delving into the Chl-a dynamics, Chapter 4 presents a time-series analysis using Landsat 8 and 9 imagery from 2013 to 2023, to reconstruct the spatial-temporal patterns and hotspots in WLO and HH. After preprocessing a collection of Level-1 images with the optimal combination of atmospheric correction method and retrieval index, as identified in Chapter 3, a time-series collection of estimated Chl-a concentration maps are produced. By applying three algal growth indicators, namely bloom intensity, extent, and severity, along with averaging annual and monthly estimated Chl-a concertation maps and conducting a Mann-Kendall trend analysis, we are able to examine algal bloom dynamics, seasonality, and delineate areas of concern. The results should help in planning monitoring and design eutrophication management strategies for the region. The findings from this thesis underscore the potential of space-borne RS in advancing water quality monitoring that can inform management practices. By identifying the most effective methods for Chl-a concentration retrieval and providing a nuanced understanding of algal growth dynamics, the research in this thesis contributes to both fields of aquatic RS and water quality monitoring. The comparative analyses, model developments, and spatial-temporal investigations not only offer practical tools for water quality assessment but also set the stage for future studies leveraging machine learning and existing satellite datasets. The work demonstrates the critical role of tailored RS applications in addressing eutrophication issues, advocating for integrated monitoring approaches to sustain aquatic ecosystems in the face of changing environmental conditions.