Coastal water quality in Australia is increasingly influenced by climatic shifts and anthropogenic pressures, presenting significant challenges for marine ecosystem management. Traditional in situ measurements have provided valuable insights into coastal ocean bio-geo-physical properties, but their limitations in spatial coverage and long-term monitoring restrict their ability to capture large-scale, multi-decadal changes. This study utilizes a unique dataset of in-situ measurements collected between 2003 and 2023 along the Australian coastline to develop advanced monitoring tools for assessing long-term water quality trends. By integrating 264 spectral datasets from field campaigns, we train a deep learning framework—Deep Learning for Aquatic Remote Sensing (DL-RS)—to model bio-optical relationships across diverse marine environments. The DL-RS model is then applied to multi-decadal MODIS Aqua satellite imagery (2002-2023) to analyse long-term spatio-temporal variations in coastal water quality. Rigorous model evaluation demonstrates its high reliability in tracking key parameters, such as Total Suspended Solids (TSS), Dissolved Organic Carbon (DOC), and Chl-a concentrations, even across regions with significant bio-optical variability. Spatio-temporal analysis in this study identifies geographical hotspots and regions experiencing substantial long-term trends in water quality changes, providing valuable insights into the impacts on Australian coastal ecosystems over extended periods.