Improving algal bloom modelling for lake management using remote sensing and process-based modelling approaches

Abstract

Freshwater ecosystems are increasingly threatened by eutrophication and harmful algal blooms, which negatively impact water quality and security. As freshwater availability declines globally, the need for effective monitoring and forecasting systems becomes more urgent. In the face of climate change, algal blooms are projected to increase in bloom frequency and severity through rising temperatures and altered hydrological cycles. Current monitoring approaches, based on in situ sampling and sensors, are accurate but resource intensive and rarely provide forecasting capabilities. Forecast models, while potentially useful for proactive water management, are often limited by data availability due to low frequency or inconsistent datasets. Remote sensing offers a scalable, cost-effective alternative for chlorophyll-a monitoring, a proxy for algal biomass, but its integration into operational forecasting remains limited due to data quality concerns and methodological challenges. This thesis investigates how we can use and advance algal bloom modelling techniques for water management under present and future climate conditions. It aims to incorporate existing data sources, such as remote sensing, to enhance process-based model calibration and improve the forecasting of algal blooms. Through three data-driven chapters, we explore the roles of data availability, remote sensing data, and model transferability in forecasting chl-a concentrations and bloom events across UK waterbodies. In Chapter 2, the relative importance of different input data for model calibration and bloom detection is assessed. Demonstrating high-frequency calibration data to improve predictions across the full chl-a spectrum, while high-frequency meteorological data are particularly critical for detecting bloom events. Chapter 3 focuses on improving model calibration with remote sensing data as a supplement to in situ monitoring. Showing that remote sensing can significantly enhance chl-a forecasts and bloom detection when subjected to robust quality filtering. Highlighting the potential for remote sensing data to improve model output and make calibration available in water bodies with low data availability. A novel, easy-to-implement quality control approach is introduced using conformal prediction to improve the reliability of satellite-derived inputs. Chapter 4 evaluates climate change impacts on three UK waterbodies using high-frequency calibrated models. By spatially relocating models across a latitudinal gradient, it is shown that northern reservoirs are likely to experience the greatest increases in chl-a under all climate scenarios. This novel approach provides a unique opportunity to explore how latitude,local waterbody characteristics, and climate interact to influence bloom dynamics across the UK. These findings underscore the importance of tailoring mitigation strategies to both geographic location and waterbody type. Together, the data chapters demonstrate that reliable forecasts of algal blooms can be achieved across different water body types and data frequencies, advancing our understanding of how data quantity and quality influence model performance. The results show that satellite Earth observation data can effectively support the calibration of lake ecosystem models, enabling their application in systems where in situ monitoring is currently limited or absent. By integrating remote sensing with process-based models, this approach offers a scalable and cost-effective way to fill data gaps while preserving site-specific characteristics. The modelling framework also proves valuable for exploring the effects of climate change on bloom dynamics across diverse lakes and regions, highlighting its broader utility for water management. Strengthening these tools will aid in building climate resilience across freshwater management systems.