Enhanced detection of harmful algal blooms using topological data analysis for clustering spatially distributed water quality and hydrodynamic data

Harmful algal blooms (HABs), particularly those caused by cyanobacteria, pose significant ecological and health risks. This study explores the application of topological data analysis (TDA) to cluster high-dimensional water quality and hydrodynamic data. This is used to identify regions prone to cyanobacterial harmful algal blooms (cyanoHABs) in the confluence area of the Nam and Nakdong rivers. This study uses diverse in-situ measurements, including bathymetry, velocity, temperature, electrical conductivity, dissolved oxygen, pH, chlorophyll-a concentration, turbidity, sediment grain size, and suspended sediment concentration. The TDA-mapper is employed to visualize and analyze these complex data structures, excluding direct algal bloom information and focusing on other water quality parameters. The TDA-mapper facilitates the identification of local clusters and boundary regions. It outperforms conventional clustering methods, such as k-means, spectral clustering, and agglomerative clustering, in discerning locational water characteristics and predicting high risk areas for cyanoHABs. Despite some computational complexity and the need for domain-specific hyperparameter tuning, the TDA-mapper is a valuable tool for environmental monitoring and management. These results underscore the potential of the TDA-mapper in providing actionable insights for the early detection and mitigation of HABs.