Early Warning of Harmful Algal Blooms (HAB): A Low-Cost Integrated IoT Device with Spectrofluorometry and Automated Plankton Imaging

This paper presents the development of a prototype automated device designed to monitor sea water quality and provide early warnings of potential harmful algal bloom (HAB) outbreaks. HABs pose a significant threat to aquaculture operations and marine ecosystems due to their ability to cause mass mortalities of fish and shellfish. The device integrates a low-cost custom spectrofluorometer capable of measuring absorption and fluorescence spectra of liquid samples, and an automated plankton imager adapted from an open-source design. Key aimed sensing parameters include nutrients, chlorophyll, water temperature, and phytoplankton presence. The affordability of the device is targeted by using low-cost components and integrating the spectrofluorometer and plankton imager into a single unit driven by an embedded computer. Machine learning algorithms are employed for real-time anomaly detection from the multivariate sensor data streams to provide early alerts of potential HAB events. Initial results demonstrate the device’s ability to detect low concentrations of fluorescent dyes and phytoplankton, and the effectiveness of an adaptive anomaly detection approach on real aquarium data.