AI-powered Forecast for Harmful Algal Blooms

Contaminants of emerging concern (CECs) were identified as a global water quality challenge with potential threats to human health and ecosystems. Significant attention was directed towards cyanotoxins in aquatic sources. These toxins, released by certain cyanobacteria during harmful algal blooms (HABs), negatively impacted water quality and affected humans and wildlife. Distinguishing harmful cyanobacteria from non-toxic ones was crucial as they appear similar to the human eye.

To address this issue, the project focused on accurately characterizing subtle visual changes such as color, texture, and translucency. With the advent of new satellites like Landsat-8 and Sentinel-2, opportunities arose for water color remote sensing, despite the challenges posed by higher spatial, spectral, and temporal resolution.

This project developed a HAB monitoring system for inland and coastal waters using satellite and drone-borne hyperspectral imaging. This system supported an AI-powered model for forecasting water quality by developing indirect measurements of pollutant presence.

The research took a holistic, multi-disciplinary approach, incorporating expertise in water quality, river hydraulics, coastal hydrodynamics, microbiology, earth observation, AI, color science, and computer vision.