Harmful algal blooms caused by cyanobacteria are a growing threat to water quality and public health around the world. One way to track these blooms is by measuring a pigment called phycocyanin, which is found in cyanobacteria. Phycocyanin levels can be estimated using satellite images, but current methods face challenges because of limited quantity and quality(noise) of data.

In this study, we developed a new AI model that combines information from satellite images with physics-based measurements of how light behaves in water. This approach helps the model better understand the unique conditions in different lakes and improves its ability to work across many locations. We tested the model using data collected in the summer of 2024 from seven lakes in the U.S., where we measured water quality at the same time NASA’s new satellite, PACE, flew overhead.

Using this satellite and our new model, we created some of the first maps showing phycocyanin levels from PACE data. Our results show that this method improves accuracy and can be used to monitor harmful algal blooms more reliably. With its global reach and daily imaging capability, the PACE satellite opens new possibilities for tracking water quality and protecting freshwater ecosystems worldwide.