How Minervans Used Satellites to Forecast Harmful Algal Blooms in Japan

When Daria Bassot, Class of 2026, and Laoise Nolan, Class of 2025, arrived in Tokyo for Minerva University's Summer 2025 Sustainability Lab, they were fascinated with marine ecosystems growing since their first Natural Science courses. What began as curiosity about Arctic kelp forests evolved into a sophisticated machine learning framework that could help protect Japan's coastal communities from one of their most persistent environmental threats: harmful algal blooms.

From Arctic Kelp to Japanese Red Tides

Daria's interest in marine carbon sequestration started in NS166: Keeping Earth Habitable, where she first encountered the ecological significance of macroalgae. During spring 2025, she and Laoise explored the Arctic Ocean as a potential habitat for expanding kelp populations, a region where rapid environmental shifts create new ecological possibilities.

But Tokyo called for a different focus. The team pivoted to tackle Harmful Algal Blooms (HABs), locally known as "red tides," which pose serious risks to fisheries, aquaculture, and coastal ecosystems throughout Japan. "When we arrived in Tokyo, we decided to localize our focus," Daria explains. "We applied our existing knowledge about marine systems while engaging directly with a sustainability challenge that has immediate regional relevance."

Building a Predictive Model from 25 Variables

The objective was to create a system that could forecast harmful algal blooms before they devastated local fisheries and coastal communities. The students developed a machine learning framework that analyzes 25 different environmental and climatic variables, including sea surface temperature, chlorophyll a concentration, and wind speed, to predict when and where blooms would occur.

What made their approach particularly innovative was its location-specific accuracy. Rather than just predicting whether a bloom might happen somewhere in Japan's Seto Inland Sea, their system pinpointed exactly where it would strike. This granular precision makes the framework far more valuable for the fishermen, local governments, and aquaculture facilities who must make rapid decisions to protect their livelihoods.

The system is operational and flexible, with three deployment options depending on the situation: a fast version for urgent warnings, a highly accurate version for detailed analysis, and a balanced option that splits the difference. All three draw on data from satellites, ocean sensors, and weather stations to paint a comprehensive picture of ocean health.

Tokyo's Innovation Ecosystem

Working in Tokyo grounded the project in real-world urgency. A meeting with the Blue Ocean Initiative provided crucial insights into how coastal communities prioritize environmental challenges and how they intersect with their economic survival. A conversation with Umitron, an AI-powered aquaculture startup, revealed a tangible gap in existing prediction services - one that the students' model could fill.

"These encounters not only clarified the applied potential of our work," Daria reflects, "but also exposed us to Japan's broader innovation ecosystem, where environmental sustainability and technological advancement intersect in dynamic ways."

The team also participated in a two-day hackathon at Mori Building and worked on a mini-project with Tanabe city government officials, applying their satellite data skills across multiple sustainability challenges throughout their time in Japan.

The Learning Curve of Geospatial Technology

The technical journey wasn't without challenges. Working with advanced geospatial technology meant navigating complex datasets and learning through trial and error. "At first, it was challenging," Daria admits. "For example, I initially underestimated the significance of the distinction between mosaic and single-date satellite images - the former being easier to use but often compromising temporal accuracy."

Through the summer, the team discovered both the remarkable versatility and the limitations of satellite data. Cloud cover could interfere with readings. Temporal resolution had gaps. The strongest results, they learned, came from hybrid approaches - combining satellite observations with ground-based measurements to achieve both scale and precision.

"Over time, what began as technical confusion turned into genuine excitement," Daria says. "Learning to test the limits of these tools and using them creatively across different sustainability topics was one of the most rewarding aspects of the project."

Looking Forward

The students have created a manuscript of their results and are exploring partnerships with a University of Tokyo researcher and The Ocean Cleanup, a leading global nonprofit. These collaborations could validate their model and test its real-world value in guiding cleanup and prevention efforts.

Their long-term vision is to publish their findings in a peer-reviewed journal and develop a publicly available application that fisheries, aquaculture facilities, and tourism vendors across Japan can use to protect themselves from harmful algal blooms.

What started as an academic curiosity about Arctic kelp has transformed into a practical tool that could safeguard coastal ecosystems and the communities that depend on them. For Daria and Laoise, the Tokyo Sustainability Lab proved that when you combine technical skills with local partnerships and a genuine commitment to solving real problems, you can create solutions that matter.

‍

If this story inspired you to begin your own Minerva journey, start your Minerva application today.