What is a Harmful Algal Bloom?

Algae are photosynthetic microorganisms found in virtually every aquatic ecosystem on Earth. Under normal conditions, they form the base of marine food webs and produce roughly half of the world's oxygen. However, when environmental conditions align [excess nutrients, warm temperatures, calm water, and sufficient sunlight], populations of certain algal species can expand explosively. These events are known as Harmful Algal Blooms (HABs).

HABs can produce powerful toxins that accumulate in shellfish and fish, contaminate drinking water supplies, trigger fish kills by depleting dissolved oxygen, and cause respiratory distress in nearby communities. As climate change drives warmer ocean temperatures and altered precipitation patterns that carry more agricultural runoff into coastal waters, HABs are becoming more frequent, more widespread, and harder to predict.

Meet Lingulodinium polyedra

Lingulodinium polyedra (L. poly) is a bioluminescent dinoflagellate responsible for the vivid red tides that periodically wash over the coastline of Southern California, particularly around San Diego. During the day, dense blooms turn the surf a rust-red or brown color. At night, wave action triggers a brilliant blue bioluminescence, a phenomenon that draws crowds to the beach but signals a serious ecological imbalance.

L. poly blooms are driven by a combination of environmental triggers. Sea surface temperature is one of the strongest predictors: this species thrives in the 17-22 C range typical of late summer and early fall in San Diego. Ocean pH, salinity, dissolved oxygen, and chlorophyll concentration all interact to either promote or suppress bloom development. Understanding how these variables relate to bloom intensity is essential for early warning systems that could protect fisheries, beaches, and public health.

Our Approach: Machine Learning Meets Ocean Sensors

Our project uses continuous environmental sensor data collected from Scripps Institution of Oceanography's pier in La Jolla, CA. Each day, instruments record a suite of ocean and atmospheric measurements. We combine these readings with historical L. poly cell count observations to train a machine learning model that estimates bloom concentration [expressed as expected mL of L. poly per liter of seawater].

The goal is to identify which environmental variables are the strongest early indicators of a bloom, and whether sensor data alone can provide enough signal for reliable short-term predictions, without requiring costly, time-consuming manual water sampling.

About the Visualization

The interactive bubble chart below plots 804 days of data spanning October 2022 through December 2025. Each bubble represents a single day and encodes four variables simultaneously:

Use the Play button to animate through time and watch bloom events develop. The 14-day trail shows how conditions evolve before and during a bloom. You can also drag the slider to scrub directly to any date, or adjust the playback speed.

Look for large bubbles clustering inside the shaded "typical bloom temp range" band, and notice how bloom events tend to follow periods of rising temperature and shifting pH.