DAtect: Forecasting Domoic Acid Levels from Algal Blooms in the Pacific Northwest Coast

Overview

In spring 2015, the U.S. Pacific Coast experienced an unprecedented spike in domoic acid concentrations, surpassing federal safety limits by over 700% in parts of Washington State and triggering widespread fishery closures. Domoic acid, a potent neurotoxin from harmful algal blooms of Pseudo-Nitzschia, poses significant risks to human health, causing amnesic shellfish poisoning, and is fatal to marine wildlife. Previous forecasting methods relied heavily on qualitative approaches, lacking quantitative accuracy, and prior machine learning models were limited in geographic and temporal scope. This project developed a machine learning model to quantify and forecast domoic acid risk across the Pacific Northwest. A comprehensive 21-year database was compiled, incorporating weekly biotic beach sampling data, environmental factors, and satellite observations to train and evaluate various forecasting algorithms. Through rigorous feature selection, key factors influencing domoic acid levels—such as temporal features, Pacific Ocean indices, latitude, and environmental metrics related to harmful algal blooms—were identified. The final model, a random forest regression and classification algorithm called DAtect, achieved a relative mean squared error of 10.57 and an R-squared of 0.57 for direct concentration forecasting. When assessing threat levels against the regulatory limit of 20 ppm, DAtect reached an accuracy of 0.79. This model offers potential to enhance previous qualitative forecasts, enabling proactive measures to protect shellfish fisheries and public health along the Pacific Coast.

Competition history

• AJAS 2025 Category not listed

Related projects

Closest projects by meaning, across every fair and year in the corpus.

Browse more like this

• All Earth and Environmental Sciences projects
• AJAS Earth and Environmental Sciences projects
• Every AJAS 2025 project