Prediction of algal blooms using genetic programming.

Area of Science:

• Marine Biology
• Environmental Science
• Computational Intelligence

Background:

• Algal blooms pose significant ecological and economic challenges.
• Predictive modeling is crucial for managing and mitigating bloom impacts.
• Genetic Programming (GP) offers a powerful tool for complex environmental modeling.

Purpose of the Study:

• To develop a mathematical model for predicting algal blooms in Tolo Harbor, Hong Kong.
• To propose a systematic procedure for selecting the most appropriate GP model from multiple feasible options.
• To identify key environmental variables influencing algal bloom dynamics.

Main Methods:

• Utilized Genetic Programming (GP) for mathematical modeling and prediction of algal blooms.
• Employed chlorophyll as a measure of algal bloom biomass, with eight other variables as inputs.
• Developed a data-driven approach combined with physical understanding to select the optimal GP model.

Main Results:

• GP successfully evolved multiple models with comparable error metrics.
• The study identified ecologically significant variables affecting algal blooms.
• Chlorophyll, total inorganic nitrogen, and dissolved oxygen were key predictors for short-term (1-week) predictions.
• Secchi-disc depth and temperature became significant for longer-term (biweekly) predictions, alongside chlorophyll.

Conclusions:

• The developed GP model provides a robust method for predicting algal blooms.
• The proposed model selection procedure enhances the reliability of GP-based environmental predictions.
• Understanding the influence of specific variables like chlorophyll, nitrogen, and oxygen is vital for effective bloom management in Tolo Harbor.