Algal Bloom Prediction Using Extreme Learning Machine Models at Artificial Weirs in the Nakdong River, Korea

Area of Science:

• Environmental Science
• Water Quality Monitoring
• Ecological Risk Assessment

Background:

• Algal blooms, indicated by chlorophyll-a concentration, pose risks to environmental management and public health.
• The Nakdong River in Korea experiences harmful annual algal blooms.
• Accurate prediction of algal blooms is crucial for effective mitigation strategies.

Purpose of the Study:

• To design and evaluate intelligent models for predicting chlorophyll-a concentration.
• To compare the performance of extreme learning machine (ELM) models with traditional methods.
• To assess the impact of incorporating upstream data on prediction accuracy.

Main Methods:

• Development of two ELM models: ELM1 (downstream data) and ELM2 (upstream and downstream data).
• Utilized weekly data from January 2013 to December 2016, including environmental factors and chlorophyll-a.
• Compared ELM models against multiple linear regression (LR), neural network with backpropagation (NN-BP), and adaptive neuro-fuzzy inference system (ANFIS).

Main Results:

• The ELM2 model, which included upstream chlorophyll-a data, significantly outperformed the ELM1 model.
• ELM2 demonstrated superior prediction and generalization capabilities compared to LR, NN-BP, and ANFIS.
• The study confirms the effectiveness of ELM models for algal bloom prediction.

Conclusions:

• ELM models offer a robust and efficient approach for predicting chlorophyll-a concentration and algal blooms.
• Incorporating upstream data enhances the predictive accuracy of ELM models.
• The findings support the use of ELM for environmental management and risk assessment in affected water bodies.