A novel single-parameter approach for forecasting algal blooms.

Area of Science:

• Environmental Science
• Data Science
• Ecology

Background:

• Harmful algal blooms (HABs) pose a global threat, necessitating effective forecasting for proactive management.
• Current aquatic environmental monitoring systems can be costly and lack predictive accuracy.

Purpose of the Study:

• To develop and validate a novel, cost-effective single-parameter approach for accurate algal bloom forecasting.
• To enhance the accuracy and reduce the cost of HAB prediction compared to existing methods.

Main Methods:

• A Wavelet Neural Network (WNN) model was developed, integrating wavelet analysis and artificial neural networks.
• The WNN model was trained and optimized using daily online monitoring data of algal density from reservoirs in China and the U.S.A.
• Model performance was evaluated by comparing its predictive accuracy against established forecasting methods.

Main Results:

• The WNN model demonstrated high accuracy in predicting cyanobacterial cell density, with one-day-ahead predictions achieving r=0.986.
• The approach precisely forecasted algal biomass (chl a) dynamics, confirming its effectiveness for various blooming species.
• The WNN model outperformed traditional methods like artificial neural networks and autoregressive integrated moving average models in accuracy.

Conclusions:

• The novel WNN approach offers a highly accurate and cost-effective solution for forecasting harmful algal blooms.
• The system's reliance on a single, inexpensive sensor (buoy-mounted fluorescent probe) significantly reduces monitoring costs (approx. 15% of typical systems).
• This method presents a promising tool for improved HAB prediction and management worldwide.