Keeping Humans in the Loop Improves Flood Forecasting

Source Domain: eos.org

Integration of Advanced Technologies: Artificial intelligence (AI) and machine learning (ML) have been increasingly incorporated into flood prediction systems to handle large datasets and identify complex patterns.
Limitations of AI and ML: While AI and ML are faster and more efficient, they require extensive data and may struggle to capture rare, extreme events which traditional models are better at.
Comparison with Traditional Systems: Most studies evaluating AI and ML in flood forecasting use historical models, lacking real-time operational data that would be used in emergency situations which might show more limitations.
Performance of Forecasters-in-the-Loop: A study comparing ML models to the actual flood forecasting system used at the California Nevada River Forecast Center suggests that the human-supervised approach outperforms ML systems in streamflow predictions and flood event detection.
Human Expertise Advantage: The study showed that human forecasters can correct errors and manage substandard data inputs, which AI and ML systems cannot do, making them more reliable for flood forecasting.
Call for Mixed Approach: Automated forecasting in flood prediction is promising but should be supplemented with human expertise to provide the most accurate predictions and better protect communities from floods.

You may have missed