When flood forecasting is limited to estimating the moment a threshold is exceeded, researchers often concentrate on predicting water levels or river discharge in a particular location.
Hydrodynamic models, such as the Hydrologic Engineering Center's River Analysis System (HEC-RAS) or the MIKE suite of models, simulate water flow and its interaction with the surrounding environment, providing detailed predictions of flood extent, depth, and velocity.
The choice of the most suitable approach depends on factors such as data availability, catchment characteristics, and desired prediction accuracy.
Here is an overview of each approach: Physically-based models simulate the underlying physical processes involved in flood generation and propagation, such as precipitation, infiltration, runoff, and routing.
Data-driven models focus on discovering patterns and relationships within historical data without explicitly representing the physical processes.
They can learn complex, non-linear relationships and adapt to changing conditions, making them useful in situations where data is abundant and accurate representation of physical processes is challenging.
Examples of data-driven models include regression techniques, Artificial Neural Networks (ANN), Support Vector Machines (SVM), and tree-based algorithms like Random Forest or XGBoost.
Real-time flood forecasting at regional area can be done within seconds by using the technology of artificial neural network.