Dash et al. (2025) Prophet-Based Artificial Intelligence Versus Seasonal Auto-Regressive Models for Flood Forecasting with Exogenous Variables

⚠️ Warning: This summary was generated from the abstract only, as the full text was not available.

Identification

• Journal: Water
• Year: 2025
• Date: 2025-12-15
• Authors: Adya Aiswarya Dash, Edward McBean
• DOI: 10.3390/w17243551

Research Groups

Not explicitly mentioned in the provided text.

Short Summary

This study compares SARIMAX and Prophet models for streamflow forecasting, demonstrating Prophet's superior accuracy and ability to capture non-linear dynamics over SARIMAX, particularly for short-term horizons, for flood risk management.

Objective

• To compare the performance of the Seasonal Auto-Regressive Integrated Moving Average with Exogenous Regressors (SARIMAX) and Prophet models for streamflow forecasting across 3-day and 15-day horizons, evaluating their effectiveness in flood risk management.

Study Configuration

• Spatial Scale: Not explicitly defined, implied to be a single-point or localized streamflow forecasting context.
• Temporal Scale: 3-day and 15-day forecasting horizons.

Methodology and Data

• Models used: Seasonal Auto-Regressive Integrated Moving Average with Exogenous Regressors (SARIMAX), Prophet (decomposable time-series forecasting model).
• Data sources: Not explicitly defined, implied to be historical streamflow observations and associated exogenous predictor data.

Main Results

• SARIMAX showed limited effectiveness, producing wide uncertainty (177.7%) and high errors (Mean Absolute Error (MAE) = 153.73; Root Mean Square Error (RMSE) = 207.10) with a negative coefficient of determination (R²) (–4.42) for longer horizons.
• For shorter horizons, SARIMAX performance remained limited (uncertainty = 28.04%; MAE = 61.52; RMSE = 94.88; R² = –0.14).
• Prophet achieved significantly lower uncertainty (16%), high accuracy (R² = 0.95), and exceptional performance on short-term forecasts (R² = 0.99).
• Prophet effectively captures non-linear dynamics and exogenous influences, outperforming SARIMAX across all evaluated horizons.

Contributions

• Provides a direct comparison highlighting the superior performance of the Prophet model over the traditional SARIMAX model for operational streamflow forecasting.
• Demonstrates Prophet's capability to effectively capture non-linear dynamics and exogenous influences, which are often missed by conventional statistical models like SARIMAX.
• Offers insights for enhancing flood risk management and preparedness through the adoption of more advanced, AI-based forecasting models.

Funding

Not mentioned in the provided text.

Citation

@article{Dash2025ProphetBased,
  author = {Dash, Adya Aiswarya and McBean, Edward},
  title = {Prophet-Based Artificial Intelligence Versus Seasonal Auto-Regressive Models for Flood Forecasting with Exogenous Variables},
  journal = {Water},
  year = {2025},
  doi = {10.3390/w17243551},
  url = {https://doi.org/10.3390/w17243551}
}