Farzin et al. (2025) Integrating Deep Learning and Copula Models for Flood–Drought Compound Analysis in Iran

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

Identification

• Journal: Water
• Year: 2025
• Date: 2025-12-08
• Authors: Saeed Farzin, Mahdi Valikhan Anaraki, Mojtaba Kadkhodazadeh, Amirreza Morshed-Bozorgdel
• DOI: 10.3390/w17243479

Research Groups

Not specified in the provided text.

Short Summary

This study developed an integrated framework combining U-Net++, quantile mapping, and Copula models to forecast the combined impacts of drought and flood under future climate change scenarios. The framework demonstrated superior performance in downscaling river flows and projected increased vulnerability to compound extreme events in future periods (2025 and 2071).

Objective

• To forecast the combined impacts of drought and flood in the future using an integrated framework.

Study Configuration

• Spatial Scale: Large-scale gridded data, specific spatial extent not detailed.
• Temporal Scale: Historical (1985–2014) and future projection periods (2021–2050, 2071–2100).

Methodology and Data

• Models used: U-Net++, Quantile Mapping (QM), Copula models, Multiple Linear Regression, Multiple Nonlinear Regression, M5 model tree, Multivariate Adaptive Regression Splines, Ensemble General Circulation Models (GCMs).
• Data sources: ISIMIP3b gridded large-scale discharge data.

Main Results

• The U-Net++QM integrated model outperformed other models in river flow downscaling, exhibiting a 58% lower Relative Root Mean Square Error (RRMSE).
• Ensemble GCMs demonstrated less uncertainty compared to other models in river flow downscaling.
• For the Ensemble model, the highest drought severity was recorded at -300 (index value), with a maximum duration of 300 months.
• Flood peak flow reached 12,000 cubic meters per second (m³/s), with flood intervals lasting up to 22 months.
• Return periods for compound drought-flood events, as projected by this model, ranged from 50 to 3000 years.
• Future river flow projections, utilizing the Ensemble model and emission scenarios (SSP126, SSP370, and SSP585), indicated increased vulnerability to extreme events in 2071 and 2025 compared to the observed period.

Contributions

• Introduction of an innovative integrated framework for forecasting the combined impacts of drought and flood.
• Provides a valuable management tool for addressing extreme compound hydrological phenomena in the context of climate change.

Funding

Not specified in the provided text.

Citation

@article{Farzin2025Integrating,
  author = {Farzin, Saeed and Anaraki, Mahdi Valikhan and Kadkhodazadeh, Mojtaba and Morshed-Bozorgdel, Amirreza},
  title = {Integrating Deep Learning and Copula Models for Flood–Drought Compound Analysis in Iran},
  journal = {Water},
  year = {2025},
  doi = {10.3390/w17243479},
  url = {https://doi.org/10.3390/w17243479}
}